Introduction

GNU PSPP is a tool for statistical analysis of sampled data. It reads the data, analyzes the data according to commands provided, and writes the results to a listing file, to the standard output or to a window of the graphical display.

The language accepted by PSPP is similar to those accepted by SPSS statistical products. The details of PSPP's language are given later in this manual.

PSPP produces tables and charts as output, which it can produce in several formats; currently, ASCII, PostScript, PDF, HTML, DocBook and TeX are supported.

PSPP is a work in progress. The authors hope to fully support all features in the products that PSPP replaces, eventually. The authors welcome questions, comments, donations, and code submissions.

License

PSPP is not in the public domain. It is copyrighted and there are restrictions on its distribution, but these restrictions are designed to permit everything that a good cooperating citizen would want to do. What is not allowed is to try to prevent others from further sharing any version of this program that they might get from you.

Specifically, we want to make sure that you have the right to give away copies of PSPP, that you receive source code or else can get it if you want it, that you can change these programs or use pieces of them in new free programs, and that you know you can do these things.

To make sure that everyone has such rights, we have to forbid you to deprive anyone else of these rights. For example, if you distribute copies of PSPP, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must tell them their rights.

Also, for our own protection, we must make certain that everyone finds out that there is no warranty for PSPP. If these programs are modified by someone else and passed on, we want their recipients to know that what they have is not what we distributed, so that any problems introduced by others will not reflect on our reputation.

Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.

PSPP is licensed under the GNU General Public License, version 3 or later. This manual is licensed under the GNU Free Documentation License, version 1.3 or later; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts.

Running pspp

This chapter describes how to run pspp, PSPP's main command-line user interface. The pspp program has a number of commands, each of which is documented in its own section.

To see a list of commands, run pspp --help. For help with a particular command, run pspp <command> --help.

Converting file formats with pspp convert

The pspp convert command reads SPSS data and viewer files and writes them out in other formats. The basic syntax is:

pspp convert <INPUT> [OUTPUT]

which reads an input file from <INPUT> and writes a copy of it to [OUTPUT]. If [OUTPUT] is omitted, output is written to the terminal.

The following sections describe how pspp convert works with different kinds of files.

Converting .sav, .por, and .sys Data Files

pspp convert can convert SPSS system files (.sav), SPSS portable files (.por), and SPSS/PC+ system files (.sys) into different formats.

If an output file is named, then pspp convert tries to guess the output format based on its extension:

  • .csv
    .txt
    Comma-separated value. Each value is formatted according to its variable's print format. The first line in the file contains variable names.

  • .sav
    .sys
    SPSS system file.

Without an output file name, the default output format is CSV. Use -O <output_format> to override the default or to specify the format for unrecognized extensions.

Converting .spv Viewer Files

pspp convert can convert SPSS viewer files (.spv files) into multiple different formats.

Options

pspp convert accepts the following general options:

  • -O csv
    -O sys
    Sets the output format.

  • -e <ENCODING>
    --encoding=<ENCODING>
    Sets the character encoding used to read text strings in the input file. This is not needed for new enough SPSS data files, but older data files do not identify their encoding, and PSPP cannot always guess correctly.

    <ENCODING> must be one of the labels for encodings in the Encoding Standard. PSPP does not support UTF-16 or EBCDIC encodings in data files.

    pspp show encodings can help figure out the correct encoding for a system file.

  • -c <MAX_CASES>
    --cases=<MAX_CASES>
    By default, all cases in the input are copied to the output. Specify this option to limit the number of copied cases.

  • -p <PASSWORD>
    --password=<PASSWORD>
    Specifies the password for reading an encrypted SPSS system file.

    pspp convert reads, but does not write, encrypted system files.

    ⚠️ The password (and other command-line options) may be visible to other users on multiuser systems.

System File Output Options

These options only affect output to SPSS system files.

  • --unicode
    Writes system file output with Unicode (UTF-8) encoding. If the input was not already in Unicode, then this causes string variables to be tripled in width.

  • --compression <COMPRESSION>
    Writes data in the system file with the specified format of compression:

    • simple: A simple form of compression that saves space writing small integer values and string segments that are all spaces. All versions of SPSS support simple compression.

    • zlib: More advanced compression that saves space in more general cases. Only SPSS 21 and later can read files written with zlib compression.

CSV Output Options

These options only affect output to CSV files.

  • --no-var-names
    By default, pspp convert writes the variable names as the first line of output. With this option, pspp convert omits this line.

  • --recode
    By default, pspp convert writes user-missing values to CSV output files as their regular values. With this option, pspp convert recodes them to system-missing values (which are written as a single space).

  • --labels
    By default, pspp convert writes variables' values to CSV output files. With this option, pspp convert writes value labels.

  • --print-formats
    By default, pspp convert writes numeric variables as plain numbers. This option makes pspp convert honor variables' print formats.

  • --decimal=DECIMAL
    This option sets the character used as a decimal point in output. The default is .. Only ASCII characters may be used.

  • --delimiter=DELIMITER
    This option sets the character used to separate fields in output. The default is ,, unless the decimal point is ,, in which case ; is used. Only ASCII characters may be used.

  • --qualifier=QUALIFIER
    The option sets the character used to quote fields that contain the delimiter. The default is ". Only ASCII characters may be used.

Inspecting System (.sav) Files with pspp show

The pspp show command reads an SPSS "system file" or data file, which usually has a .sav extension, and produces a report. The basic syntax is:

pspp show <MODE> <INPUT> [OUTPUT]

where <MODE> is a mode of operation (see below), <INPUT> is the SPSS data file to read, and [OUTPUT] is the output file name. If [OUTPUT] is omitted, output is written to the terminal.

The following <MODE>s are available:

  • identify: Outputs a line of text to stdout that identifies the basic kind of system file.

  • dictionary: Outputs the file dictionary in detail, including variables, value labels, attributes, documents, and so on. With --data, also outputs cases from the system file.

    This can be useful as an alternative to PSPP syntax commands such as SYSFILE INFO or DISPLAY DICTIONARY.

    pspp convert is a better way to convert a system file to another format.

  • encodings: Analyzes text data in the system file dictionary and (with --data) cases and produces a report that can help the user to figure out what character encoding the file uses.

    This is useful for old system files that don't identify their own encodings.

  • raw: Outputs the raw structure of the system file dictionary and (with --data) cases. This command does not assume a particular character encoding for the system file, which means that some of the dictionary can't be printed in detail, only in summary.

    This is useful for debugging how PSPP reads system files and for investigating cases of system file corruption, especially when the character encoding is unknown or uncertain.

    This command is most useful with some knowledge of the system file format.

  • decoded: Outputs the raw structure of the system file dictionary and (with --data) cases. Versus raw, this command does decode the dictionary and data with a particular character encoding, which allows it to fully interpret system file records.

    This is useful for debugging how PSPP reads system files and for investigating cases of system file corruption.

    This command is most useful with some knowledge of the system file format.

Options

The following options affect how pspp show reads <INPUT>:

  • --encoding <ENCODING>
    For modes decoded and dictionary, this reads the input file using the specified <ENCODING>, overriding the default.

    <ENCODING> must be one of the labels for encodings in the Encoding Standard. PSPP does not support UTF-16 or EBCDIC encodings in data files.

    pspp show encodings can help figure out the correct encoding for a system file.

  • --data [<MAX_CASES>]
    For modes raw, dictionary, and encodings, this instructs pspp show to read cases from the file. If <MAX_CASES> is given, then that sets a limit on the number of cases to read. Without this option, PSPP will not read any cases.

The following options affect how pspp show writes its output:

  • -f <FORMAT>
    --format <FORMAT>
    Specifies the format to use for output. <FORMAT> may be one of the following:

    • json: JSON using indentation and spaces for easy human consumption.
    • ndjson: Newline-delimited JSON.
    • output: Pivot tables with the PSPP output engine. Use -o for additional configuration.
    • discard: Do not produce any output.

    When these options are not used, the default output format is chosen based on the [OUTPUT] extension. If [OUTPUT] is not specified, then output defaults to JSON.

  • -o <OUTPUT_OPTIONS>
    Adds <OUTPUT_OPTIONS> to the output engine configuration.

Inspecting Portable (.por) Files with pspp show-por

The pspp show-por command reads an SPSS "portable file", which usually has a .por extension, and produces a report. The basic syntax is:

pspp show-por <MODE> <INPUT> [OUTPUT]

where <MODE> is a mode of operation (see below), <INPUT> is the SPSS portable file to read, and [OUTPUT] is the output file name. If [OUTPUT] is omitted, output is written to the terminal.

The portable file format is mostly obsolete. The "system file" or .sav format should be used for writing new data files. Use pspp show to inspect .sav files.

The following <MODE>s are available:

  • dictionary: Outputs the file dictionary in detail, including variables, value labels, documents, and so on. With --data, also outputs cases from the system file.

    This can be useful as an alternative to PSPP syntax commands such as DISPLAY DICTIONARY.

    pspp convert is a better way to convert a portable file to another format.

  • metadata: Outputs portable file metadata not included in the dictionary:

    • The creation date and time declared inside the file (not in the file system).

    • The name of the product and subproduct that wrote the file, if present.

    • The author of the file, if present. This is usually the name of the organization that licensed the product that wrote the file.

    • The character set translation table embedded in the file, as an array with 256 elements, one for each possible value of a byte in the file. Each array element gives the byte value as a 2-digit hexadecimal number paired with the translation table's entry for that byte. Since the file can technically be in any encoding (although the corpus universally uses extended ASCII), the entry is given as a character interpreted in two character sets: windows-1252 and code page 437, in that order. (If the two character sets agree on the code point, then it is only given once.)

      For example, consider a portable's file translation table at offset 0x9e, which in the portable character set is ±. Suppose it has value 0xb1, which is ± in windows-1252 and ▒ in code page 437. Then that array element would be ["9e", "±", "▒"].

    This command is most useful with some knowledge of the portable file format.

  • histogram: Reports on the usage of characters in the portable file. Produces output in the form of an array for each possible value of a byte in the file. Each array element gives the byte value, the byte's character, and the number of times that the byte appears in the file. A given byte is omitted from the table if it does not appear in the file at all, or if the translation table leaves it unmapped. It is also omitted if the byte's character is the ISO-8859-1 encoding of the byte (for example, if byte 0x41 represents A, which is A in ISO-8859-1).

    This command is most useful with some knowledge of the portable file format.

Options

The following options affect how pspp show-por reads <INPUT>:

  • --data [<MAX_CASES>]
    For mode dictionary, and encodings, this instructs pspp show-por to read cases from the file. If <MAX_CASES> is given, then that sets a limit on the number of cases to read. Without this option, PSPP will not read any cases.

The following options affect how pspp show-por writes its output:

  • -f <FORMAT>
    --format <FORMAT>
    Specifies the format to use for output. <FORMAT> may be one of the following:

    • json: JSON using indentation and spaces for easy human consumption.
    • ndjson: Newline-delimited JSON.
    • output: Pivot tables with the PSPP output engine. Use -o for additional configuration.
    • discard: Do not produce any output.

    When these options are not used, the default output format is chosen based on the [OUTPUT] extension. If [OUTPUT] is not specified, then output defaults to JSON.

  • -o <OUTPUT_OPTIONS>
    Adds <OUTPUT_OPTIONS> to the output engine configuration.

Inspecting SPSS/PC+ Files

The pspp show-pc command reads an SPSS/PC+ system file which usually has a .sys extension, and produces a report.

SPSS/PC+ has been obsolete since the 1990s, and its file format is also obsolete and rarely encountered. Use pspp show to inspect modern SPSS system files.

The basic syntax is:

pspp show-pc <MODE> <INPUT> [OUTPUT]

where <MODE> is a mode of operation (see below), <INPUT> is the SPSS/PC+ file to read, and [OUTPUT] is the output file name. If [OUTPUT] is omitted, output is written to the terminal.

The following <MODE>s are available:

  • dictionary: Outputs the file dictionary in detail, including variables, value labels, and so on. With --data, also outputs cases from the system file.

    This can be useful as an alternative to PSPP syntax commands such as DISPLAY DICTIONARY.

    pspp convert is a better way to convert an SPSS/PC+ file to another format.

  • metadata: Outputs metadata not included in the dictionary:

    • The creation date and time declared inside the file (not in the file system).

    • The name of the product family and product that wrote the file, if present.

    • The file name embedded inside the file, if one is present.

    • Whether the file is bytecode-compressed.

    • The number of cases in the file.

Options

The following options affect how pspp show-pc reads <INPUT>:

  • --data [<MAX_CASES>]
    For mode dictionary, and encodings, this instructs pspp show-pc to read cases from the file. If <MAX_CASES> is given, then that sets a limit on the number of cases to read. Without this option, PSPP will not read any cases.

The following options affect how pspp show-pc writes its output:

  • -f <FORMAT>
    --format <FORMAT>
    Specifies the format to use for output. <FORMAT> may be one of the following:

    • json: JSON using indentation and spaces for easy human consumption.
    • ndjson: Newline-delimited JSON.
    • output: Pivot tables with the PSPP output engine. Use -o for additional configuration.
    • discard: Do not produce any output.

    When these options are not used, the default output format is chosen based on the [OUTPUT] extension. If [OUTPUT] is not specified, then output defaults to JSON.

  • -o <OUTPUT_OPTIONS>
    Adds <OUTPUT_OPTIONS> to the output engine configuration.

Inspecting SPSS Viewer Files

The pspp show-spv command reads SPSS Viewer (SPV) files, which usually have .sav extension, and produces a report. The basic syntax is:

pspp show-spv <MODE> <INPUT> [OUTPUT]

where <MODE> is a mode of operation (see below) and <INPUT> is the SPV file to read, and [OUTPUT] is the output file name. If [OUTPUT] is omitted, output is written to the terminal.

The following <MODE>s are accepted:

  • dir: Outputs a table of contents for the SPV file, listing every selected object, which by default is every object except for hidden ones.

    The following additional option for dir is intended mainly for use by PSPP developers:

    • --member-names: Also show the names of the ZIP file members associated with each object.

  • get-table-look: Extracts the TableLook from the first table in the selected objects and outputs it in TableLook XML format. The output file should have an .stt extension.

    Use - for <INPUT> to instead write the default TableLook.

  • convert-table-look: Reads an .stt or .tlo TableLook file as <INPUT> and outputs it in TableLook XML format. The output file should have an .stt extension.

    This is useful for converting a TableLook .tlo file from SPSS 15 or earlier into the newer .stt format.

Input Selection Options

Commands that read an SPV file operate, by default, on all of the objects in the file, except for objects that are not visible in the output viewer window. The user may specify these options to select a subset of the input objects. When multiple options are used, only objects that satisfy all of them are selected:

  • --select=[^]CLASS...
    Include only objects of the given CLASS; with leading ^, include only objects not in the class. Use commas to separate multiple classes. The supported classes are:

    • charts
    • headings
    • logs
    • models
    • tables
    • texts
    • trees
    • warnings
    • outlineheaders
    • pagetitle
    • notes
    • unknown
    • other

  • --commands=[^]COMMAND...
    --subtypes=[^]SUBTYPE...
    --labels=[^]LABEL...
    Include only objects with the specified COMMAND, SUBTYPE, or LABEL. With a leading ^, include only the objects that do not match. Multiple values may be specified separated by commas. An asterisk at the end of a value acts as a wildcard.

    The --command option matches command identifiers, case insensitively. All of the objects produced by a single command use the same, unique command identifier. Command identifiers are always in English regardless of the language used for output. They often differ from the command name in PSPP syntax. Use the pspp-output program's dir command to print command identifiers in particular output.

    The --subtypes option matches particular tables within a command, case insensitively. Subtypes are not necessarily unique: two commands that produce similar output tables may use the same subtype. Only tables have subtypes, so specifying --subtypes will exclude other kinds of objects. Subtypes are always in English and dir will print them.

    The --labels option matches the labels in table output (that is, the table titles). Labels are affected by the output language, variable names and labels, split file settings, and other factors.

  • --nth-commands=N...
    Include only objects from the Nth command that matches --command (or the Nth command overall if --command is not specified), where N is 1 for the first command, 2 for the second, and so on.

  • --instances=INSTANCE...
    Include the specified INSTANCE of an object that matches the other criteria within a single command. INSTANCE may be a number (1 for the first instance, 2 for the second, and so on) or last for the last instance.

  • --show-hidden
    Include hidden output objects in the output. By default, they are excluded.

  • --or
    Separates two sets of selection options. Objects selected by either set of options are included in the output.

The following additional input selection options are intended mainly for use by PSPP developers:

  • --errors
    Include only objects that cause an error when read. With the convert command, this is most useful in conjunction with the --force option.

  • --members=MEMBER...
    Include only the objects that include a listed Zip file MEMBER. More than one name may be included, comma-separated. The members in an SPV file may be listed with the dir command by adding the --show-members option or with the zipinfo program included with many operating systems. Error messages that pspp-output prints when it reads SPV files also often include member names.

Decrypting SPSS files with pspp decrypt

The pspp decrypt command reads an encrypted SPSS file and writes out an equivalent plaintext file. The basic syntax is:

pspp decrypt <INPUT> <OUTPUT>

which reads an encrypted SPSS data, viewer, or syntax file <INPUT>, decrypts it, and writes the decrypted version to <OUTPUT>.

Other commands, such as pspp convert, can also read encrypted files directly.

PSPP does not support writing encrypted files, only reading them.

⚠️ Warning: The SPSS encryption format is insecure: when the password is unknown, it is much cheaper and faster to decrypt a file encrypted this way than if a well designed alternative were used.

Options

pspp decrypt accepts the following options:

  • -p <PASSWORD>
    --password <PASSWORD>
    Specifies the password for reading the encrypted input file. Without this option, PSPP will interactively prompt for the password.

    ⚠️ The password (and other command-line options) may be visible to other users on multiuser systems.

Output Driver Configuration

PSPP can write output in several formats. This section documents the supported formats and how they can be configured.

Text Output (.txt and .text)

PSPP can produce plain text output, drawing boxes using ASCII or Unicode line drawing characters.

Plain text output is encoded in UTF-8.

This driver has the following options:

  • width: <columns>
    Sets the maximum page width to the specified number of columns. To fit in the given width, output tables columns will be word-wrapped or, if necessary, tables will be broken into multiple chunks. The default is no maximum width.

  • boxes: unicode
    boxes: ascii
    Sets the style used for boxes in the output. The following shows an example of each style:

      unicode       ascii
┌────┬────┐  +----+----+
│    │    │  |    |    |
├────┼────┤  +----+----+
│    │    │  |    |    |
└────┴────┘  +----+----+

    Unicode boxes are generally more attractive but they can be harder to work with in some environments. The default is unicode.

  • emphasis: <bool>
    If this is set to true, then the output includes bold and underline emphasis with overstriking. This is supported by only some software, mainly on Unix. The default is false.

PDF Output (.pdf)

This driver has the following options:

  • page_setup: <PageSetup>
    Sets the page size, margins, and other parameters. The following sub-options are available:

    • initial_page_number: <number>
      The page number to use for the first page of output. The default is 1.

    • paper: <size>
      Sets the page size. <size> is a quoted string in the form <w>x<h><unit>, e.g. 8.5x11in or 210x297mm, or the name of a standard paper size, such as letter or a4. The default is system- and user-dependent.

    • margins: <trbl>
      margins: [<tb>, <lr>]
      margins: [<t>, <rl>, <b>]
      margins: [<t>, <r>, <b>, <l>]
      Sets the margins. Each variable is a quoted string with a length and a unit, e.g. 10mm. The one-value form sets all margins to the same length; the two-value form sets the top and bottom margins separately from left and right; and so on. The default is 0.5in.

    • orientation: portrait
      orientation: landscape
      Controls the output page orientation. The default is portrait.

    • object_spacing: <length>
      Sets the vertical spacing between output objects, such as tables or text. <length> is a quoted string with a length and a unit, e.g. 10mm. The default is 12pt, or 1/6 of an inch.

    • chart_spacing: as_is
      chart_spacing: full_height
      chart_spacing: half_height
      chart_spacing: quarter_height
      Sets the size of charts and graphs in the output. The default, as_is, uses the size specified in the charts themselves. The other possibilities set chart size in terms of the height of the page.

    • header: <heading>
      footer: <heading>

HTML Output (.htm and .html)

Comma-Separated Value Output (.csv)

JSON Output (.json)

SPSS Viewer Output (.spv)

PSPP Language Tutorial

PSPP is a tool for the statistical analysis of sampled data. You can use it to discover patterns in the data, to explain differences in one subset of data in terms of another subset and to find out whether certain beliefs about the data are justified. This chapter does not attempt to introduce the theory behind the statistical analysis, but it shows how such analysis can be performed using PSPP.

This tutorial assumes that you are using PSPP in its interactive mode from the command line. However, the example commands can also be typed into a file and executed in a post-hoc mode by typing pspp FILE-NAME at a shell prompt, where FILE-NAME is the name of the file containing the commands. Alternatively, from the graphical interface, you can select File → New → Syntax to open a new syntax window and use the Run menu when a syntax fragment is ready to be executed. Whichever method you choose, the syntax is identical.

When using the interactive method, PSPP tells you that it's waiting for your data with a string like PSPP> or data>. In the examples of this chapter, whenever you see text like this, it indicates the prompt displayed by PSPP, not something that you should type.

Throughout this chapter reference is made to a number of sample data files. So that you can try the examples for yourself, you should have received these files along with your copy of PSPP.1

Normally these files are installed in the directory /usr/local/share/pspp/examples. If however your system administrator or operating system vendor has chosen to install them in a different location, you will have to adjust the examples accordingly.

  1. These files contain purely fictitious data. They should not be used for research purposes. ↩

Preparation of Data Files

Before analysis can commence, the data must be loaded into PSPP and arranged such that both PSPP and humans can understand what the data represents. There are two aspects of data:

  • The variables—these are the parameters of a quantity which has been measured or estimated in some way. For example height, weight and geographic location are all variables.

  • The observations (also called 'cases') of the variables—each observation represents an instance when the variables were measured or observed.

For example, a data set which has the variables height, weight, and name, might have the observations:

1881 89.2 Ahmed
1192 107.01 Frank
1230 67 Julie

The following sections explain how to define a dataset.

Defining Variables

Variables come in two basic types: "numeric" and "string". Variables such as age, height and satisfaction are numeric, whereas name is a string variable. String variables are best reserved for commentary data to assist the human observer. However they can also be used for nominal or categorical data.

The following example defines two variables, forename and height, and reads data into them by manual input:

PSPP> data list list /forename (A12) height.
PSPP> begin data.
data> Ahmed 188
data> Bertram 167
data> Catherine 134.231
data> David 109.1
data> end data
PSPP>

There are several things to note about this example.

  • The words data list list are an example of the DATA LIST. command, which tells PSPP to prepare for reading data. The word list intentionally appears twice. The first occurrence is part of the DATA LIST call, whilst the second tells PSPP that the data is to be read as free format data with one record per line.

    Usually this manual shows command names and other fixed elements of syntax in upper case, but case doesn't matter in most parts of command syntax. In the tutorial, we usually show them in lowercase because they are easier to type that way.

  • The / character is important. It marks the start of the list of variables which you wish to define.

  • The text forename is the name of the first variable, and (A12) says that the variable forename is a string variable and that its maximum length is 12 bytes. The second variable's name is specified by the text height. Since no format is given, this variable has the default format. Normally the default format expects numeric data, which should be entered in the locale of the operating system. Thus, the example is correct for English locales and other locales which use a period (.) as the decimal separator. However if you are using a system with a locale which uses the comma (,) as the decimal separator, then you should in the subsequent lines substitute . with ,. Alternatively, you could explicitly tell PSPP that the height variable is to be read using a period as its decimal separator by appending the text DOT8.3 after the word height. For more information on data formats, see Input and Output Formats.

  • PSPP displays the prompt PSPP> when it's expecting a command. When it's expecting data, the prompt changes to data> so that you know to enter data and not a command.

  • At the end of every command there is a terminating . which tells PSPP that the end of a command has been encountered. You should not enter . when data is expected (ie. when the data> prompt is current) since it is appropriate only for terminating commands.

    You can also terminate a command with a blank line.

Listing the data

Once the data has been entered, you could type

PSPP> list /format=numbered.

to list the data. The optional text /format=numbered requests the case numbers to be shown along with the data. It should show the following output:

           Data List
┌───────────┬─────────┬──────┐
│Case Number│ forename│height│
├───────────┼─────────┼──────┤
│1          │Ahmed    │188.00│
│2          │Bertram  │167.00│
│3          │Catherine│134.23│
│4          │David    │109.10│
└───────────┴─────────┴──────┘

Note that the numeric variable height is displayed to 2 decimal places, because the format for that variable is F8.2. For a complete description of the LIST command, see LIST.

Reading data from a text file

The previous example showed how to define a set of variables and to manually enter the data for those variables. Manual entering of data is tedious work, and often a file containing the data will be have been previously prepared. Let us assume that you have a file called mydata.dat containing the ascii encoded data:

Ahmed          188.00
Bertram        167.00
Catherine      134.23
David          109.10
              .
              .
              .
Zachariah      113.02

You can can tell the DATA LIST command to read the data directly from this file instead of by manual entry, with a command like: PSPP> data list file='mydata.dat' list /forename (A12) height. Notice however, that it is still necessary to specify the names of the variables and their formats, since this information is not contained in the file. It is also possible to specify the file's character encoding and other parameters. For full details refer to DATA LIST.

Reading data from a pre-prepared PSPP file

When working with other PSPP users, or users of other software which uses the PSPP data format, you may be given the data in a pre-prepared PSPP file. Such files contain not only the data, but the variable definitions, along with their formats, labels and other meta-data. Conventionally, these files (sometimes called "system" files) have the suffix .sav, but that is not mandatory. The following syntax loads a file called my-file.sav.

PSPP> get file='my-file.sav'.

You will encounter several instances of this in future examples.

Saving data to a PSPP file.

If you want to save your data, along with the variable definitions so that you or other PSPP users can use it later, you can do this with the SAVE command.

The following syntax will save the existing data and variables to a file called my-new-file.sav.

PSPP> save outfile='my-new-file.sav'.

If my-new-file.sav already exists, then it will be overwritten. Otherwise it will be created.

Reading data from other sources

Sometimes it's useful to be able to read data from comma separated text, from spreadsheets, databases or other sources. In these instances you should use the GET DATA command.

Exiting PSPP

Use the FINISH command to exit PSPP: PSPP> finish.

Data Screening and Transformation

Once data has been entered, it is often desirable, or even necessary, to transform it in some way before performing analysis upon it. At the very least, it's good practice to check for errors.

Identifying incorrect data

Data from real sources is rarely error free. PSPP has a number of procedures which can be used to help identify data which might be incorrect.

The DESCRIPTIVES command is used to generate simple linear statistics for a dataset. It is also useful for identifying potential problems in the data. The example file physiology.sav contains a number of physiological measurements of a sample of healthy adults selected at random. However, the data entry clerk made a number of mistakes when entering the data. The following example illustrates the use of DESCRIPTIVES to screen this data and identify the erroneous values:

PSPP> get file='/usr/local/share/pspp/examples/physiology.sav'.
PSPP> descriptives sex, weight, height.

For this example, PSPP produces the following output:

                  Descriptive Statistics
┌─────────────────────┬──┬───────┬───────┬───────┬───────┐
│                     │ N│  Mean │Std Dev│Minimum│Maximum│
├─────────────────────┼──┼───────┼───────┼───────┼───────┤
│Sex of subject       │40│    .45│    .50│Male   │Female │
│Weight in kilograms  │40│  72.12│  26.70│  ─55.6│   92.1│
│Height in millimeters│40│1677.12│ 262.87│    179│   1903│
│Valid N (listwise)   │40│       │       │       │       │
│Missing N (listwise) │ 0│       │       │       │       │
└─────────────────────┴──┴───────┴───────┴───────┴───────┘

The most interesting column in the output is the minimum value. The weight variable has a minimum value of less than zero, which is clearly erroneous. Similarly, the height variable's minimum value seems to be very low. In fact, it is more than 5 standard deviations from the mean, and is a seemingly bizarre height for an adult person.

We can look deeper into these discrepancies by issuing an additional EXAMINE command:

PSPP> examine height, weight /statistics=extreme(3).

This command produces the following additional output (in part):

                   Extreme Values
┌───────────────────────────────┬───────────┬─────┐
│                               │Case Number│Value│
├───────────────────────────────┼───────────┼─────┤
│Height in millimeters Highest 1│         14│ 1903│
│                              2│         15│ 1884│
│                              3│         12│ 1802│
│                     ──────────┼───────────┼─────┤
│                      Lowest  1│         30│  179│
│                              2│         31│ 1598│
│                              3│         28│ 1601│
├───────────────────────────────┼───────────┼─────┤
│Weight in kilograms   Highest 1│         13│ 92.1│
│                              2│          5│ 92.1│
│                              3│         17│ 91.7│
│                     ──────────┼───────────┼─────┤
│                      Lowest  1│         38│─55.6│
│                              2│         39│ 54.5│
│                              3│         33│ 55.4│
└───────────────────────────────┴───────────┴─────┘

From this new output, you can see that the lowest value of height is 179 (which we suspect to be erroneous), but the second lowest is 1598 which we know from DESCRIPTIVES is within 1 standard deviation from the mean. Similarly, the lowest value of weight is negative, but its second lowest value is plausible. This suggests that the two extreme values are outliers and probably represent data entry errors.

The output also identifies the case numbers for each extreme value, so we can see that cases 30 and 38 are the ones with the erroneous values.

Dealing with suspicious data

If possible, suspect data should be checked and re-measured. However, this may not always be feasible, in which case the researcher may decide to disregard these values. PSPP has a feature for missing values, whereby data can assume the special value 'SYSMIS', and will be disregarded in future analysis. You can set the two suspect values to the SYSMIS value using the RECODE command.

PSPP> recode height (179 = SYSMIS).
PSPP> recode weight (LOWEST THRU 0 = SYSMIS).

The first command says that for any observation which has a height value of 179, that value should be changed to the SYSMIS value. The second command says that any weight values of zero or less should be changed to SYSMIS. From now on, they will be ignored in analysis.

If you now re-run the DESCRIPTIVES or EXAMINE commands from the previous section, you will see a data summary with more plausible parameters. You will also notice that the data summaries indicate the two missing values.

Inverting negatively coded variables

Data entry errors are not the only reason for wanting to recode data. The sample file hotel.sav comprises data gathered from a customer satisfaction survey of clients at a particular hotel. The following commands load the file and display its variables and associated data:

PSPP> get file='/usr/local/share/pspp/examples/hotel.sav'.
PSPP> display dictionary.

It yields the following output:

                                   Variables
┌────┬────────┬─────────────┬────────────┬─────┬─────┬─────────┬──────┬───────┐
│    │        │             │ Measurement│     │     │         │ Print│ Write │
│Name│Position│    Label    │    Level   │ Role│Width│Alignment│Format│ Format│
├────┼────────┼─────────────┼────────────┼─────┼─────┼─────────┼──────┼───────┤
│v1  │       1│I am         │Ordinal     │Input│    8│Right    │F8.0  │F8.0   │
│    │        │satisfied    │            │     │     │         │      │       │
│    │        │with the     │            │     │     │         │      │       │
│    │        │level of     │            │     │     │         │      │       │
│    │        │service      │            │     │     │         │      │       │
│v2  │       2│The value for│Ordinal     │Input│    8│Right    │F8.0  │F8.0   │
│    │        │money was    │            │     │     │         │      │       │
│    │        │good         │            │     │     │         │      │       │
│v3  │       3│The staff    │Ordinal     │Input│    8│Right    │F8.0  │F8.0   │
│    │        │were slow in │            │     │     │         │      │       │
│    │        │responding   │            │     │     │         │      │       │
│v4  │       4│My concerns  │Ordinal     │Input│    8│Right    │F8.0  │F8.0   │
│    │        │were dealt   │            │     │     │         │      │       │
│    │        │with in an   │            │     │     │         │      │       │
│    │        │efficient    │            │     │     │         │      │       │
│    │        │manner       │            │     │     │         │      │       │
│v5  │       5│There was too│Ordinal     │Input│    8│Right    │F8.0  │F8.0   │
│    │        │much noise in│            │     │     │         │      │       │
│    │        │the rooms    │            │     │     │         │      │       │
└────┴────────┴─────────────┴────────────┴─────┴─────┴─────────┴──────┴───────┘

                              Value Labels
┌────────────────────────────────────────────────────┬─────────────────┐
│Variable Value                                      │      Label      │
├────────────────────────────────────────────────────┼─────────────────┤
│I am satisfied with the level of service           1│Strongly Disagree│
│                                                   2│Disagree         │
│                                                   3│No Opinion       │
│                                                   4│Agree            │
│                                                   5│Strongly Agree   │
├────────────────────────────────────────────────────┼─────────────────┤
│The value for money was good                       1│Strongly Disagree│
│                                                   2│Disagree         │
│                                                   3│No Opinion       │
│                                                   4│Agree            │
│                                                   5│Strongly Agree   │
├────────────────────────────────────────────────────┼─────────────────┤
│The staff were slow in responding                  1│Strongly Disagree│
│                                                   2│Disagree         │
│                                                   3│No Opinion       │
│                                                   4│Agree            │
│                                                   5│Strongly Agree   │
├────────────────────────────────────────────────────┼─────────────────┤
│My concerns were dealt with in an efficient manner 1│Strongly Disagree│
│                                                   2│Disagree         │
│                                                   3│No Opinion       │
│                                                   4│Agree            │
│                                                   5│Strongly Agree   │
├────────────────────────────────────────────────────┼─────────────────┤
│There was too much noise in the rooms              1│Strongly Disagree│
│                                                   2│Disagree         │
│                                                   3│No Opinion       │
│                                                   4│Agree            │
│                                                   5│Strongly Agree   │
└────────────────────────────────────────────────────┴─────────────────┘

The output shows that all of the variables v1 through v5 are measured on a 5 point Likert scale, with 1 meaning "Strongly disagree" and 5 meaning "Strongly agree". However, some of the questions are positively worded (v1, v2, v4) and others are negatively worded (v3, v5). To perform meaningful analysis, we need to recode the variables so that they all measure in the same direction. We could use the RECODE command, with syntax such as:

recode v3 (1 = 5) (2 = 4) (4 = 2) (5 = 1).

However an easier and more elegant way uses the COMPUTE command. Since the variables are Likert variables in the range (1 ... 5), subtracting their value from 6 has the effect of inverting them:

compute VAR = 6 - VAR.

The following section uses this technique to recode the variables v3 and v5. After applying COMPUTE for both variables, all subsequent commands will use the inverted values.

Testing data consistency

A sensible check to perform on survey data is the calculation of reliability. This gives the statistician some confidence that the questionnaires have been completed thoughtfully. If you examine the labels of variables v1, v3 and v4, you will notice that they ask very similar questions. One would therefore expect the values of these variables (after recoding) to closely follow one another, and we can test that with the RELIABILITY command. The following example shows a PSPP session where the user recodes negatively scaled variables and then requests reliability statistics for v1, v3, and v4.

PSPP> get file='/usr/local/share/pspp/examples/hotel.sav'.
PSPP> compute v3 = 6 - v3.
PSPP> compute v5 = 6 - v5.
PSPP> reliability v1, v3, v4.

This yields the following output:

Scale: ANY

Case Processing Summary
┌────────┬──┬───────┐
│Cases   │ N│Percent│
├────────┼──┼───────┤
│Valid   │17│ 100.0%│
│Excluded│ 0│    .0%│
│Total   │17│ 100.0%│
└────────┴──┴───────┘

    Reliability Statistics
┌────────────────┬──────────┐
│Cronbach's Alpha│N of Items│
├────────────────┼──────────┤
│             .81│         3│
└────────────────┴──────────┘

As a rule of thumb, many statisticians consider a value of Cronbach's Alpha of 0.7 or higher to indicate reliable data.

Here, the value is 0.81, which suggests a high degree of reliability among variables v1, v3 and v4, so the data and the recoding that we performed are vindicated.

Testing for normality

Many statistical tests rely upon certain properties of the data. One common property, upon which many linear tests depend, is that of normality -- the data must have been drawn from a normal distribution. It is necessary then to ensure normality before deciding upon the test procedure to use. One way to do this uses the EXAMINE command.

In the following example, a researcher was examining the failure rates of equipment produced by an engineering company. The file repairs.sav contains the mean time between failures (mtbf) of some items of equipment subject to the study. Before performing linear analysis on the data, the researcher wanted to ascertain that the data is normally distributed.

PSPP> get file='/usr/local/share/pspp/examples/repairs.sav'.
PSPP> examine mtbf /statistics=descriptives.

This produces the following output:

                                  Descriptives
┌──────────────────────────────────────────────────────────┬─────────┬────────┐
│                                                          │         │  Std.  │
│                                                          │Statistic│  Error │
├──────────────────────────────────────────────────────────┼─────────┼────────┤
│Mean time between        Mean                             │     8.78│    1.10│
│failures (months)       ──────────────────────────────────┼─────────┼────────┤
│                         95% Confidence Interval Lower    │     6.53│        │
│                         for Mean                Bound    │         │        │
│                                                 Upper    │    11.04│        │
│                                                 Bound    │         │        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         5% Trimmed Mean                  │     8.20│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Median                           │     8.29│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Variance                         │    36.34│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Std. Deviation                   │     6.03│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Minimum                          │     1.63│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Maximum                          │    26.47│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Range                            │    24.84│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Interquartile Range              │     6.03│        │
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Skewness                         │     1.65│     .43│
│                        ──────────────────────────────────┼─────────┼────────┤
│                         Kurtosis                         │     3.41│     .83│
└──────────────────────────────────────────────────────────┴─────────┴────────┘

A normal distribution has a skewness and kurtosis of zero. The skewness of mtbf in the output above makes it clear that the mtbf figures have a lot of positive skew and are therefore not drawn from a normally distributed variable. Positive skew can often be compensated for by applying a logarithmic transformation, as in the following continuation of the example:

PSPP> compute mtbf_ln = ln (mtbf).
PSPP> examine mtbf_ln /statistics=descriptives.

which produces the following additional output:

                                Descriptives
┌────────────────────────────────────────────────────┬─────────┬──────────┐
│                                                    │Statistic│Std. Error│
├────────────────────────────────────────────────────┼─────────┼──────────┤
│mtbf_ln Mean                                        │     1.95│       .13│
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        95% Confidence Interval for Mean Lower Bound│     1.69│          │
│                                         Upper Bound│     2.22│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        5% Trimmed Mean                             │     1.96│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Median                                      │     2.11│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Variance                                    │      .49│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Std. Deviation                              │      .70│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Minimum                                     │      .49│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Maximum                                     │     3.28│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Range                                       │     2.79│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Interquartile Range                         │      .88│          │
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Skewness                                    │     ─.37│       .43│
│       ─────────────────────────────────────────────┼─────────┼──────────┤
│        Kurtosis                                    │      .01│       .83│
└────────────────────────────────────────────────────┴─────────┴──────────┘

The COMPUTE command in the first line above performs the logarithmic transformation: compute mtbf_ln = ln (mtbf). Rather than redefining the existing variable, this use of COMPUTE defines a new variable mtbf_ln which is the natural logarithm of mtbf. The final command in this example calls EXAMINE on this new variable. The results show that both the skewness and kurtosis for mtbf_ln are very close to zero. This provides some confidence that the mtbf_ln variable is normally distributed and thus safe for linear analysis. In the event that no suitable transformation can be found, then it would be worth considering an appropriate non-parametric test instead of a linear one. See NPAR TESTS, for information about non-parametric tests.

Hypothesis Testing

One of the most fundamental purposes of statistical analysis is hypothesis testing. Researchers commonly need to test hypotheses about a set of data. For example, she might want to test whether one set of data comes from the same distribution as another, or whether the mean of a dataset significantly differs from a particular value. This section presents just some of the possible tests that PSPP offers.

The researcher starts by making a "null hypothesis". Often this is a hypothesis which he suspects to be false. For example, if he suspects that A is greater than B he will state the null hypothesis as A = B.1

The "p-value" is a recurring concept in hypothesis testing. It is the highest acceptable probability that the evidence implying a null hypothesis is false, could have been obtained when the null hypothesis is in fact true. Note that this is not the same as "the probability of making an error" nor is it the same as "the probability of rejecting a hypothesis when it is true".

Testing for differences of means

A common statistical test involves hypotheses about means. The T-TEST command is used to find out whether or not two separate subsets have the same mean.

A researcher suspected that the heights and core body temperature of persons might be different depending upon their sex. To investigate this, he posed two null hypotheses based on the data from physiology.sav previously encountered:

  • The mean heights of males and females in the population are equal.

  • The mean body temperature of males and females in the population are equal.

For the purposes of the investigation the researcher decided to use a p-value of 0.05.

In addition to the T-test, the T-TEST command also performs the Levene test for equal variances. If the variances are equal, then a more powerful form of the T-test can be used. However if it is unsafe to assume equal variances, then an alternative calculation is necessary. PSPP performs both calculations.

For the height variable, the output shows the significance of the Levene test to be 0.33 which means there is a 33% probability that the Levene test produces this outcome when the variances are equal. Had the significance been less than 0.05, then it would have been unsafe to assume that the variances were equal. However, because the value is higher than 0.05 the homogeneity of variances assumption is safe and the "Equal Variances" row (the more powerful test) can be used. Examining this row, the two tailed significance for the height t-test is less than 0.05, so it is safe to reject the null hypothesis and conclude that the mean heights of males and females are unequal.

For the temperature variable, the significance of the Levene test is 0.58 so again, it is safe to use the row for equal variances. The equal variances row indicates that the two tailed significance for temperature is 0.20. Since this is greater than 0.05 we must reject the null hypothesis and conclude that there is insufficient evidence to suggest that the body temperature of male and female persons are different.

The syntax for this analysis is:

PSPP> get file='/usr/local/share/pspp/examples/physiology.sav'.
PSPP> recode height (179 = SYSMIS).
PSPP> t-test group=sex(0,1) /variables = height temperature.

PSPP produces the following output for this syntax:

                                Group Statistics
┌───────────────────────────────────────────┬──┬───────┬─────────────┬────────┐
│                                           │  │       │     Std.    │  S.E.  │
│                                     Group │ N│  Mean │  Deviation  │  Mean  │
├───────────────────────────────────────────┼──┼───────┼─────────────┼────────┤
│Height in millimeters                Male  │22│1796.49│        49.71│   10.60│
│                                     Female│17│1610.77│        25.43│    6.17│
├───────────────────────────────────────────┼──┼───────┼─────────────┼────────┤
│Internal body temperature in degrees Male  │22│  36.68│         1.95│     .42│
│Celcius                              Female│18│  37.43│         1.61│     .38│
└───────────────────────────────────────────┴──┴───────┴─────────────┴────────┘

                          Independent Samples Test
┌─────────────────────┬──────────┬──────────────────────────────────────────
│                     │ Levene's │
│                     │ Test for │
│                     │ Equality │
│                     │    of    │
│                     │ Variances│              T─Test for Equality of Means
│                     ├────┬─────┼─────┬─────┬───────┬──────────┬──────────┐
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │  Sig. │          │          │
│                     │    │     │     │     │  (2─  │   Mean   │Std. Error│
│                     │  F │ Sig.│  t  │  df │tailed)│Difference│Difference│
├─────────────────────┼────┼─────┼─────┼─────┼───────┼──────────┼──────────┤
│Height in   Equal    │ .97│ .331│14.02│37.00│   .000│    185.72│     13.24│
│millimeters variances│    │     │     │     │       │          │          │
│            assumed  │    │     │     │     │       │          │          │
│            Equal    │    │     │15.15│32.71│   .000│    185.72│     12.26│
│            variances│    │     │     │     │       │          │          │
│            not      │    │     │     │     │       │          │          │
│            assumed  │    │     │     │     │       │          │          │
├─────────────────────┼────┼─────┼─────┼─────┼───────┼──────────┼──────────┤
│Internal    Equal    │ .31│ .581│─1.31│38.00│   .198│      ─.75│       .57│
│body        variances│    │     │     │     │       │          │          │
│temperature assumed  │    │     │     │     │       │          │          │
│in degrees  Equal    │    │     │─1.33│37.99│   .190│      ─.75│       .56│
│Celcius     variances│    │     │     │     │       │          │          │
│            not      │    │     │     │     │       │          │          │
│            assumed  │    │     │     │     │       │          │          │
└─────────────────────┴────┴─────┴─────┴─────┴───────┴──────────┴──────────┘

┌─────────────────────┬─────────────┐
│                     │             │
│                     │             │
│                     │             │
│                     │             │
│                     │             │
│                     ├─────────────┤
│                     │     95%     │
│                     │  Confidence │
│                     │ Interval of │
│                     │     the     │
│                     │  Difference │
│                     ├──────┬──────┤
│                     │ Lower│ Upper│
├─────────────────────┼──────┼──────┤
│Height in   Equal    │158.88│212.55│
│millimeters variances│      │      │
│            assumed  │      │      │
│            Equal    │160.76│210.67│
│            variances│      │      │
│            not      │      │      │
│            assumed  │      │      │
├─────────────────────┼──────┼──────┤
│Internal    Equal    │ ─1.91│   .41│
│body        variances│      │      │
│temperature assumed  │      │      │
│in degrees  Equal    │ ─1.89│   .39│
│Celcius     variances│      │      │
│            not      │      │      │
│            assumed  │      │      │
└─────────────────────┴──────┴──────┘

The T-TEST command tests for differences of means. Here, the height variable's two tailed significance is less than 0.05, so the null hypothesis can be rejected. Thus, the evidence suggests there is a difference between the heights of male and female persons. However the significance of the test for the temperature variable is greater than 0.05 so the null hypothesis cannot be rejected, and there is insufficient evidence to suggest a difference in body temperature.

Linear Regression

Linear regression is a technique used to investigate if and how a variable is linearly related to others. If a variable is found to be linearly related, then this can be used to predict future values of that variable.

In the following example, the service department of the company wanted to be able to predict the time to repair equipment, in order to improve the accuracy of their quotations. It was suggested that the time to repair might be related to the time between failures and the duty cycle of the equipment. The p-value of 0.1 was chosen for this investigation. In order to investigate this hypothesis, the REGRESSION command was used. This command not only tests if the variables are related, but also identifies the potential linear relationship.

A first attempt includes duty_cycle:

PSPP> get file='/usr/local/share/pspp/examples/repairs.sav'.
PSPP> regression /variables = mtbf duty_cycle /dependent = mttr.

This attempt yields the following output (in part):

                  Coefficients (Mean time to repair (hours) )
┌────────────────────────┬─────────────────────┬───────────────────┬─────┬────┐
│                        │    Unstandardized   │    Standardized   │     │    │
│                        │     Coefficients    │    Coefficients   │     │    │
│                        ├─────────┬───────────┼───────────────────┤     │    │
│                        │    B    │ Std. Error│        Beta       │  t  │Sig.│
├────────────────────────┼─────────┼───────────┼───────────────────┼─────┼────┤
│(Constant)              │    10.59│       3.11│                .00│ 3.40│.002│
│Mean time between       │     3.02│        .20│                .95│14.88│.000│
│failures (months)       │         │           │                   │     │    │
│Ratio of working to non─│    ─1.12│       3.69│               ─.02│ ─.30│.763│
│working time            │         │           │                   │     │    │
└────────────────────────┴─────────┴───────────┴───────────────────┴─────┴────┘

The coefficients in the above table suggest that the formula \(\textrm{MTTR} = 9.81 + 3.1 \times \textrm{MTBF} + 1.09 \times \textrm{DUTY_CYCLE}\) can be used to predict the time to repair. However, the significance value for the DUTY_CYCLE coefficient is very high, which would make this an unsafe predictor. For this reason, the test was repeated, but omitting the duty_cycle variable:

PSPP> regression /variables = mtbf /dependent = mttr.

This second try produces the following output (in part):

                  Coefficients (Mean time to repair (hours) )
┌───────────────────────┬──────────────────────┬───────────────────┬─────┬────┐
│                       │    Unstandardized    │    Standardized   │     │    │
│                       │     Coefficients     │    Coefficients   │     │    │
│                       ├─────────┬────────────┼───────────────────┤     │    │
│                       │    B    │ Std. Error │        Beta       │  t  │Sig.│
├───────────────────────┼─────────┼────────────┼───────────────────┼─────┼────┤
│(Constant)             │     9.90│        2.10│                .00│ 4.71│.000│
│Mean time between      │     3.01│         .20│                .94│15.21│.000│
│failures (months)      │         │            │                   │     │    │
└───────────────────────┴─────────┴────────────┴───────────────────┴─────┴────┘

This time, the significance of all coefficients is no higher than 0.06, suggesting that at the 0.06 level, the formula \(\textrm{MTTR} = 10.5 + 3.11 \times \textrm{MTBF}\) is a reliable predictor of the time to repair.

  1. This example assumes that it is already proven that B is not greater than A. ↩

This chapter discusses elements common to many PSPP commands. Later chapters describe individual commands in detail.

Tokens

PSPP divides most syntax file lines into series of short chunks called "tokens". Tokens are then grouped to form commands, each of which tells PSPP to take some action—read in data, write out data, perform a statistical procedure, etc. Each type of token is described below.

Identifiers

Identifiers are names that typically specify variables, commands, or subcommands. The first character in an identifier must be a letter, #, or @. The remaining characters in the identifier must be letters, digits, or one of the following special characters:

. _ $ # @

Identifiers may be any length, but only the first 64 bytes are significant. Identifiers are not case-sensitive: foobar, Foobar, FooBar, FOOBAR, and FoObaR are different representations of the same identifier.

Some identifiers are reserved. Reserved identifiers may not be used in any context besides those explicitly described in this manual. The reserved identifiers are:

ALL AND BY EQ GE GT LE LT NE NOT OR TO WITH

Keywords

Keywords are a subclass of identifiers that form a fixed part of command syntax. For example, command and subcommand names are keywords. Keywords may be abbreviated to their first 3 characters if this abbreviation is unambiguous. (Unique abbreviations of 3 or more characters are also accepted: FRE, FREQ, and FREQUENCIES are equivalent when the last is a keyword.)

Reserved identifiers are always used as keywords. Other identifiers may be used both as keywords and as user-defined identifiers, such as variable names.

Numbers

Numbers are expressed in decimal. A decimal point is optional. Numbers may be expressed in scientific notation by adding e and a base-10 exponent, so that 1.234e3 has the value 1234. Here are some more examples of valid numbers:

-5  3.14159265359  1e100  -.707  8945.

Negative numbers are expressed with a - prefix. However, in situations where a literal - token is expected, what appears to be a negative number is treated as - followed by a positive number.

No white space is allowed within a number token, except for horizontal white space between - and the rest of the number.

The last example above, 8945. is interpreted as two tokens, 8945 and ., if it is the last token on a line (see Forming Commands).

Strings

Strings are literal sequences of characters enclosed in pairs of single quotes (') or double quotes ("). To include the character used for quoting in the string, double it, e.g. 'it''s an apostrophe'. White space and case of letters are significant inside strings.

Strings can be concatenated using +, so that "a" + 'b' + 'c' is equivalent to 'abc'. So that a long string may be broken across lines, a line break may precede or follow, or both precede and follow, the +. (However, an entirely blank line preceding or following the + is interpreted as ending the current command.)

Strings may also be expressed as hexadecimal character values by prefixing the initial quote character by x or X. Regardless of the syntax file or active dataset's encoding, the hexadecimal digits in the string are interpreted as Unicode characters in UTF-8 encoding.

Individual Unicode code points may also be expressed by specifying the hexadecimal code point number in single or double quotes preceded by u or U. For example, Unicode code point U+1D11E, the musical G clef character, could be expressed as U'1D11E'. Invalid Unicode code points (above U+10FFFF or in between U+D800 and U+DFFF) are not allowed.

When strings are concatenated with +, each segment's prefix is considered individually. For example, 'The G clef symbol is:' + u"1d11e" + "." inserts a G clef symbol in the middle of an otherwise plain text string.

Punctuators and Operators

These tokens are the punctuators and operators:

, / = ( ) + - * / ** < <= <> > >= ~= & | .

Most of these appear within the syntax of commands, but the period (.) punctuator is used only at the end of a command. It is a punctuator only as the last character on a line (except white space). When it is the last non-space character on a line, a period is not treated as part of another token, even if it would otherwise be part of, e.g., an identifier or a floating-point number.

Forming Commands

Most PSPP commands share a common structure. A command begins with a command name, such as FREQUENCIES, DATA LIST, or N OF CASES. The command name may be abbreviated to its first word, and each word in the command name may be abbreviated to its first three or more characters, where these abbreviations are unambiguous.

The command name may be followed by one or more "subcommands". Each subcommand begins with a subcommand name, which may be abbreviated to its first three letters. Some subcommands accept a series of one or more specifications, which follow the subcommand name, optionally separated from it by an equals sign (=). Specifications may be separated from each other by commas or spaces. Each subcommand must be separated from the next (if any) by a forward slash (/).

There are multiple ways to mark the end of a command. The most common way is to end the last line of the command with a period (.) as described in the previous section. A blank line, or one that consists only of white space or comments, also ends a command.

Syntax Variants

There are three variants of command syntax, which vary only in how they detect the end of one command and the start of the next.

In "interactive mode", which is the default for syntax typed at a command prompt, a period as the last non-blank character on a line ends a command. A blank line also ends a command.

In "batch mode", an end-of-line period or a blank line also ends a command. Additionally, it treats any line that has a non-blank character in the leftmost column as beginning a new command. Thus, in batch mode the second and subsequent lines in a command must be indented.

Regardless of the syntax mode, a plus sign, minus sign, or period in the leftmost column of a line is ignored and causes that line to begin a new command. This is most useful in batch mode, in which the first line of a new command could not otherwise be indented, but it is accepted regardless of syntax mode.

The default mode for reading commands from a file is "auto mode". It is the same as batch mode, except that a line with a non-blank in the leftmost column only starts a new command if that line begins with the name of a PSPP command. This correctly interprets most valid PSPP syntax files regardless of the syntax mode for which they are intended.

The --interactive (or -i) or --batch (or -b) options set the syntax mode for files listed on the PSPP command line.

Handling Missing Values

PSPP includes special support for unknown numeric data values. Missing observations are assigned a special value, called the "system-missing value". This "value" actually indicates the absence of a value; it means that the actual value is unknown. Procedures automatically exclude from analyses those observations or cases that have missing values. Details of missing value exclusion depend on the procedure and can often be controlled by the user; refer to descriptions of individual procedures for details.

The system-missing value exists only for numeric variables. String variables always have a defined value, even if it is only a string of spaces.

Variables, whether numeric or string, can have designated "user-missing values". Every user-missing value is an actual value for that variable. However, most of the time user-missing values are treated in the same way as the system-missing value.

Datasets

PSPP works with data organized into "datasets". A dataset consists of a set of "variables", which taken together are said to form a "dictionary", and one or more "cases", each of which has one value for each variable.

At any given time PSPP has exactly one distinguished dataset, called the "active dataset". Most PSPP commands work only with the active dataset. In addition to the active dataset, PSPP also supports any number of additional open datasets. The DATASET commands can choose a new active dataset from among those that are open, as well as create and destroy datasets.

Attributes of Variables

Each variable has a number of attributes, including:

  • Name
    An identifier, up to 64 bytes long. Each variable must have a different name.

    User-defined variable names may not begin with $.

    A variable name can with ., but it should not, because such an identifier will be misinterpreted when it is the final token on a line: FOO. is divided into two separate tokens, FOO and ., indicating end-of-command.

    A variable name can end with _, but it should not, because some PSPP procedures reserve those names for special purposes.

    Variable names are not case-sensitive. PSPP capitalizes variable names on output the same way they were capitalized at their point of definition in the input.

  • Type
    Numeric or string.

  • Width (string variables only)
    String variables with a width of 8 characters or fewer are called "short string variables", and wider ones are called "long string variables". In a few contexts, long string variables are not allowed.

  • Position
    Variables in the dictionary are arranged in a specific order. DISPLAY can show this order.

  • Initialization
    Either reinitialized to 0 or spaces for each case, or left at its existing value. Use LEAVE to avoid reinitializing a variable.

  • Missing values
    Optionally, up to three values, or a range of values, or a specific value plus a range, can be specified as "user-missing values". There is also a "system-missing value" that is assigned to an observation when there is no other obvious value for that observation. Observations with missing values are automatically excluded from analyses. User-missing values are actual data values, while the system-missing value is not a value at all. See Handling Missing Values for more information on missing values. The MISSING VALUES command sets missing values.

  • Variable label
    A string that describes the variable. The VARIABLE LABELS command sets variable labels.

  • Value label
    Optionally, these associate each possible value of the variable with a string. The VALUE LABELS and ADD VALUE LABELS commands set value labels.

  • Print format
    Display width, format, and (for numeric variables) number of decimal places. This attribute does not affect how data are stored, just how they are displayed. See Input and Output Formats for details. The FORMATS and PRINT FORMATS commands set print formats.

  • Write format
    Similar to print format, but used by the WRITE command. The FORMATS and WRITE FORMATS commands set write formats.

  • Measurement level
    One of the following:

    • Nominal: Each value of a nominal variable represents a distinct category. The possible categories are finite and often have value labels. The order of categories is not significant. Political parties, US states, and yes/no choices are nominal. Numeric and string variables can be nominal.

    • Ordinal: Ordinal variables also represent distinct categories, but their values are arranged according to some natural order. Likert scales, e.g. from strongly disagree to strongly agree, are ordinal. Data grouped into ranges, e.g. age groups or income groups, are ordinal. Both numeric and string variables can be ordinal. String values are ordered alphabetically, so letter grades from A to F will work as expected, but poor, satisfactory, excellent will not.

    • Scale: Scale variables are ones for which differences and ratios are meaningful. These are often values which have a natural unit attached, such as age in years, income in dollars, or distance in miles. Only numeric variables are scalar.

    The VARIABLE LEVEL command sets measurement levels.

    Variables created by COMPUTE and similar transformations, obtained from external sources, etc., initially have an unknown measurement level. Any procedure that reads the data will then assign a default measurement level. PSPP can assign some defaults without reading the data:

    • Nominal, if it's a string variable.

    • Nominal, if the variable has a WKDAY or MONTH print format.

    • Scale, if the variable has a DOLLAR, CCA through CCE, or time or date print format.

    Otherwise, PSPP reads the data and decides based on its distribution:

    • Nominal, if all observations are missing.

    • Scale, if one or more valid observations are noninteger or negative.

    • Scale, if no valid observation is less than 10.

    • Scale, if the variable has 24 or more unique valid values. The value 24 is the default. Use SET SCALEMIN to change the default.

    Finally, if none of the above is true, PSPP assigns the variable a nominal measurement level.

  • Custom attributes
    User-defined associations between names and values. The VARIABLE ATTRIBUTE command sets variable atributes.

  • Role
    The intended role of a variable for use in dialog boxes in graphical user interfaces. The VARIABLE ROLE command sets variable roles.

Variable Lists

To refer to a set of variables, list their names one after another. Optionally, their names may be separated by commas. To include a range of variables from the dictionary in the list, write the name of the first and last variable in the range, separated by TO. For instance, if the dictionary contains six variables with the names ID, X1, X2, GOAL, MET, and NEXTGOAL, in that order, then X2 TO MET would include variables X2, GOAL, and MET.

Commands that define variables, such as DATA LIST, give TO an alternate meaning. With these commands, TO define sequences of variables whose names end in consecutive integers. The syntax is two identifiers that begin with the same root and end with numbers, separated by TO. The syntax X1 TO X5 defines 5 variables, named X1, X2, X3, X4, and X5. The syntax ITEM0008 TO ITEM0013 defines 6 variables, named ITEM0008, ITEM0009, ITEM0010, ITEM0011, ITEM0012, and ITEM00013. The syntaxes QUES001 TO QUES9 and QUES6 TO QUES3 are invalid.

After a set of variables has been defined with DATA LIST or another command with this method, the same set can be referenced on later commands using the same syntax.

Input and Output Formats

An "input format" describes how to interpret the contents of an input field as a number or a string. It might specify that the field contains an ordinary decimal number, a time or date, a number in binary or hexadecimal notation, or one of several other notations. Input formats are used by commands such as DATA LIST that read data or syntax files into the PSPP active dataset.

Every input format corresponds to a default "output format" that specifies the formatting used when the value is output later. It is always possible to explicitly specify an output format that resembles the input format. Usually, this is the default, but in cases where the input format is unfriendly to human readability, such as binary or hexadecimal formats, the default output format is an easier-to-read decimal format.

Every variable has two output formats, called its "print format" and "write format". Print formats are used in most output contexts; only the WRITE command uses write formats. Newly created variables have identical print and write formats, and FORMATS, the most commonly used command for changing formats, sets both of them to the same value as well. This means that the distinction between print and write formats is usually unimportant.

Input and output formats are specified to PSPP with a "format specification" of the form TypeW or TypeW.D, where Type is one of the format types described later, W is a field width measured in columns, and D is an optional number of decimal places. If D is omitted, a value of 0 is assumed. Some formats do not allow a nonzero D to be specified.

Basic Numeric Formats

The basic numeric formats are used for input and output of real numbers in standard or scientific notation. The following table shows an example of how each format displays positive and negative numbers with the default decimal point setting:

Format3141.59-3141.59
F8.2 3141.59-3141.59
COMMA9.2 3,141.59-3,141.59
DOT9.2 3.141,59-3.141,59
DOLLAR10.2 $3,141.59-$3,141.59
PCT9.2 3141.59%-3141.59%
E8.1 3.1E+003-3.1E+003

On output, numbers in F format are expressed in standard decimal notation with the requested number of decimal places. The other formats output some variation on this style:

  • Numbers in COMMA format are additionally grouped every three digits by inserting a grouping character. The grouping character is ordinarily a comma, but it can be changed to a period (with SET DECIMAL).

  • DOT format is like COMMA format, but it interchanges the role of the decimal point and grouping characters. That is, the current grouping character is used as a decimal point and vice versa.

  • DOLLAR format is like COMMA format, but it prefixes the number with $.

  • PCT format is like F format, but adds % after the number.

  • The E format always produces output in scientific notation.

On input, the basic numeric formats accept positive and numbers in standard decimal notation or scientific notation. Leading and trailing spaces are allowed. An empty or all-spaces field, or one that contains only a single period, is treated as the system missing value.

In scientific notation, the exponent may be introduced by a sign (+ or -), or by one of the letters e or d (in uppercase or lowercase), or by a letter followed by a sign. A single space may follow the letter or the sign or both.

On fixed-format DATA LIST and in a few other contexts, decimals are implied when the field does not contain a decimal point. In F6.5 format, for example, the field 314159 is taken as the value 3.14159 with implied decimals. Decimals are never implied if an explicit decimal point is present or if scientific notation is used.

E and F formats accept the basic syntax already described. The other formats allow some additional variations:

  • COMMA, DOLLAR, and DOT formats ignore grouping characters within the integer part of the input field. The identity of the grouping character depends on the format.

  • DOLLAR format allows a dollar sign to precede the number. In a negative number, the dollar sign may precede or follow the minus sign.

  • PCT format allows a percent sign to follow the number.

    All of the basic number formats have a maximum field width of 40 and accept no more than 16 decimal places, on both input and output. Some additional restrictions apply:

  • As input formats, the basic numeric formats allow no more decimal places than the field width. As output formats, the field width must be greater than the number of decimal places; that is, large enough to allow for a decimal point and the number of requested decimal places. DOLLAR and PCT formats must allow an additional column for $ or %.

  • The default output format for a given input format increases the field width enough to make room for optional input characters. If an input format calls for decimal places, the width is increased by 1 to make room for an implied decimal point. COMMA, DOT, and DOLLAR formats also increase the output width to make room for grouping characters. DOLLAR and PCT further increase the output field width by 1 to make room for $ or %. The increased output width is capped at 40, the maximum field width.

  • The E format is exceptional. For output, E format has a minimum width of 7 plus the number of decimal places. The default output format for an E input format is an E format with at least 3 decimal places and thus a minimum width of 10.

More details of basic numeric output formatting are given below:

  • Output rounds to nearest, with ties rounded away from zero. Thus, 2.5 is output as 3 in F1.0 format, and -1.125 as -1.13 in F5.1 format.

  • The system-missing value is output as a period in a field of spaces, placed in the decimal point's position, or in the rightmost column if no decimal places are requested. A period is used even if the decimal point character is a comma.

  • A number that does not fill its field is right-justified within the field.

  • A number is too large for its field causes decimal places to be dropped to make room. If dropping decimals does not make enough room, scientific notation is used if the field is wide enough. If a number does not fit in the field, even in scientific notation, the overflow is indicated by filling the field with asterisks (*).

  • COMMA, DOT, and DOLLAR formats insert grouping characters only if space is available for all of them. Grouping characters are never inserted when all decimal places must be dropped. Thus, 1234.56 in COMMA5.2 format is output as 1235 without a comma, even though there is room for one, because all decimal places were dropped.

  • DOLLAR or PCT format drop the $ or % only if the number would not fit at all without it. Scientific notation with $ or % is preferred to ordinary decimal notation without it.

  • Except in scientific notation, a decimal point is included only when it is followed by a digit. If the integer part of the number being output is 0, and a decimal point is included, then PSPP ordinarily drops the zero before the decimal point. However, in F, COMMA, or DOT formats, PSPP keeps the zero if SET LEADZERO is set to ON.

    In scientific notation, the number always includes a decimal point, even if it is not followed by a digit.

  • A negative number includes a minus sign only in the presence of a nonzero digit: -0.01 is output as -.01 in F4.2 format but as .0 in F4.1 format. Thus, a "negative zero" never includes a minus sign.

  • In negative numbers output in DOLLAR format, the dollar sign follows the negative sign. Thus, -9.99 in DOLLAR6.2 format is output as -$9.99.

  • In scientific notation, the exponent is output as E followed by + or - and exactly three digits. Numbers with magnitude less than 10**-999 or larger than 10**999 are not supported by most computers, but if they are supported then their output is considered to overflow the field and they are output as asterisks.

  • On most computers, no more than 15 decimal digits are significant in output, even if more are printed. In any case, output precision cannot be any higher than input precision; few data sets are accurate to 15 digits of precision. Unavoidable loss of precision in intermediate calculations may also reduce precision of output.

  • Special values such as infinities and "not a number" values are usually converted to the system-missing value before printing. In a few circumstances, these values are output directly. In fields of width 3 or greater, special values are output as however many characters fit from +Infinity or -Infinity for infinities, from NaN for "not a number," or from Unknown for other values (if any are supported by the system). In fields under 3 columns wide, special values are output as asterisks.

Custom Currency Formats

The custom currency formats are closely related to the basic numeric formats, but they allow users to customize the output format. The SET command configures custom currency formats, using the syntax

SET CCX="STRING".

where X is A, B, C, D, or E, and STRING is no more than 16 characters long.

STRING must contain exactly three commas or exactly three periods (but not both), except that a single quote character may be used to "escape" a following comma, period, or single quote. If three commas are used, commas are used for grouping in output, and a period is used as the decimal point. Uses of periods reverses these roles.

The commas or periods divide STRING into four fields, called the "negative prefix", "prefix", "suffix", and "negative suffix", respectively. The prefix and suffix are added to output whenever space is available. The negative prefix and negative suffix are always added to a negative number when the output includes a nonzero digit.

The following syntax shows how custom currency formats could be used to reproduce basic numeric formats:

SET CCA="-,,,".  /* Same as COMMA.
SET CCB="-...".  /* Same as DOT.
SET CCC="-,$,,". /* Same as DOLLAR.
SET CCD="-,,%,". /* Like PCT, but groups with commas.

Here are some more examples of custom currency formats. The final example shows how to use a single quote to escape a delimiter:

SET CCA=",EUR,,-".   /* Euro.
SET CCB="(,USD ,,)". /* US dollar.
SET CCC="-.R$..".    /* Brazilian real.
SET CCD="-,, NIS,".  /* Israel shekel.
SET CCE="-.Rp'. ..". /* Indonesia Rupiah.

These formats would yield the following output:

Format 3145.59-3145.59
CCA12.2 EUR3,145.59EUR3,145.59-
CCB14.2 USD 3,145.59(USD 3,145.59)
CCC11.2 R$3.145,59-R$3.145,59
CCD13.2 3,145.59 NIS-3,145.59 NIS
CCE10.0 Rp. 3.146-Rp. 3.146

The default for all the custom currency formats is -,,,, equivalent to COMMA format.

Legacy Numeric Formats

The N and Z numeric formats provide compatibility with legacy file formats. They have much in common:

  • Output is rounded to the nearest representable value, with ties rounded away from zero.

  • Numbers too large to display are output as a field filled with asterisks (*).

  • The decimal point is always implicitly the specified number of digits from the right edge of the field, except that Z format input allows an explicit decimal point.

  • Scientific notation may not be used.

  • The system-missing value is output as a period in a field of spaces. The period is placed just to the right of the implied decimal point in Z format, or at the right end in N format or in Z format if no decimal places are requested. A period is used even if the decimal point character is a comma.

  • Field width may range from 1 to 40. Decimal places may range from 0 up to the field width, to a maximum of 16.

  • When a legacy numeric format used for input is converted to an output format, it is changed into the equivalent F format. The field width is increased by 1 if any decimal places are specified, to make room for a decimal point. For Z format, the field width is increased by 1 more column, to make room for a negative sign. The output field width is capped at 40 columns.

N Format

The N format supports input and output of fields that contain only digits. On input, leading or trailing spaces, a decimal point, or any other non-digit character causes the field to be read as the system-missing value. As a special exception, an N format used on DATA LIST FREE or DATA LIST LIST is treated as the equivalent F format.

On output, N pads the field on the left with zeros. Negative numbers are output like the system-missing value.

Z Format

The Z format is a "zoned decimal" format used on IBM mainframes. Z format encodes the sign as part of the final digit, which must be one of the following:

0123456789
{ABCDEFGHI
}JKLMNOPQR

where the characters on each line represent digits 0 through 9 in order. Characters on the first two lines indicate a positive sign; those on the third indicate a negative sign.

On output, Z fields are padded on the left with spaces. On input, leading and trailing spaces are ignored. Any character in an input field other than spaces, the digit characters above, and . causes the field to be read as system-missing.

The decimal point character for input and output is always ., even if the decimal point character is a comma (see SET DECIMAL).

Nonzero, negative values output in Z format are marked as negative even when no nonzero digits are output. For example, -0.2 is output in Z1.0 format as J. The "negative zero" value supported by most machines is output as positive.

Binary and Hexadecimal Numeric Formats

The binary and hexadecimal formats are primarily designed for compatibility with existing machine formats, not for human readability. All of them therefore have a F format as default output format. Some of these formats are only portable between machines with compatible byte ordering (endianness).

Binary formats use byte values that in text files are interpreted as special control functions, such as carriage return and line feed. Thus, data in binary formats should not be included in syntax files or read from data files with variable-length records, such as ordinary text files. They may be read from or written to data files with fixed-length records. See FILE HANDLE, for information on working with fixed-length records.

P and PK Formats

These are binary-coded decimal formats, in which every byte (except the last, in P format) represents two decimal digits. The most-significant 4 bits of the first byte is the most-significant decimal digit, the least-significant 4 bits of the first byte is the next decimal digit, and so on.

In P format, the most-significant 4 bits of the last byte are the least-significant decimal digit. The least-significant 4 bits represent the sign: decimal 15 indicates a negative value, decimal 13 indicates a positive value.

Numbers are rounded downward on output. The system-missing value and numbers outside representable range are output as zero.

The maximum field width is 16. Decimal places may range from 0 up to the number of decimal digits represented by the field.

The default output format is an F format with twice the input field width, plus one column for a decimal point (if decimal places were requested).

IB and PIB Formats

These are integer binary formats. IB reads and writes 2's complement binary integers, and PIB reads and writes unsigned binary integers. The byte ordering is by default the host machine's, but SET RIB may be used to select a specific byte ordering for reading and SET WIB, similarly, for writing.

The maximum field width is 8. Decimal places may range from 0 up to the number of decimal digits in the largest value representable in the field width.

The default output format is an F format whose width is the number of decimal digits in the largest value representable in the field width, plus 1 if the format has decimal places.

RB Format

This is a binary format for real numbers. It reads and writes the host machine's floating-point format. The byte ordering is by default the host machine's, but SET RIB may be used to select a specific byte ordering for reading and SET WIB, similarly, for writing.

The field width should be 4, for 32-bit floating-point numbers, or 8, for 64-bit floating-point numbers. Other field widths do not produce useful results. The maximum field width is 8. No decimal places may be specified.

The default output format is F8.2.

PIBHEX and RBHEX Formats

These are hexadecimal formats, for reading and writing binary formats where each byte has been recoded as a pair of hexadecimal digits.

A hexadecimal field consists solely of hexadecimal digits 0...9 and A...F. Uppercase and lowercase are accepted on input; output is in uppercase.

Other than the hexadecimal representation, these formats are equivalent to PIB and RB formats, respectively. However, bytes in PIBHEX format are always ordered with the most-significant byte first (big-endian order), regardless of the host machine's native byte order or PSPP settings.

Field widths must be even and between 2 and 16. RBHEX format allows no decimal places; PIBHEX allows as many decimal places as a PIB format with half the given width.

Time and Date Formats

In PSPP, a "time" is an interval. The time formats translate between human-friendly descriptions of time intervals and PSPP's internal representation of time intervals, which is simply the number of seconds in the interval. PSPP has three time formats:

Time FormatTemplateExample
MTIMEMM:SS.ss91:17.01
TIMEhh:MM:SS.ss01:31:17.01
DTIMEDD HH:MM:SS.ss00 04:31:17.01

A "date" is a moment in the past or the future. Internally, PSPP represents a date as the number of seconds since the "epoch", midnight, Oct. 14, 1582. The date formats translate between human-readable dates and PSPP's numeric representation of dates and times. PSPP has several date formats:

Date FormatTemplateExample
DATEdd-mmm-yyyy01-OCT-1978
ADATEmm/dd/yyyy10/01/1978
EDATEdd.mm.yyyy01.10.1978
JDATEyyyyjjj1978274
SDATEyyyy/mm/dd1978/10/01
QYRq Q yyyy3 Q 1978
MOYRmmm yyyyOCT 1978
WKYRww WK yyyy40 WK 1978
DATETIMEdd-mmm-yyyy HH:MM:SS.ss01-OCT-1978 04:31:17.01
YMDHMSyyyy-mm-dd HH:MM:SS.ss1978-01-OCT 04:31:17.01

The templates in the preceding tables describe how the time and date formats are input and output:

  • dd
    Day of month, from 1 to 31. Always output as two digits.

  • mm
    mmm
    Month. In output, mm is output as two digits, mmm as the first three letters of an English month name (January, February, ...). In input, both of these formats, plus Roman numerals, are accepted.

  • yyyy
    Year. In output, DATETIME and YMDHMS always produce 4-digit years; other formats can produce a 2- or 4-digit year. The century assumed for 2-digit years depends on the EPOCH setting. In output, a year outside the epoch causes the whole field to be filled with asterisks (*).

  • jjj
    Day of year (Julian day), from 1 to 366. This is exactly three digits giving the count of days from the start of the year. January 1 is considered day 1.

  • q
    Quarter of year, from 1 to 4. Quarters start on January 1, April 1, July 1, and October 1.

  • ww
    Week of year, from 1 to 53. Output as exactly two digits. January 1 is the first day of week 1.

  • DD
    Count of days, which may be positive or negative. Output as at least two digits.

  • hh
    Count of hours, which may be positive or negative. Output as at least two digits.

  • HH
    Hour of day, from 0 to 23. Output as exactly two digits.

  • MM
    In MTIME, count of minutes, which may be positive or negative. Output as at least two digits.

    In other formats, minute of hour, from 0 to 59. Output as exactly two digits.

  • SS.ss
    Seconds within minute, from 0 to 59. The integer part is output as exactly two digits. On output, seconds and fractional seconds may or may not be included, depending on field width and decimal places. On input, seconds and fractional seconds are optional. The DECIMAL setting controls the character accepted and displayed as the decimal point (see SET DECIMAL).

    For output, the date and time formats use the delimiters indicated in the table. For input, date components may be separated by spaces or by one of the characters -, /, ., or ,, and time components may be separated by spaces or :. On input, the Q separating quarter from year and the WK separating week from year may be uppercase or lowercase, and the spaces around them are optional.

    On input, all time and date formats accept any amount of leading and trailing white space.

    The maximum width for time and date formats is 40 columns. Minimum input and output width for each of the time and date formats is shown below:

FormatMin. Input WidthMin. Output WidthOption
DATE894-digit year
ADATE884-digit year
EDATE884-digit year
JDATE554-digit year
SDATE884-digit year
QYR464-digit year
MOYR664-digit year
WKYR684-digit year
DATETIME1717seconds
YMDHMS1216seconds
MTIME45
TIME55seconds
DTIME88seconds

In the table, "Option" describes what increased output width enables:

  • "4-digit year": A field 2 columns wider than the minimum includes a 4-digit year. (DATETIME and YMDHMS formats always include a 4-digit year.)

  • "seconds": A field 3 columns wider than the minimum includes seconds as well as minutes. A field 5 columns wider than minimum, or more, can also include a decimal point and fractional seconds (but no more than allowed by the format's decimal places).

    For the time and date formats, the default output format is the same as the input format, except that PSPP increases the field width, if necessary, to the minimum allowed for output.

    Time or dates narrower than the field width are right-justified within the field.

    When a time or date exceeds the field width, characters are trimmed from the end until it fits. This can occur in an unusual situation, e.g. with a year greater than 9999 (which adds an extra digit), or for a negative value on MTIME, TIME, or DTIME (which adds a leading minus sign).

    The system-missing value is output as a period at the right end of the field.

Date Component Formats

The WKDAY and MONTH formats provide input and output for the names of weekdays and months, respectively.

On output, these formats convert a number between 1 and 7, for WKDAY, or between 1 and 12, for MONTH, into the English name of a day or month, respectively. If the name is longer than the field, it is trimmed to fit. If the name is shorter than the field, it is padded on the right with spaces. Values outside the valid range, and the system-missing value, are output as all spaces.

On input, English weekday or month names (in uppercase or lowercase) are converted back to their corresponding numbers. Weekday and month names may be abbreviated to their first 2 or 3 letters, respectively.

The field width may range from 2 to 40, for WKDAY, or from 3 to 40, for MONTH. No decimal places are allowed.

The default output format is the same as the input format.

String Formats

The A and AHEX formats are the only ones that may be assigned to string variables. Neither format allows any decimal places.

In A format, the entire field is treated as a string value. The field width may range from 1 to 32,767, the maximum string width. The default output format is the same as the input format.

In AHEX format, the field is composed of characters in a string encoded as hex digit pairs. On output, hex digits are output in uppercase; on input, uppercase and lowercase are both accepted. The default output format is A format with half the input width.

Scratch Variables

Most of the time, variables don't retain their values between cases. Instead, either they're being read from a data file or the active dataset, in which case they assume the value read, or, if created with COMPUTE or another transformation, they're initialized to the system-missing value or to blanks, depending on type.

However, sometimes it's useful to have a variable that keeps its value between cases. You can do this with LEAVE, or you can use a "scratch variable". Scratch variables are variables whose names begin with an octothorpe (#).

Scratch variables have the same properties as variables left with LEAVE: they retain their values between cases, and for the first case they are initialized to 0 or blanks. They have the additional property that they are deleted before the execution of any procedure. For this reason, scratch variables can't be used for analysis. To use a scratch variable in an analysis, use COMPUTE to copy its value into an ordinary variable, then use that ordinary variable in the analysis.

Files Used by PSPP

PSPP makes use of many files each time it runs. Some of these it reads, some it writes, some it creates. Here is a table listing the most important of these files:

  • command file
    syntax file
    These names (synonyms) refer to the file that contains instructions that tell PSPP what to do. The syntax file's name is specified on the PSPP command line. Syntax files can also be read with INCLUDE or INSERT.

  • data file
    Data files contain raw data in text or binary format. Data can also be embedded in a syntax file with BEGIN DATA and END DATA.

  • listing file
    One or more output files are created by PSPP each time it is run. The output files receive the tables and charts produced by statistical procedures. The output files may be in any number of formats, depending on how PSPP is configured.

  • system file
    System files are binary files that store a dictionary and a set of cases. GET and SAVE read and write system files.

  • portable file
    Portable files are files in a text-based format that store a dictionary and a set of cases. IMPORT and EXPORT read and write portable files.

File Handles

A "file handle" is a reference to a data file, system file, or portable file. Most often, a file handle is specified as the name of a file as a string, that is, enclosed within ' or ".

A file name string that begins or ends with | is treated as the name of a command to pipe data to or from. You can use this feature to read data over the network using a program such as curl (e.g. GET '|curl -s -S http://example.com/mydata.sav'), to read compressed data from a file using a program such as zcat (e.g. GET '|zcat mydata.sav.gz'), and for many other purposes.

PSPP also supports declaring named file handles with the FILE HANDLE command. This command associates an identifier of your choice (the file handle's name) with a file. Later, the file handle name can be substituted for the name of the file. When PSPP syntax accesses a file multiple times, declaring a named file handle simplifies updating the syntax later to use a different file. Use of FILE HANDLE is also required to read data files in binary formats.

In some circumstances, PSPP must distinguish whether a file handle refers to a system file or a portable file. When this is necessary to read a file, e.g. as an input file for GET or MATCH FILES, PSPP uses the file's contents to decide. In the context of writing a file, e.g. as an output file for SAVE or AGGREGATE, PSPP decides based on the file's name: if it ends in .por (with any capitalization), then PSPP writes a portable file; otherwise, PSPP writes a system file.

INLINE is reserved as a file handle name. It refers to the "data file" embedded into the syntax file between BEGIN DATA and END DATA.

The file to which a file handle refers may be reassigned on a later FILE HANDLE command if it is first closed using CLOSE FILE HANDLE.

Syntax Diagrams

The syntax of PSPP commands is presented in this manual with syntax diagrams.

A syntax diagram is a series of definitions of "nonterminals". Each nonterminal is defined its name, then ::=, then what the nonterminal consists of. If a nonterminal has multiple definitions, then any of them is acceptable. If the definition is empty, then one possible expansion of that nonterminal is nothing. Otherwise, the definition consists of a series of nonterminals and "terminals". The latter represent single tokens and consist of:

  • KEYWORD
    Any word written in uppercase is that literal syntax keyword.

  • number
    A real number.

  • integer
    An integer number.

  • string
    A string.

  • var-name
    A single variable name.

  • =, /, +, -, etc.
    Operators and punctuators.

  • .
    The end of the command. This is not necessarily an actual dot in the syntax file (see Forming Commands).

Some nonterminals are very common, so they are defined here in English for clarity:

  • var-list
    A list of one or more variable names or the keyword ALL.

  • expression
    An expression.

The first nonterminal defined in a syntax diagram for a command is the entire syntax for that command.

Mathematical Expressions

Expressions share a common syntax each place they appear in PSPP commands. Expressions are made up of "operands", which can be numbers, strings, variable names, or invocations of functions, separated by "operators".

Boolean Values

Some PSPP operators and expressions work with Boolean values, which represent true/false conditions. Booleans have only three possible values: 0 (false), 1 (true), and system-missing (unknown). System-missing is neither true nor false and indicates that the true value is unknown.

Boolean-typed operands or function arguments must take on one of these three values. Other values are considered false, but provoke a warning when the expression is evaluated.

Strings and Booleans are not compatible, and neither may be used in place of the other.

Missing Values

Most numeric operators yield system-missing when given any system-missing operand. A string operator given any system-missing operand typically results in the empty string. Exceptions are listed under particular operator descriptions.

String user-missing values are not treated specially in expressions.

User-missing values for numeric variables are always transformed into the system-missing value, except inside the arguments to the VALUE and SYSMIS functions.

The missing-value functions can be used to precisely control how missing values are treated in expressions.

Order of Operations

The following table describes operator precedence. Smaller-numbered levels in the table have higher precedence. Within a level, operations are always performed from left to right.

  1. ()
  2. **
  3. Unary + and -
  4. * /
  5. Binary + and -
  6. = >= > <= < <>
  7. NOT
  8. AND
  9. OR

Operators

Every operator takes one or more operands as input and yields exactly one result as output. Depending on the operator, operands accept strings or numbers as operands. With few exceptions, operands may be full-fledged expressions in themselves.

Grouping Operators

Parentheses (()) are the grouping operators. Surround an expression with parentheses to force early evaluation.

Parentheses also surround the arguments to functions, but in that situation they act as punctuators, not as operators.

Arithmetic Operators

The arithmetic operators take numeric operands and produce numeric results.

  • A + B
    A - B
    Addition and subtraction.

  • A * B
    Multiplication. If either A or B is 0, then the result is 0, even if the other operand is missing.

  • A / B
    Division. If A is 0, then the result is 0, even if B is missing. If B is zero, the result is system-missing.

  • A ** B
    A raised to the power B. If A is negative and B is not an integer, the result is system-missing. 0**0 is also system-missing.

  • -A
    Reverses the sign of A.

Logical Operators

The logical operators take logical operands and produce logical results, meaning "true or false." Logical operators are not true Boolean operators because they may also result in a system-missing value. See Boolean Values, above, for more information.

  • A AND B
    A & B
    True if both A and B are true, false otherwise. If one operand is false, the result is false even if the other is missing. If both operands are missing, the result is missing.

  • A OR B
    A | B
    True if at least one of A and B is true. If one operand is true, the result is true even if the other operand is missing. If both operands are missing, the result is missing.

  • NOT A
    ~A
    True if A is false. If the operand is missing, then the result is missing.

The overall truth table for the binary logical operators is:

ABA AND BA OR B
falsefalsefalsefalse
falsetruefalsetrue
truefalsefalsetrue
truetruetruetrue
falsemissingfalsemissing
truemissingmissingtrue
missingfalsefalsemissing
missingtruemissingtrue
missingmissingmissingmissing

Relational Operators

The relational operators take numeric or string operands and produce Boolean results.

Strings cannot be compared to numbers. When strings of different lengths are compared, the shorter string is right-padded with spaces to match the length of the longer string.

The results of string comparisons, other than tests for equality or inequality, depend on the character set in use. String comparisons are case-sensitive.

  • A EQ B
    A = B
    True if A is equal to B.

  • A LE B
    A <= B
    True if A is less than or equal to B.

  • A LT B
    A < B
    True if A is less than B.

  • A GE B
    A >= B
    True if A is greater than or equal to B.

  • A GT B
    A > B
    True if A is greater than B.

  • A NE B
    A ~= B
    A <> B
    True if A is not equal to B.

Functions

PSPP functions provide mathematical abilities above and beyond those possible using simple operators. Functions have a common syntax: each is composed of a function name followed by a left parenthesis, one or more arguments, and a right parenthesis.

Function names are not reserved. Their names are specially treated only when followed by a left parenthesis, so that EXP(10) refers to the constant value e raised to the 10th power, but EXP by itself refers to the value of a variable called EXP.

Mathematical Functions

Mathematical functions take numeric arguments and produce numeric results.

  • ABS(X)
    Results in the absolute value of X.

  • EXP(EXPONENT)
    Returns e (approximately 2.71828) raised to power EXPONENT.

  • LG10(X)
    Takes the base-10 logarithm of X. If X is not positive, the result is system-missing.

  • LN(X)
    Takes the base-e logarithm of X. If X is not positive, the result is system-missing.

  • LNGAMMA(X)
    Yields the base-e logarithm of the complete gamma of X. If X is a negative integer, the result is system-missing.

  • MOD(A, B)
    Returns the remainder (modulus) of A divided by B. If A is 0, then the result is 0, even if B is missing. If B is 0, the result is system-missing.

  • MOD10(X)
    Returns the remainder when X is divided by 10. If X is negative, MOD10(X) is negative or zero.

  • RND(X [, MULT[, FUZZBITS]])
    Rounds X and rounds it to a multiple of MULT (by default 1). Halves are rounded away from zero, as are values that fall short of halves by less than FUZZBITS of errors in the least-significant bits of X. If FUZZBITS is not specified then the default is taken from SET FUZZBITS, which is 6 unless overridden.

  • SQRT(X)
    Takes the square root of X. If X is negative, the result is system-missing.

  • TRUNC(X [, MULT[, FUZZBITS]])
    Rounds X to a multiple of MULT, toward zero. For the default MULT of 1, this is equivalent to discarding the fractional part of X. Values that fall short of a multiple of MULT by less than FUZZBITS of errors in the least-significant bits of X are rounded away from zero. If FUZZBITS is not specified then the default is taken from SET FUZZBITS, which is 6 unless overridden.

Trigonometric Functions

Trigonometric functions take numeric arguments and produce numeric results.

  • ARCOS(X)
    ACOS(X)
    Takes the arccosine, in radians, of X. Results in system-missing if X is not between -1 and 1 inclusive. This function is a PSPP extension.

  • ARSIN(X)
    ASIN(X)
    Takes the arcsine, in radians, of X. Results in system-missing if X is not between -1 and 1 inclusive.

  • ARTAN(X)
    ATAN(X)
    Takes the arctangent, in radians, of X.

  • COS(ANGLE)
    Takes the cosine of ANGLE which should be in radians.

  • SIN(ANGLE)
    Takes the sine of ANGLE which should be in radians.

  • TAN(ANGLE)
    Takes the tangent of ANGLE which should be in radians. Results in system-missing at values of ANGLE that are too close to odd multiples of π/2.

Missing-Value Functions

Missing-value functions take various numeric arguments and yield various types of results. Except where otherwise stated below, the normal rules of evaluation apply within expression arguments to these functions. In particular, user-missing values for numeric variables are converted to system-missing values.

  • MISSING(EXPR)
    When EXPR is simply the name of a numeric variable, returns 1 if the variable has the system-missing value or if it is user-missing. For any other value 0 is returned. If EXPR is any other kind of expression, the function returns 1 if the value is system-missing, 0 otherwise.

  • NMISS(EXPR [, EXPR]...)
    Each argument must be a numeric expression. Returns the number of system-missing values in the list, which may include variable ranges using the VAR1 TO VAR2 syntax.

  • NVALID(EXPR [, EXPR]...)
    Each argument must be a numeric expression. Returns the number of values in the list that are not system-missing. The list may include variable ranges using the VAR1 TO VAR2 syntax.

  • SYSMIS(EXPR)
    Returns 1 if EXPR has the system-missing value, 0 otherwise.

  • VALUE(VARIABLE)
    VALUE(VECTOR(INDEX))
    Prevents the user-missing values of the variable or vector element from being transformed into system-missing values, and always results in its actual value, whether it is valid, user-missing, or system-missing.

Set Membership Functions

Set membership functions determine whether a value is a member of a set. They take a set of numeric arguments or a set of string arguments, and produce Boolean results.

String comparisons are performed according to the rules given for Relational Operators. User-missing string values are treated as valid values.

  • ANY(VALUE, SET [, SET]...)
    Returns true if VALUE is equal to any of the SET values, and false otherwise. For numeric arguments, returns system-missing if VALUE is system-missing or if all the values in SET are system-missing.

  • RANGE(VALUE, LOW, HIGH [, LOW, HIGH]...)
    Returns true if VALUE is in any of the intervals bounded by LOW and HIGH, inclusive, and false otherwise. LOW and HIGH must be given in pairs. Returns system-missing if any HIGH is less than its LOW or, for numeric arguments, if VALUE is system-missing or if all the LOW-HIGH pairs contain one (or two) system-missing values. A pair does not match VALUE if either LOW or HIGH is missing, even if VALUE equals the non-missing endpoint.

Statistical Functions

Statistical functions compute descriptive statistics on a list of values. Some statistics can be computed on numeric or string values; other can only be computed on numeric values. Their results have the same type as their arguments. The current case's weight has no effect on statistical functions.

These functions' argument lists may include entire ranges of variables using the VAR1 TO VAR2 syntax.

Unlike most functions, statistical functions can return non-missing values even when some of their arguments are missing. Most statistical functions, by default, require only one non-missing value to have a non-missing return; CFVAR, SD, and VARIANCE require 2. These defaults can be increased (but not decreased) by appending a dot and the minimum number of valid arguments to the function name. For example, MEAN.3(X, Y, Z) would only return non-missing if all of X, Y, and Z were valid.

  • CFVAR(NUMBER, NUMBER[, ...])
    Results in the coefficient of variation of the values of NUMBER. (The coefficient of variation is the standard deviation divided by the mean.)

  • MAX(VALUE, VALUE[, ...])
    Results in the value of the greatest VALUE. The VALUEs may be numeric or string.

  • MEAN(NUMBER, NUMBER[, ...])
    Results in the mean of the values of NUMBER.

  • MEDIAN(NUMBER, NUMBER[, ...])
    Results in the median of the values of NUMBER. Given an even number of nonmissing arguments, yields the mean of the two middle values.

  • MIN(NUMBER, NUMBER[, ...])
    Results in the value of the least VALUE. The VALUEs may be numeric or string.

  • SD(NUMBER, NUMBER[, ...])
    Results in the standard deviation of the values of NUMBER.

  • SUM(NUMBER, NUMBER[, ...])
    Results in the sum of the values of NUMBER.

  • VARIANCE(NUMBER, NUMBER[, ...])
    Results in the variance of the values of NUMBER.

String Functions

String functions take various arguments and return various results.

  • CONCAT(STRING, STRING[, ...])
    Returns a string consisting of each STRING in sequence. CONCAT("abc", "def", "ghi") has a value of "abcdefghi". The resultant string is truncated to a maximum of 32767 bytes.

  • INDEX(HAYSTACK, NEEDLE)
    RINDEX(HAYSTACK, NEEDLE)
    Returns a positive integer indicating the position of the first (for INDEX) or last (for RINDEX) occurrence of NEEDLE in HAYSTACK. Returns 0 if HAYSTACK does not contain NEEDLE. Returns 1 if NEEDLE is the empty string.

  • INDEX(HAYSTACK, NEEDLES, NEEDLE_LEN)
    RINDEX(HAYSTACK, NEEDLE, NEEDLE_LEN)
    Divides NEEDLES into multiple needles, each with length NEEDLE_LEN, which must be a positive integer that evenly divides the length of NEEDLES. Searches HAYSTACK for the occurrences of each needle and returns a positive integer indicating the byte index of the beginning of the first (for INDEX) or last (for RINDEX) needle it finds. Returns 0 if HAYSTACK does not contain any of the needles, or if NEEDLES is the empty string.

  • LENGTH(STRING)
    Returns the number of bytes in STRING.

  • LOWER(STRING)
    Returns a string identical to STRING except that all uppercase letters are changed to lowercase letters. The definitions of "uppercase" and "lowercase" are encoding-dependent.

  • LPAD(STRING, LENGTH[, PADDING])
    RPAD(STRING, LENGTH[, PADDING])
    If STRING is at least LENGTH bytes long, these functions return STRING unchanged. Otherwise, they return STRING padded with PADDING on the left side (for LPAD) or right side (for RPAD) to LENGTH bytes. These functions report an error and return STRING unchanged if LENGTH is missing or bigger than 32767.

    The PADDING argument must not be an empty string and defaults to a space if not specified. If its length does not evenly fit the amount of space needed for padding, the returned string will be shorter than LENGTH.

  • LTRIM(STRING[, PADDING])
    RTRIM(STRING[, PADDING])
    These functions return STRING, after removing leading (for LTRIM) or trailing (for RTRIM) copies of PADDING. If PADDING is omitted, these functions remove spaces (but not tabs or other white space). These functions return STRING unchanged if PADDING is the empty string.

  • NUMBER(STRING, FORMAT)
    Returns the number produced when STRING is interpreted according to format specifier FORMAT. If the format width W is less than the length of STRING, then only the first W bytes in STRING are used, e.g. NUMBER("123", F3.0) and NUMBER("1234", F3.0) both have value 123. If W is greater than STRING's length, then it is treated as if it were right-padded with spaces. If STRING is not in the correct format for FORMAT, system-missing is returned.

  • REPLACE(HAYSTACK, NEEDLE, REPLACEMENT[, N])
    Returns string HAYSTACK with instances of NEEDLE replaced by REPLACEMENT. If nonnegative integer N is specified, it limits the maximum number of replacements; otherwise, all instances of NEEDLE are replaced.

  • STRING(NUMBER, FORMAT)
    Returns a string corresponding to NUMBER in the format given by format specifier FORMAT. For example, STRING(123.56, F5.1) has the value "123.6".

  • STRUNC(STRING, N)
    Returns STRING, first trimming it to at most N bytes, then removing trailing spaces (but not tabs or other white space). Returns an empty string if N is zero or negative, or STRING unchanged if N is missing.

  • SUBSTR(STRING, START)
    Returns a string consisting of the value of STRING from position START onward. Returns an empty string if START is system-missing, less than 1, or greater than the length of STRING.

  • SUBSTR(STRING, START, COUNT)
    Returns a string consisting of the first COUNT bytes from STRING beginning at position START. Returns an empty string if START or COUNT is system-missing, if START is less than 1 or greater than the number of bytes in STRING, or if COUNT is less than 1. Returns a string shorter than COUNT bytes if START + COUNT - 1 is greater than the number of bytes in STRING. Examples: SUBSTR("abcdefg", 3, 2) has value "cd"; SUBSTR("nonsense", 4, 10) has the value "sense".

  • UPCASE(STRING)
    Returns STRING, changing lowercase letters to uppercase letters.

Time and Date Functions

For compatibility, PSPP considers dates before 15 Oct 1582 invalid. Most time and date functions will not accept earlier dates.

Time and Date Representations

Times and dates are handled by PSPP as single numbers. A "time" is an interval. PSPP measures times in seconds. Thus, the following intervals correspond with the numeric values given:

IntervalNumeric Value
10 minutes600
1 hour3,600
1 day, 3 hours, 10 seconds97,210
40 days3,456,000

A "date", on the other hand, is a particular instant in the past or the future. PSPP represents a date as a number of seconds since midnight preceding 14 Oct 1582. Because midnight preceding the dates given below correspond with the numeric PSPP dates given:

DateNumeric Value
15 Oct 158286,400
4 Jul 17766,113,318,400
1 Jan 190010,010,390,400
1 Oct 197812,495,427,200
24 Aug 199513,028,601,600

Constructing Times

These functions take numeric arguments and return numeric values that represent times.

  • TIME.DAYS(NDAYS)
    Returns a time corresponding to NDAYS days.

  • TIME.HMS(NHOURS, NMINS, NSECS)
    Returns a time corresponding to NHOURS hours, NMINS minutes, and NSECS seconds. The arguments may not have mixed signs: if any of them are positive, then none may be negative, and vice versa.

Examining Times

These functions take numeric arguments in PSPP time format and give numeric results.

  • CTIME.DAYS(TIME)
    Results in the number of days and fractional days in TIME.

  • CTIME.HOURS(TIME)
    Results in the number of hours and fractional hours in TIME.

  • CTIME.MINUTES(TIME)
    Results in the number of minutes and fractional minutes in TIME.

  • CTIME.SECONDS(TIME)
    Results in the number of seconds and fractional seconds in TIME. (CTIME.SECONDS does nothing; CTIME.SECONDS(X) is equivalent to X.)

Constructing Dates

These functions take numeric arguments and give numeric results that represent dates. Arguments taken by these functions are:

  • DAY
    Refers to a day of the month between 1 and 31. Day 0 is also accepted and refers to the final day of the previous month. Days 29, 30, and 31 are accepted even in months that have fewer days and refer to a day near the beginning of the following month.

  • MONTH
    Refers to a month of the year between 1 and 12. Months 0 and 13 are also accepted and refer to the last month of the preceding year and the first month of the following year, respectively.

  • QUARTER
    Refers to a quarter of the year between 1 and 4. The quarters of the year begin on the first day of months 1, 4, 7, and 10.

  • WEEK
    Refers to a week of the year between 1 and 53.

  • YDAY
    Refers to a day of the year between 1 and 366.

  • YEAR
    Refers to a year, 1582 or greater. Years between 0 and 99 are treated according to the epoch set on SET EPOCH, by default beginning 69 years before the current date.

If these functions' arguments are out-of-range, they are correctly normalized before conversion to date format. Non-integers are rounded toward zero.

  • DATE.DMY(DAY, MONTH, YEAR)
    DATE.MDY(MONTH, DAY, YEAR)
    Results in a date value corresponding to the midnight before day DAY of month MONTH of year YEAR.

  • DATE.MOYR(MONTH, YEAR)
    Results in a date value corresponding to the midnight before the first day of month MONTH of year YEAR.

  • DATE.QYR(QUARTER, YEAR)
    Results in a date value corresponding to the midnight before the first day of quarter QUARTER of year YEAR.

  • DATE.WKYR(WEEK, YEAR)
    Results in a date value corresponding to the midnight before the first day of week WEEK of year YEAR.

  • DATE.YRDAY(YEAR, YDAY)
    Results in a date value corresponding to the day YDAY of year YEAR.

Examining Dates

These functions take numeric arguments in PSPP date or time format and give numeric results. These names are used for arguments:

  • DATE
    A numeric value in PSPP date format.

  • TIME
    A numeric value in PSPP time format.

  • TIME-OR-DATE
    A numeric value in PSPP time or date format.

The functions for examining dates are:

  • XDATE.DATE(TIME-OR-DATE)
    For a time, results in the time corresponding to the number of whole days DATE-OR-TIME includes. For a date, results in the date corresponding to the latest midnight at or before DATE-OR-TIME; that is, gives the date that DATE-OR-TIME is in.

  • XDATE.HOUR(TIME-OR-DATE)
    For a time, results in the number of whole hours beyond the number of whole days represented by DATE-OR-TIME. For a date, results in the hour(as an integer between 0 and 23) corresponding to DATE-OR-TIME.

  • XDATE.JDAY(DATE)
    Results in the day of the year (as an integer between 1 and 366) corresponding to DATE.

  • XDATE.MDAY(DATE)
    Results in the day of the month (as an integer between 1 and 31) corresponding to DATE.

  • XDATE.MINUTE(TIME-OR-DATE)
    Results in the number of minutes (as an integer between 0 and 59) after the last hour in TIME-OR-DATE.

  • XDATE.MONTH(DATE)
    Results in the month of the year (as an integer between 1 and 12) corresponding to DATE.

  • XDATE.QUARTER(DATE)
    Results in the quarter of the year (as an integer between 1 and 4) corresponding to DATE.

  • XDATE.SECOND(TIME-OR-DATE)
    Results in the number of whole seconds after the last whole minute (as an integer between 0 and 59) in TIME-OR-DATE.

  • XDATE.TDAY(DATE)
    Results in the number of whole days from 14 Oct 1582 to DATE.

  • XDATE.TIME(DATE)
    Results in the time of day at the instant corresponding to DATE, as a time value. This is the number of seconds since midnight on the day corresponding to DATE.

  • XDATE.WEEK(DATE)
    Results in the week of the year (as an integer between 1 and 53) corresponding to DATE.

  • XDATE.WKDAY(DATE)
    Results in the day of week (as an integer between 1 and 7) corresponding to DATE, where 1 represents Sunday.

  • XDATE.YEAR(DATE)
    Returns the year (as an integer 1582 or greater) corresponding to DATE.

Time and Date Arithmetic

Ordinary arithmetic operations on dates and times often produce sensible results. Adding a time to, or subtracting one from, a date produces a new date that much earlier or later. The difference of two dates yields the time between those dates. Adding two times produces the combined time. Multiplying a time by a scalar produces a time that many times longer. Since times and dates are just numbers, the ordinary addition and subtraction operators are employed for these purposes.

Adding two dates does not produce a useful result.

Dates and times may have very large values. Thus, it is not a good idea to take powers of these values; also, the accuracy of some procedures may be affected. If necessary, convert times or dates in seconds to some other unit, like days or years, before performing analysis.

PSPP supplies a few functions for date arithmetic:

  • DATEDIFF(DATE2, DATE1, UNIT)
    Returns the span of time from DATE1 to DATE2 in terms of UNIT, which must be a quoted string, one of years, quarters, months, weeks, days, hours, minutes, and seconds. The result is an integer, truncated toward zero.

    One year is considered to span from a given date to the same month, day, and time of day the next year. Thus, from January 1 of one year to January 1 the next year is considered to be a full year, but February 29 of a leap year to the following February 28 is not. Similarly, one month spans from a given day of the month to the same day of the following month. Thus, there is never a full month from Jan. 31 of a given year to any day in the following February.

  • DATESUM(DATE, QUANTITY, UNIT[, METHOD])
    Returns DATE advanced by the given QUANTITY of the specified UNIT, which must be one of the strings years, quarters, months, weeks, days, hours, minutes, and seconds.

    When UNIT is years, quarters, or months, only the integer part of QUANTITY is considered. Adding one of these units can cause the day of the month to exceed the number of days in the month. In this case, the METHOD comes into play: if it is omitted or specified as closest (as a quoted string), then the resulting day is the last day of the month; otherwise, if it is specified as rollover, then the extra days roll over into the following month.

    When UNIT is weeks, days, hours, minutes, or seconds, the QUANTITY is not rounded to an integer and METHOD, if specified, is ignored.

Miscellaneous Functions

  • LAG(VARIABLE[, N])
    VARIABLE must be a numeric or string variable name. LAG yields the value of that variable for the case N before the current one. Results in system-missing (for numeric variables) or blanks (for string variables) for the first N cases.

    LAG obtains values from the cases that become the new active dataset after a procedure executes. Thus, LAG will not return values from cases dropped by transformations such as SELECT IF, and transformations like COMPUTE that modify data will change the values returned by LAG. These are both the case whether these transformations precede or follow the use of LAG.

    If LAG is used before TEMPORARY, then the values it returns are those in cases just before TEMPORARY. LAG may not be used after TEMPORARY.

    If omitted, N defaults to 1. Otherwise, N must be a small positive constant integer. There is no explicit limit, but use of a large value will increase memory consumption.

  • YRMODA(YEAR, MONTH, DAY)
    YEAR is a year, either between 0 and 99 or at least 1582. Unlike other PSPP date functions, years between 0 and 99 always correspond to 1900 through 1999. MONTH is a month between 1 and 13. DAY is a day between 0 and 31. A DAY of 0 refers to the last day of the previous month, and a MONTH of 13 refers to the first month of the next year. YEAR must be in range. YEAR, MONTH, and DAY must all be integers.

    YRMODA results in the number of days between 15 Oct 1582 and the date specified, plus one. The date passed to YRMODA must be on or after 15 Oct 1582. 15 Oct 1582 has a value of 1.

  • VALUELABEL(VARIABLE)
    Returns a string matching the label associated with the current value of VARIABLE. If the current value of VARIABLE has no associated label, then this function returns the empty string. VARIABLE may be a numeric or string variable.

Statistical Distribution Functions

PSPP can calculate several functions of standard statistical distributions. These functions are named systematically based on the function and the distribution. The table below describes the statistical distribution functions in general:

  • PDF.DIST(X[, PARAM...])
    Probability density function for DIST. The domain of X depends on DIST. For continuous distributions, the result is the density of the probability function at X, and the range is nonnegative real numbers. For discrete distributions, the result is the probability of X.

  • CDF.DIST(X[, PARAM...])
    Cumulative distribution function for DIST, that is, the probability that a random variate drawn from the distribution is less than X. The domain of X depends DIST. The result is a probability.

  • SIG.DIST(X[, PARAM...)
    Tail probability function for DIST, that is, the probability that a random variate drawn from the distribution is greater than X. The domain of X depends DIST. The result is a probability. Only a few distributions include an SIG function.

  • IDF.DIST(P[, PARAM...])
    Inverse distribution function for DIST, the value of X for which the CDF would yield P. The value of P is a probability. The range depends on DIST and is identical to the domain for the corresponding CDF.

  • RV.DIST([PARAM...])
    Random variate function for DIST. The range depends on the distribution.

  • NPDF.DIST(X[, PARAM...])
    Noncentral probability density function. The result is the density of the given noncentral distribution at X. The domain of X depends on DIST. The range is nonnegative real numbers. Only a few distributions include an NPDF function.

  • NCDF.DIST(X[, PARAM...])
    Noncentral cumulative distribution function for DIST, that is, the probability that a random variate drawn from the given noncentral distribution is less than X. The domain of X depends DIST. The result is a probability. Only a few distributions include an NCDF function.

Continuous Distributions

The following continuous distributions are available:

  • PDF.BETA(X)
    CDF.BETA(X, A, B)
    IDF.BETA(P, A, B)
    RV.BETA(A, B)
    NPDF.BETA(X, A, B, ꟛ)
    NCDF.BETA(X, A, B, ꟛ)
    Beta distribution with shape parameters A and B. The noncentral distribution takes an additional parameter ꟛ. Constraints: A > 0, B > 0, ꟛ >= 0, 0 <= X <= 1, 0 <= P <= 1.

  • PDF.BVNOR(X0, X1, ρ)
    CDF.BVNOR(X0, X1, ρ)
    Bivariate normal distribution of two standard normal variables with correlation coefficient ρ. Two variates X0 and X1 must be provided. Constraints: 0 <= ρ <= 1, 0 <= P <= 1.

  • PDF.CAUCHY(X, A, B)
    CDF.CAUCHY(X, A, B)
    IDF.CAUCHY(P, A, B)
    RV.CAUCHY(A, B)
    Cauchy distribution with location parameter A and scale parameter B. Constraints: B > 0, 0 < P < 1.

  • CDF.CHISQ(X, DF)
    SIG.CHISQ(X, DF)
    IDF.CHISQ(P, DF)
    RV.CHISQ(DF)
    NCDF.CHISQ(X, DF, ꟛ)
    Chi-squared distribution with DF degrees of freedom. The noncentral distribution takes an additional parameter ꟛ. Constraints: DF > 0, ꟛ > 0, X >= 0, 0 <= P < 1.

  • PDF.EXP(X, A)
    CDF.EXP(X, A)
    IDF.EXP(P, A)
    RV.EXP(A)
    Exponential distribution with scale parameter A. The inverse of A represents the rate of decay. Constraints: A > 0, X >= 0, 0 <= P < 1.

  • PDF.XPOWER(X, A, B)
    RV.XPOWER(A, B)
    Exponential power distribution with positive scale parameter A and nonnegative power parameter B. Constraints: A > 0, B >= 0, X >= 0, 0 <= P <= 1. This distribution is a PSPP extension.

  • PDF.F(X, DF1, DF2)
    CDF.F(X, DF1, DF2)
    SIG.F(X, DF1, DF2)
    IDF.F(P, DF1, DF2)
    RV.F(DF1, DF2)
    F-distribution of two chi-squared deviates with DF1 and DF2 degrees of freedom. The noncentral distribution takes an additional parameter ꟛ. Constraints: DF1 > 0, DF2 > 0, ꟛ >= 0, X >= 0, 0 <= P < 1.

  • PDF.GAMMA(X, A, B)
    CDF.GAMMA(X, A, B)
    IDF.GAMMA(P, A, B)
    RV.GAMMA(A, B)
    Gamma distribution with shape parameter A and scale parameter B. Constraints: A > 0, B > 0, X >= 0, 0 <= P < 1.

  • PDF.LANDAU(X)
    RV.LANDAU()
    Landau distribution.

  • PDF.LAPLACE(X, A, B)
    CDF.LAPLACE(X, A, B)
    IDF.LAPLACE(P, A, B)
    RV.LAPLACE(A, B)
    Laplace distribution with location parameter A and scale parameter B. Constraints: B > 0, 0 < P < 1.

  • RV.LEVY(C, ɑ)
    Levy symmetric alpha-stable distribution with scale C and exponent ɑ. Constraints: 0 < ɑ <= 2.

  • RV.LVSKEW(C, ɑ, β)
    Levy skew alpha-stable distribution with scale C, exponent ɑ, and skewness parameter β. Constraints: 0 < ɑ <= 2, -1 <= β <= 1.

  • PDF.LOGISTIC(X, A, B)
    CDF.LOGISTIC(X, A, B)
    IDF.LOGISTIC(P, A, B)
    RV.LOGISTIC(A, B)
    Logistic distribution with location parameter A and scale parameter B. Constraints: B > 0, 0 < P < 1.

  • PDF.LNORMAL(X, A, B)
    CDF.LNORMAL(X, A, B)
    IDF.LNORMAL(P, A, B)
    RV.LNORMAL(A, B)
    Lognormal distribution with parameters A and B. Constraints: A > 0, B > 0, X >= 0, 0 <= P < 1.

  • PDF.NORMAL(X, μ, σ)
    CDF.NORMAL(X, μ, σ)
    IDF.NORMAL(P, μ, σ)
    RV.NORMAL(μ, σ)
    Normal distribution with mean μ and standard deviation σ. Constraints: B > 0, 0 < P < 1. Three additional functions are available as shorthand:

    • CDFNORM(X)
      Equivalent to CDF.NORMAL(X, 0, 1).

    • PROBIT(P)
      Equivalent to IDF.NORMAL(P, 0, 1).

    • NORMAL(σ)
      Equivalent to RV.NORMAL(0, σ).

  • PDF.NTAIL(X, A, σ)
    RV.NTAIL(A, σ)
    Normal tail distribution with lower limit A and standard deviation σ. This distribution is a PSPP extension. Constraints: A > 0, X > A, 0 < P < 1.

  • PDF.PARETO(X, A, B)
    CDF.PARETO(X, A, B)
    IDF.PARETO(P, A, B)
    RV.PARETO(A, B)
    Pareto distribution with threshold parameter A and shape parameter B. Constraints: A > 0, B > 0, X >= A, 0 <= P < 1.

  • PDF.RAYLEIGH(X, σ)
    CDF.RAYLEIGH(X, σ)
    IDF.RAYLEIGH(P, σ)
    RV.RAYLEIGH(σ)
    Rayleigh distribution with scale parameter σ. This distribution is a PSPP extension. Constraints: σ > 0, X > 0.

  • PDF.RTAIL(X, A, σ)
    RV.RTAIL(A, σ)
    Rayleigh tail distribution with lower limit A and scale parameter σ. This distribution is a PSPP extension. Constraints: A > 0, σ > 0, X > A.

  • PDF.T(X, DF)
    CDF.T(X, DF)
    IDF.T(P, DF)
    RV.T(DF)
    T-distribution with DF degrees of freedom. The noncentral distribution takes an additional parameter ꟛ. Constraints: DF > 0, 0 < P < 1.

  • PDF.T1G(X, A, B)
    CDF.T1G(X, A, B)
    IDF.T1G(P, A, B)
    Type-1 Gumbel distribution with parameters A and B. This distribution is a PSPP extension. Constraints: 0 < P < 1.

  • PDF.T2G(X, A, B)
    CDF.T2G(X, A, B)
    IDF.T2G(P, A, B)
    Type-2 Gumbel distribution with parameters A and B. This distribution is a PSPP extension. Constraints: X > 0, 0 < P < 1.

  • PDF.UNIFORM(X, A, B)
    CDF.UNIFORM(X, A, B)
    IDF.UNIFORM(P, A, B)
    RV.UNIFORM(A, B)
    Uniform distribution with parameters A and B. Constraints: A <= X <= B, 0 <= P <= 1. An additional function is available as shorthand:

    • UNIFORM(B)
      Equivalent to RV.UNIFORM(0, B).

  • PDF.WEIBULL(X, A, B)
    CDF.WEIBULL(X, A, B)
    IDF.WEIBULL(P, A, B)
    RV.WEIBULL(A, B)
    Weibull distribution with parameters A and B. Constraints: A > 0, B > 0, X >= 0, 0 <= P < 1.

Discrete Distributions

The following discrete distributions are available:

  • PDF.BERNOULLI(X)
    CDF.BERNOULLI(X, P)
    RV.BERNOULLI(P)
    Bernoulli distribution with probability of success P. Constraints: X = 0 or 1, 0 <= P <= 1.

  • PDF.BINOM(X, N, P)
    CDF.BINOM(X, N, P)
    RV.BINOM(N, P)
    Binomial distribution with N trials and probability of success P. Constraints: integer N > 0, 0 <= P <= 1, integer X <= N.

  • PDF.GEOM(X, N, P)
    CDF.GEOM(X, N, P)
    RV.GEOM(N, P)
    Geometric distribution with probability of success P. Constraints: 0 <= P <= 1, integer X > 0.

  • PDF.HYPER(X, A, B, C)
    CDF.HYPER(X, A, B, C)
    RV.HYPER(A, B, C)
    Hypergeometric distribution when B objects out of A are drawn and C of the available objects are distinctive. Constraints: integer A > 0, integer B <= A, integer C <= A, integer X >= 0.

  • PDF.LOG(X, P)
    RV.LOG(P)
    Logarithmic distribution with probability parameter P. Constraints: 0 <= P < 1, X >= 1.

  • PDF.NEGBIN(X, N, P)
    CDF.NEGBIN(X, N, P)
    RV.NEGBIN(N, P)
    Negative binomial distribution with number of successes parameter N and probability of success parameter P. Constraints: integer N >= 0, 0 < P <= 1, integer X >= 1.

  • PDF.POISSON(X, μ)
    CDF.POISSON(X, μ)
    RV.POISSON(μ)
    Poisson distribution with mean μ. Constraints: μ > 0, integer X >= 0.

System Variables

The system variables described below may be used only in expressions.

  • $CASENUM
    Case number of the case being processed. This changes as cases are added, deleted, and reordered.

  • $DATE
    Date the PSPP process was started, in format A9, following the pattern DD-MMM-YY.

  • $DATE11
    Date the PSPP process was started, in format A11, following the pattern DD-MMM-YYYY.

  • $JDATE
    Number of days between 15 Oct 1582 and the time the PSPP process was started.

  • $LENGTH
    Page length, in lines, in format F11.

  • $SYSMIS
    System missing value, in format F1.

  • $TIME
    Number of seconds between midnight 14 Oct 1582 and the time the active dataset was read, in format F20.

  • $WIDTH
    Page width, in characters, in format F3.

Data Input and Output

Data are the focus of the PSPP language. Each datum belongs to a “case” (also called an “observation”). Each case represents an individual or "experimental unit". For example, in the results of a survey, the names of the respondents, their sex, age, etc. and their responses are all data and the data pertaining to single respondent is a case. This chapter examines the PSPP commands for defining variables and reading and writing data. There are alternative commands to read data from predefined sources such as system files or databases.

These commands tell PSPP how to read data, but the data will not actually be read until a procedure is executed.

BEGIN DATA…END DATA

BEGIN DATA.
...
END DATA.

BEGIN DATA and END DATA can be used to embed raw ASCII data in a PSPP syntax file. DATA LIST or another input procedure must be used before BEGIN DATA. BEGIN DATA and END DATA must be used together. END DATA must appear by itself on a single line, with no leading white space and exactly one space between the words END and DATA.

CLOSE FILE HANDLE

CLOSE FILE HANDLE HANDLE_NAME.

CLOSE FILE HANDLE disassociates the name of a file handle with a given file. The only specification is the name of the handle to close. Afterward FILE HANDLE.

The file named INLINE, which represents data entered between BEGIN DATA and END DATA, cannot be closed. Attempts to close it with CLOSE FILE HANDLE have no effect.

CLOSE FILE HANDLE is a PSPP extension.

DATAFILE ATTRIBUTE

DATAFILE ATTRIBUTE
         ATTRIBUTE=NAME('VALUE') [NAME('VALUE')]...
         ATTRIBUTE=NAME[INDEX]('VALUE') [NAME[INDEX]('VALUE')]...
         DELETE=NAME [NAME]...
         DELETE=NAME[INDEX] [NAME[INDEX]]...

DATAFILE ATTRIBUTE adds, modifies, or removes user-defined attributes associated with the active dataset. Custom data file attributes are not interpreted by PSPP, but they are saved as part of system files and may be used by other software that reads them.

Use the ATTRIBUTE subcommand to add or modify a custom data file attribute. Specify the name of the attribute, followed by the desired value, in parentheses, as a quoted string. Attribute names that begin with $ are reserved for PSPP's internal use, and attribute names that begin with @ or $@ are not displayed by most PSPP commands that display other attributes. Other attribute names are not treated specially.

Attributes may also be organized into arrays. To assign to an array element, add an integer array index enclosed in square brackets ([ and ]) between the attribute name and value. Array indexes start at 1, not 0. An attribute array that has a single element (number 1) is not distinguished from a non-array attribute.

Use the DELETE subcommand to delete an attribute. Specify an attribute name by itself to delete an entire attribute, including all array elements for attribute arrays. Specify an attribute name followed by an array index in square brackets to delete a single element of an attribute array. In the latter case, all the array elements numbered higher than the deleted element are shifted down, filling the vacated position.

To associate custom attributes with particular variables, instead of with the entire active dataset, use VARIABLE ATTRIBUTE instead.

DATAFILE ATTRIBUTE takes effect immediately. It is not affected by conditional and looping structures such as DO IF or LOOP.

DATASET commands

DATASET NAME NAME [WINDOW={ASIS,FRONT}].
DATASET ACTIVATE NAME [WINDOW={ASIS,FRONT}].
DATASET COPY NAME [WINDOW={MINIMIZED,HIDDEN,FRONT}].
DATASET DECLARE NAME [WINDOW={MINIMIZED,HIDDEN,FRONT}].
DATASET CLOSE {NAME,*,ALL}.
DATASET DISPLAY.

The DATASET commands simplify use of multiple datasets within a PSPP session. They allow datasets to be created and destroyed. At any given time, most PSPP commands work with a single dataset, called the active dataset.

The DATASET NAME command gives the active dataset the specified name, or if it already had a name, it renames it. If another dataset already had the given name, that dataset is deleted.

The DATASET ACTIVATE command selects the named dataset, which must already exist, as the active dataset. Before switching the active dataset, any pending transformations are executed, as if EXECUTE had been specified. If the active dataset is unnamed before switching, then it is deleted and becomes unavailable after switching.

The DATASET COPY command creates a new dataset with the specified name, whose contents are a copy of the active dataset. Any pending transformations are executed, as if EXECUTE had been specified, before making the copy. If a dataset with the given name already exists, it is replaced. If the name is the name of the active dataset, then the active dataset becomes unnamed.

The DATASET DECLARE command creates a new dataset that is initially "empty," that is, it has no dictionary or data. If a dataset with the given name already exists, this has no effect. The new dataset can be used with commands that support output to a dataset, such as. AGGREGATE.

The DATASET CLOSE command deletes a dataset. If the active dataset is specified by name, or if * is specified, then the active dataset becomes unnamed. If a different dataset is specified by name, then it is deleted and becomes unavailable. Specifying ALL deletes all datasets except for the active dataset, which becomes unnamed.

The DATASET DISPLAY command lists all the currently defined datasets.

Many DATASET commands accept an optional WINDOW subcommand. In the PSPPIRE GUI, the value given for this subcommand influences how the dataset's window is displayed. Outside the GUI, the WINDOW subcommand has no effect. The valid values are:

  • ASIS
    Do not change how the window is displayed. This is the default for DATASET NAME and DATASET ACTIVATE.

  • FRONT
    Raise the dataset's window to the top. Make it the default dataset for running syntax.

  • MINIMIZED
    Display the window "minimized" to an icon. Prefer other datasets for running syntax. This is the default for DATASET COPY and DATASET DECLARE.

  • HIDDEN
    Hide the dataset's window. Prefer other datasets for running syntax.

DATA LIST

Used to read text or binary data, DATA LIST is the most fundamental data-reading command. Even the more sophisticated input methods use DATA LIST commands as a building block. Understanding DATA LIST is important to understanding how to use PSPP to read your data files.

There are two major variants of DATA LIST, which are fixed format and free format. In addition, free format has a minor variant, list format, which is discussed in terms of its differences from vanilla free format.

Each form of DATA LIST is described in detail below.

See GET DATA for a command that offers a few enhancements over DATA LIST and that may be substituted for DATA LIST in many situations.

DATA LIST FIXED

DATA LIST [FIXED]
        {TABLE,NOTABLE}
        [FILE='FILE_NAME' [ENCODING='ENCODING']]
        [RECORDS=RECORD_COUNT]
        [END=END_VAR]
        [SKIP=RECORD_COUNT]
        /[line_no] VAR_SPEC...

where each VAR_SPEC takes one of the forms
        VAR_LIST START-END [TYPE_SPEC]
        VAR_LIST (FORTRAN_SPEC)

DATA LIST FIXED is used to read data files that have values at fixed positions on each line of single-line or multiline records. The keyword FIXED is optional.

The FILE subcommand must be used if input is to be taken from an external file. It may be used to specify a file name as a string or a file handle. If the FILE subcommand is not used, then input is assumed to be specified within the command file using BEGIN DATA...END DATA. The ENCODING subcommand may only be used if the FILE subcommand is also used. It specifies the character encoding of the file. See INSERT, for information on supported encodings.

The optional RECORDS subcommand, which takes a single integer as an argument, is used to specify the number of lines per record. If RECORDS is not specified, then the number of lines per record is calculated from the list of variable specifications later in DATA LIST.

The END subcommand is only useful in conjunction with INPUT PROGRAM.

The optional SKIP subcommand specifies a number of records to skip at the beginning of an input file. It can be used to skip over a row that contains variable names, for example.

DATA LIST can optionally output a table describing how the data file is read. The TABLE subcommand enables this output, and NOTABLE disables it. The default is to output the table.

The list of variables to be read from the data list must come last. Each line in the data record is introduced by a slash (/). Optionally, a line number may follow the slash. Following, any number of variable specifications may be present.

Each variable specification consists of a list of variable names followed by a description of their location on the input line. Sets of variables may be with TO, e.g. VAR1 TO VAR5. There are two ways to specify the location of the variable on the line: columnar style and FORTRAN style.

In columnar style, the starting column and ending column for the field are specified after the variable name, separated by a dash (-). For instance, the third through fifth columns on a line would be specified 3-5. By default, variables are considered to be in F format. (Use SET FORMAT to change the default.)

In columnar style, to use a variable format other than the default, specify the format type in parentheses after the column numbers. For instance, for alphanumeric A format, use (A).

In addition, implied decimal places can be specified in parentheses after the column numbers. As an example, suppose that a data file has a field in which the characters 1234 should be interpreted as having the value 12.34. Then this field has two implied decimal places, and the corresponding specification would be (2). If a field that has implied decimal places contains a decimal point, then the implied decimal places are not applied.

Changing the variable format and adding implied decimal places can be done together; for instance, (N,5).

When using columnar style, the input and output width of each variable is computed from the field width. The field width must be evenly divisible into the number of variables specified.

FORTRAN style is an altogether different approach to specifying field locations. With this approach, a list of variable input format specifications, separated by commas, are placed after the variable names inside parentheses. Each format specifier advances as many characters into the input line as it uses.

Implied decimal places also exist in FORTRAN style. A format specification with D decimal places also has D implied decimal places.

In addition to the standard formats, FORTRAN style defines some extensions:

  • X
    Advance the current column on this line by one character position.

  • T<X>
    Set the current column on this line to column <X>, with column numbers considered to begin with 1 at the left margin.

  • NEWREC<X>
    Skip forward <X> lines in the current record, resetting the active column to the left margin.

  • Repeat count
    Any format specifier may be preceded by a number. This causes the action of that format specifier to be repeated the specified number of times.

  • (SPEC1, ..., SPECN)
    Use () to group specifiers together. This is most useful when preceded by a repeat count. Groups may be nested.

    FORTRAN and columnar styles may be freely intermixed. Columnar style leaves the active column immediately after the ending column specified. Record motion using NEWREC in FORTRAN style also applies to later FORTRAN and columnar specifiers.

Example 1

DATA LIST TABLE /NAME 1-10 (A) INFO1 TO INFO3 12-17 (1).

BEGIN DATA.
John Smith 102311
Bob Arnold 122015
Bill Yates  918 6
END DATA.

Defines the following variables:

  • NAME, a 10-character-wide string variable, in columns 1 through 10.

  • INFO1, a numeric variable, in columns 12 through 13.

  • INFO2, a numeric variable, in columns 14 through 15.

  • INFO3, a numeric variable, in columns 16 through 17.

The BEGIN DATA/END DATA commands cause three cases to be defined:

CaseNAMEINFO1INFO2INFO3
1John Smith102311
2Bob Arnold122015
3Bill Yates9186

The TABLE keyword causes PSPP to print out a table describing the four variables defined.

Example 2

DATA LIST FILE="survey.dat"
        /ID 1-5 NAME 7-36 (A) SURNAME 38-67 (A) MINITIAL 69 (A)
        /Q01 TO Q50 7-56
        /.

Defines the following variables:

  • ID, a numeric variable, in columns 1-5 of the first record.

  • NAME, a 30-character string variable, in columns 7-36 of the first record.

  • SURNAME, a 30-character string variable, in columns 38-67 of the first record.

  • MINITIAL, a 1-character string variable, in column 69 of the first record.

  • Fifty variables Q01, Q02, Q03, ..., Q49, Q50, all numeric, Q01 in column 7, Q02 in column 8, ..., Q49 in column 55, Q50 in column 56, all in the second record.

Cases are separated by a blank record.

Data is read from file survey.dat in the current directory.

DATA LIST FREE

DATA LIST FREE
        [({TAB,'C'}, ...)]
        [{NOTABLE,TABLE}]
        [FILE='FILE_NAME' [ENCODING='ENCODING']]
        [SKIP=N_RECORDS]
        /VAR_SPEC...

where each VAR_SPEC takes one of the forms
        VAR_LIST [(TYPE_SPEC)]
        VAR_LIST *

In free format, the input data is, by default, structured as a series of fields separated by spaces, tabs, or line breaks. If the current DECIMAL separator is DOT, then commas are also treated as field separators. Each field's content may be unquoted, or it may be quoted with a pairs of apostrophes (') or double quotes ("). Unquoted white space separates fields but is not part of any field. Any mix of spaces, tabs, and line breaks is equivalent to a single space for the purpose of separating fields, but consecutive commas will skip a field.

Alternatively, delimiters can be specified explicitly, as a parenthesized, comma-separated list of single-character strings immediately following FREE. The word TAB may also be used to specify a tab character as a delimiter. When delimiters are specified explicitly, only the given characters, plus line breaks, separate fields. Furthermore, leading spaces at the beginnings of fields are not trimmed, consecutive delimiters define empty fields, and no form of quoting is allowed.

The NOTABLE and TABLE subcommands are as in DATA LIST FIXED above. NOTABLE is the default.

The FILE, SKIP, and ENCODING subcommands are as in DATA LIST FIXED above.

The variables to be parsed are given as a single list of variable names. This list must be introduced by a single slash (/). The set of variable names may contain format specifications in parentheses. Format specifications apply to all variables back to the previous parenthesized format specification.

An asterisk on its own has the same effect as (F8.0), assigning the variables preceding it input/output format F8.0.

Specified field widths are ignored on input, although all normal limits on field width apply, but they are honored on output.

DATA LIST LIST

DATA LIST LIST
        [({TAB,'C'}, ...)]
        [{NOTABLE,TABLE}]
        [FILE='FILE_NAME' [ENCODING='ENCODING']]
        [SKIP=RECORD_COUNT]
        /VAR_SPEC...

where each VAR_SPEC takes one of the forms
        VAR_LIST [(TYPE_SPEC)]
        VAR_LIST *

With one exception, DATA LIST LIST is syntactically and semantically equivalent to DATA LIST FREE. The exception is that each input line is expected to correspond to exactly one input record. If more or fewer fields are found on an input line than expected, an appropriate diagnostic is issued.

END CASE

END CASE.

END CASE is used only within INPUT PROGRAM to output the current case.

END FILE

END FILE.

END FILE is used only within INPUT PROGRAM to terminate the current input program.

FILE HANDLE

Syntax Overview

For text files:

FILE HANDLE HANDLE_NAME
        /NAME='FILE_NAME
        [/MODE=CHARACTER]
        [/ENDS={CR,CRLF}]
        /TABWIDTH=TAB_WIDTH
        [ENCODING='ENCODING']

For binary files in native encoding with fixed-length records:

FILE HANDLE HANDLE_NAME
        /NAME='FILE_NAME'
        /MODE=IMAGE
        [/LRECL=REC_LEN]
        [ENCODING='ENCODING']

For binary files in native encoding with variable-length records:

FILE HANDLE HANDLE_NAME
        /NAME='FILE_NAME'
        /MODE=BINARY
        [/LRECL=REC_LEN]
        [ENCODING='ENCODING']

For binary files encoded in EBCDIC:

FILE HANDLE HANDLE_NAME
        /NAME='FILE_NAME'
        /MODE=360
        /RECFORM={FIXED,VARIABLE,SPANNED}
        [/LRECL=REC_LEN]
        [ENCODING='ENCODING']

Details

Use FILE HANDLE to associate a file handle name with a file and its attributes, so that later commands can refer to the file by its handle name. Names of text files can be specified directly on commands that access files, so that FILE HANDLE is only needed when a file is not an ordinary file containing lines of text. However, FILE HANDLE may be used even for text files, and it may be easier to specify a file's name once and later refer to it by an abstract handle.

Specify the file handle name as the identifier immediately following the FILE HANDLE command name. The identifier INLINE is reserved for representing data embedded in the syntax file (see BEGIN DATA). The file handle name must not already have been used in a previous invocation of FILE HANDLE, unless it has been closed with CLOSE FILE HANDLE.

The effect and syntax of FILE HANDLE depends on the selected MODE:

  • In CHARACTER mode, the default, the data file is read as a text file. Each text line is read as one record.

    In CHARACTER mode only, tabs are expanded to spaces by input programs, except by DATA LIST FREE with explicitly specified delimiters. Each tab is 4 characters wide by default, but TABWIDTH (a PSPP extension) may be used to specify an alternate width. Use a TABWIDTH of 0 to suppress tab expansion.

    A file written in CHARACTER mode by default uses the line ends of the system on which PSPP is running, that is, on Windows, the default is CR LF line ends, and on other systems the default is LF only. Specify ENDS as CR or CRLF to override the default. PSPP reads files using either convention on any kind of system, regardless of ENDS.

  • In IMAGE mode, the data file is treated as a series of fixed-length binary records. LRECL should be used to specify the record length in bytes, with a default of 1024. On input, it is an error if an IMAGE file's length is not a integer multiple of the record length. On output, each record is padded with spaces or truncated, if necessary, to make it exactly the correct length.

  • In BINARY mode, the data file is treated as a series of variable-length binary records. LRECL may be specified, but its value is ignored. The data for each record is both preceded and followed by a 32-bit signed integer in little-endian byte order that specifies the length of the record. (This redundancy permits records in these files to be efficiently read in reverse order, although PSPP always reads them in forward order.) The length does not include either integer.

  • Mode 360 reads and writes files in formats first used for tapes in the 1960s on IBM mainframe operating systems and still supported today by the modern successors of those operating systems. For more information, see OS/400 Tape and Diskette Device Programming, available on IBM's website.

    Alphanumeric data in mode 360 files are encoded in EBCDIC. PSPP translates EBCDIC to or from the host's native format as necessary on input or output, using an ASCII/EBCDIC translation that is one-to-one, so that a "round trip" from ASCII to EBCDIC back to ASCII, or vice versa, always yields exactly the original data.

    The RECFORM subcommand is required in mode 360. The precise file format depends on its setting:

    • F
      FIXED
      This record format is equivalent to IMAGE mode, except for EBCDIC translation.

      IBM documentation calls this *F (fixed-length, deblocked) format.

    • V
      VARIABLE
      The file comprises a sequence of zero or more variable-length blocks. Each block begins with a 4-byte "block descriptor word" (BDW). The first two bytes of the BDW are an unsigned integer in big-endian byte order that specifies the length of the block, including the BDW itself. The other two bytes of the BDW are ignored on input and written as zeros on output.

      Following the BDW, the remainder of each block is a sequence of one or more variable-length records, each of which in turn begins with a 4-byte "record descriptor word" (RDW) that has the same format as the BDW. Following the RDW, the remainder of each record is the record data.

      The maximum length of a record in VARIABLE mode is 65,527 bytes: 65,535 bytes (the maximum value of a 16-bit unsigned integer), minus 4 bytes for the BDW, minus 4 bytes for the RDW.

      In mode VARIABLE, LRECL specifies a maximum, not a fixed, record length, in bytes. The default is 8,192.

      IBM documentation calls this *VB (variable-length, blocked, unspanned) format.

    • VS
      SPANNED
      This format is like VARIABLE, except that logical records may be split among multiple physical records (called "segments") or blocks. In SPANNED mode, the third byte of each RDW is called the segment control character (SCC). Odd SCC values cause the segment to be appended to a record buffer maintained in memory; even values also append the segment and then flush its contents to the input procedure. Canonically, SCC value 0 designates a record not spanned among multiple segments, and values 1 through 3 designate the first segment, the last segment, or an intermediate segment, respectively, within a multi-segment record. The record buffer is also flushed at end of file regardless of the final record's SCC.

      The maximum length of a logical record in VARIABLE mode is limited only by memory available to PSPP. Segments are limited to 65,527 bytes, as in VARIABLE mode.

      This format is similar to what IBM documentation call *VS (variable-length, deblocked, spanned) format.

    In mode 360, fields of type A that extend beyond the end of a record read from disk are padded with spaces in the host's native character set, which are then translated from EBCDIC to the native character set. Thus, when the host's native character set is based on ASCII, these fields are effectively padded with character X'80'. This wart is implemented for compatibility.

The NAME subcommand specifies the name of the file associated with the handle. It is required in all modes but SCRATCH mode, in which its use is forbidden.

The ENCODING subcommand specifies the encoding of text in the file. For reading text files in CHARACTER mode, all of the forms described for ENCODING on the INSERT command are supported. For reading in other file-based modes, encoding autodetection is not supported; if the specified encoding requests autodetection then the default encoding is used. This is also true when a file handle is used for writing a file in any mode.

INPUT PROGRAM…END INPUT PROGRAM

INPUT PROGRAM.
... input commands ...
END INPUT PROGRAM.

INPUT PROGRAM...END INPUT PROGRAM specifies a complex input program. By placing data input commands within INPUT PROGRAM, PSPP programs can take advantage of more complex file structures than available with only DATA LIST.

The first sort of extended input program is to simply put multiple DATA LIST commands within the INPUT PROGRAM. This will cause all of the data files to be read in parallel. Input will stop when end of file is reached on any of the data files.

Transformations, such as conditional and looping constructs, can also be included within INPUT PROGRAM. These can be used to combine input from several data files in more complex ways. However, input will still stop when end of file is reached on any of the data files.

To prevent INPUT PROGRAM from terminating at the first end of file, use the END subcommand on DATA LIST. This subcommand takes a variable name, which should be a numeric scratch variable. (It need not variable](../language/datasets/scratch-variables.md). (It need not be a scratch variable but otherwise the results can be surprising.) The value of this variable is set to 0 when reading the data file, or 1 when end of file is encountered.

Two additional commands are useful in conjunction with INPUT PROGRAM. END CASE is the first. Normally each loop through the INPUT PROGRAM structure produces one case. END CASE controls exactly when cases are output. When END CASE is used, looping from the end of INPUT PROGRAM to the beginning does not cause a case to be output.

END FILE is the second. When the END subcommand is used on DATA LIST, there is no way for the INPUT PROGRAM construct to stop looping, so an infinite loop results. END FILE, when executed, stops the flow of input data and passes out of the INPUT PROGRAM structure.

INPUT PROGRAM must contain at least one DATA LIST or END FILE command.

Example 1: Read two files in parallel to the end of the shorter

The following example reads variable X from file a.txt and variable Y from file b.txt. If one file is shorter than the other then the extra data in the longer file is ignored.

INPUT PROGRAM.
    DATA LIST NOTABLE FILE='a.txt'/X 1-10.
    DATA LIST NOTABLE FILE='b.txt'/Y 1-10.
END INPUT PROGRAM.
LIST.

Example 2: Read two files in parallel, supplementing the shorter

The following example also reads variable X from a.txt and variable Y from b.txt. If one file is shorter than the other then it continues reading the longer to its end, setting the other variable to system-missing.

INPUT PROGRAM.
    NUMERIC #A #B.

    DO IF NOT #A.
        DATA LIST NOTABLE END=#A FILE='a.txt'/X 1-10.
    END IF.
    DO IF NOT #B.
        DATA LIST NOTABLE END=#B FILE='b.txt'/Y 1-10.
    END IF.
    DO IF #A AND #B.
        END FILE.
    END IF.
    END CASE.
END INPUT PROGRAM.
LIST.

Example 3: Concatenate two files (version 1)

The following example reads data from file a.txt, then from b.txt, and concatenates them into a single active dataset.

INPUT PROGRAM.
    NUMERIC #A #B.

    DO IF #A.
        DATA LIST NOTABLE END=#B FILE='b.txt'/X 1-10.
        DO IF #B.
            END FILE.
        ELSE.
            END CASE.
        END IF.
    ELSE.
        DATA LIST NOTABLE END=#A FILE='a.txt'/X 1-10.
        DO IF NOT #A.
            END CASE.
        END IF.
    END IF.
END INPUT PROGRAM.
LIST.

Example 4: Concatenate two files (version 2)

This is another way to do the same thing as Example 3.

INPUT PROGRAM.
    NUMERIC #EOF.

    LOOP IF NOT #EOF.
        DATA LIST NOTABLE END=#EOF FILE='a.txt'/X 1-10.
        DO IF NOT #EOF.
            END CASE.
        END IF.
    END LOOP.

    COMPUTE #EOF = 0.
    LOOP IF NOT #EOF.
        DATA LIST NOTABLE END=#EOF FILE='b.txt'/X 1-10.
        DO IF NOT #EOF.
            END CASE.
        END IF.
    END LOOP.

    END FILE.
END INPUT PROGRAM.
LIST.

Example 5: Generate random variates

The follows example creates a dataset that consists of 50 random variates between 0 and 10.

INPUT PROGRAM.
    LOOP #I=1 TO 50.
        COMPUTE X=UNIFORM(10).
        END CASE.
    END LOOP.
    END FILE.
END INPUT PROGRAM.
LIST /FORMAT=NUMBERED.

LIST

LIST
        /VARIABLES=VAR_LIST
        /CASES=FROM START_INDEX TO END_INDEX BY INCR_INDEX
        /FORMAT={UNNUMBERED,NUMBERED} {WRAP,SINGLE}

The LIST procedure prints the values of specified variables to the listing file.

The VARIABLES subcommand specifies the variables whose values are to be printed. Keyword VARIABLES is optional. If the VARIABLES subcommand is omitted then all variables in the active dataset are printed.

The CASES subcommand can be used to specify a subset of cases to be printed. Specify FROM and the case number of the first case to print, TO and the case number of the last case to print, and BY and the number of cases to advance between printing cases, or any subset of those settings. If CASES is not specified then all cases are printed.

The FORMAT subcommand can be used to change the output format. NUMBERED will print case numbers along with each case; UNNUMBERED, the default, causes the case numbers to be omitted. The WRAP and SINGLE settings are currently not used.

Case numbers start from 1. They are counted after all transformations have been considered.

LIST is a procedure. It causes the data to be read.

NEW FILE

NEW FILE.

The NEW FILE command clears the dictionary and data from the current active dataset.

PRINT

PRINT
        [OUTFILE='FILE_NAME']
        [RECORDS=N_LINES]
        [{NOTABLE,TABLE}]
        [ENCODING='ENCODING']
        [/[LINE_NO] ARG...]

ARG takes one of the following forms:
        'STRING' [START]
        VAR_LIST START-END [TYPE_SPEC]
        VAR_LIST (FORTRAN_SPEC)
        VAR_LIST *

The PRINT transformation writes variable data to the listing file or an output file. PRINT is executed when a procedure causes the data to be read. Follow PRINT by EXECUTE to print variable data without invoking a procedure.

All PRINT subcommands are optional. If no strings or variables are specified, PRINT outputs a single blank line.

The OUTFILE subcommand specifies the file to receive the output. The file may be a file name as a string or a file handle. If OUTFILE is not present then output is sent to PSPP's output listing file. When OUTFILE is present, the output is written to the file in a plain text format, with a space inserted at beginning of each output line, even lines that otherwise would be blank.

The ENCODING subcommand may only be used if the OUTFILE subcommand is also used. It specifies the character encoding of the file. See INSERT, for information on supported encodings.

The RECORDS subcommand specifies the number of lines to be output. The number of lines may optionally be surrounded by parentheses.

TABLE will cause the PRINT command to output a table to the listing file that describes what it will print to the output file. NOTABLE, the default, suppresses this output table.

Introduce the strings and variables to be printed with a slash (/). Optionally, the slash may be followed by a number indicating which output line is specified. In the absence of this line number, the next line number is specified. Multiple lines may be specified using multiple slashes with the intended output for a line following its respective slash.

Literal strings may be printed. Specify the string itself. Optionally the string may be followed by a column number, specifying the column on the line where the string should start. Otherwise, the string is printed at the current position on the line.

Variables to be printed can be specified in the same ways as available for DATA LIST FIXED. In addition, a variable list may be followed by an asterisk (*), which indicates that the variables should be printed in their dictionary print formats, separated by spaces. A variable list followed by a slash or the end of command is interpreted in the same way.

If a FORTRAN type specification is used to move backwards on the current line, then text is written at that point on the line, the line is truncated to that length, although additional text being added will again extend the line to that length.

PRINT EJECT

PRINT EJECT
        OUTFILE='FILE_NAME'
        RECORDS=N_LINES
        {NOTABLE,TABLE}
        /[LINE_NO] ARG...

ARG takes one of the following forms:
        'STRING' [START-END]
        VAR_LIST START-END [TYPE_SPEC]
        VAR_LIST (FORTRAN_SPEC)
        VAR_LIST *

PRINT EJECT advances to the beginning of a new output page in the listing file or output file. It can also output data in the same way as PRINT.

All PRINT EJECT subcommands are optional.

Without OUTFILE, PRINT EJECT ejects the current page in the listing file, then it produces other output, if any is specified.

With OUTFILE, PRINT EJECT writes its output to the specified file. The first line of output is written with 1 inserted in the first column. Commonly, this is the only line of output. If additional lines of output are specified, these additional lines are written with a space inserted in the first column, as with PRINT.

See PRINT for more information on syntax and usage.

PRINT SPACE

PRINT SPACE [OUTFILE='file_name'] [ENCODING='ENCODING'] [n_lines].

PRINT SPACE prints one or more blank lines to an output file.

The OUTFILE subcommand is optional. It may be used to direct output to a file specified by file name as a string or file handle. If OUTFILE is not handle](../language/files/file-handles.md). If OUTFILE is not specified then output is directed to the listing file.

The ENCODING subcommand may only be used if OUTFILE is also used. It specifies the character encoding of the file. See INSERT, for information on supported encodings.

n_lines is also optional. If present, it is an expression for the number of blank lines to be printed. The expression must evaluate to a nonnegative value.

REREAD

REREAD [FILE=handle] [COLUMN=column] [ENCODING='ENCODING'].

The REREAD transformation allows the previous input line in a data file already processed by DATA LIST or another input command to be re-read for further processing.

The FILE subcommand, which is optional, is used to specify the file to have its line re-read. The file must be specified as the name of a file handle. If FILE is not file handle. If FILE is not specified then the file specified on the most recent DATA LIST command is assumed.

By default, the line re-read is re-read in its entirety. With the COLUMN subcommand, a prefix of the line can be exempted from re-reading. Specify an expression evaluating to the expression evaluating to the first column that should be included in the re-read line. Columns are numbered from 1 at the left margin.

The ENCODING subcommand may only be used if the FILE subcommand is also used. It specifies the character encoding of the file. See INSERT for information on supported encodings.

Issuing REREAD multiple times will not back up in the data file. Instead, it will re-read the same line multiple times.

REPEATING DATA

REPEATING DATA
        /STARTS=START-END
        /OCCURS=N_OCCURS
        /FILE='FILE_NAME'
        /LENGTH=LENGTH
        /CONTINUED[=CONT_START-CONT_END]
        /ID=ID_START-ID_END=ID_VAR
        /{TABLE,NOTABLE}
        /DATA=VAR_SPEC...

where each VAR_SPEC takes one of the forms
        VAR_LIST START-END [TYPE_SPEC]
        VAR_LIST (FORTRAN_SPEC)

REPEATING DATA parses groups of data repeating in a uniform format, possibly with several groups on a single line. Each group of data corresponds with one case. REPEATING DATA may only be used within INPUT PROGRAM. When used with DATA LIST, it can be used to parse groups of cases that share a subset of variables but differ in their other data.

The STARTS subcommand is required. Specify a range of columns, using literal numbers or numeric variable names. This range specifies the columns on the first line that are used to contain groups of data. The ending column is optional. If it is not specified, then the record width of the input file is used. For the inline file, this is 80 columns; for a file with fixed record widths it is the record width; for other files it is 1024 characters by default.

The OCCURS subcommand is required. It must be a number or the name of a numeric variable. Its value is the number of groups present in the current record.

The DATA subcommand is required. It must be the last subcommand specified. It is used to specify the data present within each repeating group. Column numbers are specified relative to the beginning of a group at column 1. Data is specified in the same way as with DATA LIST FIXED.

All other subcommands are optional.

FILE specifies the file to read, either a file name as a string or a file handle. If FILE is not file handle. If FILE is not present then the default is the last file handle used on the most recent DATA LIST command.

By default REPEATING DATA will output a table describing how it will parse the input data. Specifying NOTABLE will disable this behavior; specifying TABLE will explicitly enable it.

The LENGTH subcommand specifies the length in characters of each group. If it is not present then length is inferred from the DATA subcommand. LENGTH may be a number or a variable name.

Normally all the data groups are expected to be present on a single line. Use the CONTINUED command to indicate that data can be continued onto additional lines. If data on continuation lines starts at the left margin and continues through the entire field width, no column specifications are necessary on CONTINUED. Otherwise, specify the possible range of columns in the same way as on STARTS.

When data groups are continued from line to line, it is easy for cases to get out of sync through careless hand editing. The ID subcommand allows a case identifier to be present on each line of repeating data groups. REPEATING DATA will check for the same identifier on each line and report mismatches. Specify the range of columns that the identifier will occupy, followed by an equals sign (=) and the identifier variable name. The variable must already have been declared with NUMERIC or another command.

REPEATING DATA should be the last command given within an INPUT PROGRAM. It should not be enclosed within LOOP…END LOOP. Use DATA LIST before, not after, REPEATING DATA.

WRITE

WRITE
        OUTFILE='FILE_NAME'
        RECORDS=N_LINES
        {NOTABLE,TABLE}
        /[LINE_NO] ARG...

ARG takes one of the following forms:
        'STRING' [START-END]
        VAR_LIST START-END [TYPE_SPEC]
        VAR_LIST (FORTRAN_SPEC)
        VAR_LIST *

WRITE writes text or binary data to an output file. WRITE differs from PRINT in only a few ways:

  • WRITE uses write formats by default, whereas PRINT uses print formats.

  • PRINT inserts a space between variables unless a format is explicitly specified, but WRITE never inserts space between variables in output.

  • PRINT inserts a space at the beginning of each line that it writes to an output file (and PRINT EJECT inserts 1 at the beginning of each line that should begin a new page), but WRITE does not.

  • PRINT outputs the system-missing value according to its specified output format, whereas WRITE outputs the system-missing value as a field filled with spaces. Binary formats are an exception.

Working with SPSS Data Files

These commands read and write data files in SPSS and other proprietary or specialized data formats.

APPLY DICTIONARY

APPLY DICTIONARY FROM={'FILE_NAME',FILE_HANDLE}.

APPLY DICTIONARY applies the variable labels, value labels, and missing values taken from a file to corresponding variables in the active dataset. In some cases it also updates the weighting variable.

The FROM clause is mandatory. Use it to specify a system file or portable file's name in single quotes, or a file handle name. The dictionary in the file is read, but it does not replace the active dataset's dictionary. The file's data is not read.

Only variables with names that exist in both the active dataset and the system file are considered. Variables with the same name but different types (numeric, string) cause an error message. Otherwise, the system file variables' attributes replace those in their matching active dataset variables:

  • If a system file variable has a variable label, then it replaces the variable label of the active dataset variable. If the system file variable does not have a variable label, then the active dataset variable's variable label, if any, is retained.

  • If the system file variable has variable attributes, then those attributes replace the active dataset variable's variable attributes. If the system file variable does not have varaible attributes, then the active dataset variable's variable attributes, if any, is retained.

  • If the active dataset variable is numeric or short string, then value labels and missing values, if any, are copied to the active dataset variable. If the system file variable does not have value labels or missing values, then those in the active dataset variable, if any, are not disturbed.

In addition to properties of variables, some properties of the active file dictionary as a whole are updated:

  • If the system file has custom attributes (see DATAFILE ATTRIBUTE), then those attributes replace the active dataset variable's custom attributes.

  • If the active dataset has a weight variable, and the system file does not, or if the weighting variable in the system file does not exist in the active dataset, then the active dataset weighting variable, if any, is retained. Otherwise, the weighting variable in the system file becomes the active dataset weighting variable.

APPLY DICTIONARY takes effect immediately. It does not read the active dataset. The system file is not modified.

EXPORT

EXPORT
        /OUTFILE='FILE_NAME'
        /UNSELECTED={RETAIN,DELETE}
        /DIGITS=N
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /RENAME=(SRC_NAMES=TARGET_NAMES)...
        /TYPE={COMM,TAPE}
        /MAP

The EXPORT procedure writes the active dataset's dictionary and data to a specified portable file.

EXPORT is obsolete and retained only for compatibility. New syntax should use SAVE instead.

UNSELECTED controls whether cases excluded with FILTER are written to the file. These can be excluded by specifying DELETE on the UNSELECTED subcommand. The default is RETAIN.

Portable files express real numbers in base 30. Integers are always expressed to the maximum precision needed to make them exact. Non-integers are, by default, expressed to the machine's maximum natural precision (approximately 15 decimal digits on many machines). If many numbers require this many digits, the portable file may significantly increase in size. As an alternative, the DIGITS subcommand may be used to specify the number of decimal digits of precision to write. DIGITS applies only to non-integers.

The OUTFILE subcommand, which is the only required subcommand, specifies the portable file to be written as a file name string or a file handle.

DROP, KEEP, and RENAME have the same syntax and meaning as for the SAVE command.

The TYPE subcommand specifies the character set for use in the portable file. Its value is currently not used.

The MAP subcommand is currently ignored.

EXPORT is a procedure. It causes the active dataset to be read.

GET

GET
        /FILE={'FILE_NAME',FILE_HANDLE}
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /RENAME=(SRC_NAMES=TARGET_NAMES)...
        /ENCODING='ENCODING'

GET clears the current dictionary and active dataset and replaces them with the dictionary and data from a specified file.

The FILE subcommand is the only required subcommand. Specify the SPSS system file, SPSS/PC+ system file, or SPSS portable file to be read as a string file name or a file handle. handle](../language/files/file-handles.md).

By default, all the variables in a file are read. The DROP subcommand can be used to specify a list of variables that are not to be read. By contrast, the KEEP subcommand can be used to specify variable that are to be read, with all other variables not read.

Normally variables in a file retain the names that they were saved under. Use the RENAME subcommand to change these names. Specify, within parentheses, a list of variable names followed by an equals sign (=) and the names that they should be renamed to. Multiple parenthesized groups of variable names can be included on a single RENAME subcommand. Variables' names may be swapped using a RENAME subcommand of the form /RENAME=(A B=B A).

Alternate syntax for the RENAME subcommand allows the parentheses to be omitted. When this is done, only a single variable may be renamed at once. For instance, /RENAME=A=B. This alternate syntax is discouraged.

DROP, KEEP, and RENAME are executed in left-to-right order. Each may be present any number of times. GET never modifies a file on disk. Only the active dataset read from the file is affected by these subcommands.

PSPP automatically detects the encoding of string data in the file, when possible. The character encoding of old SPSS system files cannot always be guessed correctly, and SPSS/PC+ system files do not include any indication of their encoding. Specify the ENCODING subcommand with an IANA character set name as its string argument to override the default. Use SYSFILE INFO to analyze the encodings that might be valid for a system file. The ENCODING subcommand is a PSPP extension.

GET does not cause the data to be read, only the dictionary. The data is read later, when a procedure is executed.

Use of GET to read a portable file is a PSPP extension.

GET DATA

GET DATA
        /TYPE={GNM,ODS,PSQL,TXT}
        ...additional subcommands depending on TYPE...

The GET DATA command is used to read files and other data sources created by other applications. When this command is executed, the current dictionary and active dataset are replaced with variables and data read from the specified source.

The TYPE subcommand is mandatory and must be the first subcommand specified. It determines the type of the file or source to read. PSPP currently supports the following TYPEs:

Each supported file type has additional subcommands, explained in separate sections below.

Spreadsheet Files

GET DATA /TYPE={GNM, ODS}
         /FILE={'FILE_NAME'}
         /SHEET={NAME 'SHEET_NAME', INDEX N}
         /CELLRANGE={RANGE 'RANGE', FULL}
         /READNAMES={ON, OFF}
         /ASSUMEDSTRWIDTH=N.

GET DATA can read Gnumeric spreadsheets (http://gnumeric.org), and spreadsheets in OpenDocument format (http://libreplanet.org/wiki/Group:OpenDocument/Software). Use the TYPE subcommand to indicate the file's format. /TYPE=GNM indicates Gnumeric files, /TYPE=ODS indicates OpenDocument. The FILE subcommand is mandatory. Use it to specify the name file to be read. All other subcommands are optional.

The format of each variable is determined by the format of the spreadsheet cell containing the first datum for the variable. If this cell is of string (text) format, then the width of the variable is determined from the length of the string it contains, unless the ASSUMEDSTRWIDTH subcommand is given.

The SHEET subcommand specifies the sheet within the spreadsheet file to read. There are two forms of the SHEET subcommand. In the first form, /SHEET=name SHEET_NAME, the string SHEET_NAME is the name of the sheet to read. In the second form, /SHEET=index IDX, IDX is a integer which is the index of the sheet to read. The first sheet has the index 1. If the SHEET subcommand is omitted, then the command reads the first sheet in the file.

The CELLRANGE subcommand specifies the range of cells within the sheet to read. If the subcommand is given as /CELLRANGE=FULL, then the entire sheet is read. To read only part of a sheet, use the form /CELLRANGE=range 'TOP_LEFT_CELL:BOTTOM_RIGHT_CELL'. For example, the subcommand /CELLRANGE=range 'C3:P19' reads columns C-P and rows 3-19, inclusive. Without the CELLRANGE subcommand, the entire sheet is read.

If /READNAMES=ON is specified, then the contents of cells of the first row are used as the names of the variables in which to store the data from subsequent rows. This is the default. If /READNAMES=OFF is used, then the variables receive automatically assigned names.

The ASSUMEDSTRWIDTH subcommand specifies the maximum width of string variables read from the file. If omitted, the default value is determined from the length of the string in the first spreadsheet cell for each variable.

Postgres Database Queries

GET DATA /TYPE=PSQL
         /CONNECT={CONNECTION INFO}
         /SQL={QUERY}
         [/ASSUMEDSTRWIDTH=W]
         [/UNENCRYPTED]
         [/BSIZE=N].

GET DATA /TYPE=PSQL imports data from a local or remote Postgres database server. It automatically creates variables based on the table column names or the names specified in the SQL query. PSPP cannot support the full precision of some Postgres data types, so data of those types will lose some precision when PSPP imports them. PSPP does not support all Postgres data types. If PSPP cannot support a datum, GET DATA issues a warning and substitutes the system-missing value.

The CONNECT subcommand must be a string for the parameters of the database server from which the data should be fetched. The format of the string is given in the Postgres manual.

The SQL subcommand must be a valid SQL statement to retrieve data from the database.

The ASSUMEDSTRWIDTH subcommand specifies the maximum width of string variables read from the database. If omitted, the default value is determined from the length of the string in the first value read for each variable.

The UNENCRYPTED subcommand allows data to be retrieved over an insecure connection. If the connection is not encrypted, and the UNENCRYPTED subcommand is not given, then an error occurs. Whether or not the connection is encrypted depends upon the underlying psql library and the capabilities of the database server.

The BSIZE subcommand serves only to optimise the speed of data transfer. It specifies an upper limit on number of cases to fetch from the database at once. The default value is 4096. If your SQL statement fetches a large number of cases but only a small number of variables, then the data transfer may be faster if you increase this value. Conversely, if the number of variables is large, or if the machine on which PSPP is running has only a small amount of memory, then a smaller value is probably better.

Example

GET DATA /TYPE=PSQL
     /CONNECT='host=example.com port=5432 dbname=product user=fred passwd=xxxx'
     /SQL='select * from manufacturer'.

Textual Data Files

GET DATA /TYPE=TXT
        /FILE={'FILE_NAME',FILE_HANDLE}
        [ENCODING='ENCODING']
        [/ARRANGEMENT={DELIMITED,FIXED}]
        [/FIRSTCASE={FIRST_CASE}]
        [/IMPORTCASES=...]
        ...additional subcommands depending on ARRANGEMENT...

When TYPE=TXT is specified, GET DATA reads data in a delimited or fixed columnar format, much like DATA LIST.

The FILE subcommand must specify the file to be read as a string file name or (for textual data only) a file handle).

The ENCODING subcommand specifies the character encoding of the file to be read. See INSERT, for information on supported encodings.

The ARRANGEMENT subcommand determines the file's basic format. DELIMITED, the default setting, specifies that fields in the input data are separated by spaces, tabs, or other user-specified delimiters. FIXED specifies that fields in the input data appear at particular fixed column positions within records of a case.

By default, cases are read from the input file starting from the first line. To skip lines at the beginning of an input file, set FIRSTCASE to the number of the first line to read: 2 to skip the first line, 3 to skip the first two lines, and so on.

IMPORTCASES is ignored, for compatibility. Use N OF CASES to limit the number of cases read from a file, or SAMPLE to obtain a random sample of cases.

The remaining subcommands apply only to one of the two file arrangements, described below.

Delimited Data

GET DATA /TYPE=TXT
        /FILE={'FILE_NAME',FILE_HANDLE}
        [/ARRANGEMENT={DELIMITED,FIXED}]
        [/FIRSTCASE={FIRST_CASE}]
        [/IMPORTCASE={ALL,FIRST MAX_CASES,PERCENT PERCENT}]

        /DELIMITERS="DELIMITERS"
        [/QUALIFIER="QUOTES"
        [/DELCASE={LINE,VARIABLES N_VARIABLES}]
        /VARIABLES=DEL_VAR1 [DEL_VAR2]...
where each DEL_VAR takes the form:
        variable format

The GET DATA command with TYPE=TXT and ARRANGEMENT=DELIMITED reads input data from text files in delimited format, where fields are separated by a set of user-specified delimiters. Its capabilities are similar to those of DATA LIST FREE, with a few enhancements.

The required FILE subcommand and optional FIRSTCASE and IMPORTCASE subcommands are described above.

DELIMITERS, which is required, specifies the set of characters that may separate fields. Each character in the string specified on DELIMITERS separates one field from the next. The end of a line also separates fields, regardless of DELIMITERS. Two consecutive delimiters in the input yield an empty field, as does a delimiter at the end of a line. A space character as a delimiter is an exception: consecutive spaces do not yield an empty field and neither does any number of spaces at the end of a line.

To use a tab as a delimiter, specify \t at the beginning of the DELIMITERS string. To use a backslash as a delimiter, specify \\ as the first delimiter or, if a tab should also be a delimiter, immediately following \t. To read a data file in which each field appears on a separate line, specify the empty string for DELIMITERS.

The optional QUALIFIER subcommand names one or more characters that can be used to quote values within fields in the input. A field that begins with one of the specified quote characters ends at the next matching quote. Intervening delimiters become part of the field, instead of terminating it. The ability to specify more than one quote character is a PSPP extension.

The character specified on QUALIFIER can be embedded within a field that it quotes by doubling the qualifier. For example, if ' is specified on QUALIFIER, then 'a''b' specifies a field that contains a'b.

The DELCASE subcommand controls how data may be broken across lines in the data file. With LINE, the default setting, each line must contain all the data for exactly one case. For additional flexibility, to allow a single case to be split among lines or multiple cases to be contained on a single line, specify VARIABLES n_variables, where n_variables is the number of variables per case.

The VARIABLES subcommand is required and must be the last subcommand. Specify the name of each variable and its input format, in the order they should be read from the input file.

Example 1

On a Unix-like system, the /etc/passwd file has a format similar to this:

root:$1$nyeSP5gD$pDq/:0:0:,,,:/root:/bin/bash
blp:$1$BrP/pFg4$g7OG:1000:1000:Ben Pfaff,,,:/home/blp:/bin/bash
john:$1$JBuq/Fioq$g4A:1001:1001:John Darrington,,,:/home/john:/bin/bash
jhs:$1$D3li4hPL$88X1:1002:1002:Jason Stover,,,:/home/jhs:/bin/csh

The following syntax reads a file in the format used by /etc/passwd:

GET DATA /TYPE=TXT /FILE='/etc/passwd' /DELIMITERS=':'
        /VARIABLES=username A20
                   password A40
                   uid F10
                   gid F10
                   gecos A40
                   home A40
                   shell A40.

Example 2

Consider the following data on used cars:

model   year    mileage price   type    age
Civic   2002    29883   15900   Si      2
Civic   2003    13415   15900   EX      1
Civic   1992    107000  3800    n/a     12
Accord  2002    26613   17900   EX      1

The following syntax can be used to read the used car data:

GET DATA /TYPE=TXT /FILE='cars.data' /DELIMITERS=' ' /FIRSTCASE=2
        /VARIABLES=model A8
                   year F4
                   mileage F6
                   price F5
                   type A4
                   age F2.

Example 3

Consider the following information on animals in a pet store:

'Pet''s Name', "Age", "Color", "Date Received", "Price", "Height", "Type"
, (Years), , , (Dollars), ,
"Rover", 4.5, Brown, "12 Feb 2004", 80, '1''4"', "Dog"
"Charlie", , Gold, "5 Apr 2007", 12.3, "3""", "Fish"
"Molly", 2, Black, "12 Dec 2006", 25, '5"', "Cat"
"Gilly", , White, "10 Apr 2007", 10, "3""", "Guinea Pig"

The following syntax can be used to read the pet store data:

GET DATA /TYPE=TXT /FILE='pets.data' /DELIMITERS=', ' /QUALIFIER='''"' /ESCAPE
        /FIRSTCASE=3
        /VARIABLES=name A10
                   age F3.1
                   color A5
                   received EDATE10
                   price F5.2
                   height a5
                   type a10.

Fixed Columnar Data

GET DATA /TYPE=TXT
        /FILE={'file_name',FILE_HANDLE}
        [/ARRANGEMENT={DELIMITED,FIXED}]
        [/FIRSTCASE={FIRST_CASE}]
        [/IMPORTCASE={ALL,FIRST MAX_CASES,PERCENT PERCENT}]

        [/FIXCASE=N]
        /VARIABLES FIXED_VAR [FIXED_VAR]...
            [/rec# FIXED_VAR [FIXED_VAR]...]...
where each FIXED_VAR takes the form:
        VARIABLE START-END FORMAT

The GET DATA command with TYPE=TXT and ARRANGEMENT=FIXED reads input data from text files in fixed format, where each field is located in particular fixed column positions within records of a case. Its capabilities are similar to those of DATA LIST FIXED, with a few enhancements.

The required FILE subcommand and optional FIRSTCASE and IMPORTCASE subcommands are described above.

The optional FIXCASE subcommand may be used to specify the positive integer number of input lines that make up each case. The default value is 1.

The VARIABLES subcommand, which is required, specifies the positions at which each variable can be found. For each variable, specify its name, followed by its start and end column separated by - (e.g. 0-9), followed by an input format type (e.g. F) or a full format specification (e.g. DOLLAR12.2). For this command, columns are numbered starting from 0 at the left column. Introduce the variables in the second and later lines of a case by a slash followed by the number of the line within the case, e.g. /2 for the second line.

Example

Consider the following data on used cars:

model   year    mileage price   type    age
Civic   2002    29883   15900   Si      2
Civic   2003    13415   15900   EX      1
Civic   1992    107000  3800    n/a     12
Accord  2002    26613   17900   EX      1

The following syntax can be used to read the used car data:

GET DATA /TYPE=TXT /FILE='cars.data' /ARRANGEMENT=FIXED /FIRSTCASE=2
        /VARIABLES=model 0-7 A
                   year 8-15 F
                   mileage 16-23 F
                   price 24-31 F
                   type 32-40 A
                   age 40-47 F.

IMPORT

IMPORT
        /FILE='FILE_NAME'
        /TYPE={COMM,TAPE}
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /RENAME=(SRC_NAMES=TARGET_NAMES)...

The IMPORT transformation clears the active dataset dictionary and data and replaces them with a dictionary and data from a system file or portable file.

IMPORT is obsolete and retained only for compatibility with existing portable files. New syntax should use SAVE to write system files instead, and GET to read them.

The FILE subcommand, which is the only required subcommand, specifies the portable file to be read as a file name string or a file handle.

The TYPE subcommand is currently not used.

DROP, KEEP, and RENAME follow the syntax used by GET.

IMPORT does not cause the data to be read; only the dictionary. The data is read later, when a procedure is executed.

Use of IMPORT to read a system file is a PSPP extension.

SAVE

SAVE
        /OUTFILE={'FILE_NAME',FILE_HANDLE}
        /UNSELECTED={RETAIN,DELETE}
        /{UNCOMPRESSED,COMPRESSED,ZCOMPRESSED}
        /PERMISSIONS={WRITEABLE,READONLY}
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /VERSION=VERSION
        /RENAME=(SRC_NAMES=TARGET_NAMES)...
        /NAMES
        /MAP

The SAVE procedure causes the dictionary and data in the active dataset to be written to a system file.

OUTFILE is the only required subcommand. Specify the system file to be written as a string file name or a file handle. handle](../language/files/file-handles.md).

By default, cases excluded with FILTER are written to the system file. These can be excluded by specifying DELETE on the UNSELECTED subcommand. Specifying RETAIN makes the default explicit.

The UNCOMPRESSED, COMPRESSED, and ZCOMPRESSED subcommand determine the system file's compression level:

  • UNCOMPRESSED
    Data is not compressed. Each numeric value uses 8 bytes of disk space. Each string value uses one byte per column width, rounded up to a multiple of 8 bytes.

  • COMPRESSED
    Data is compressed in a simple way. Each integer numeric value between −99 and 151, inclusive, or system missing value uses one byte of disk space. Each 8-byte segment of a string that consists only of spaces uses 1 byte. Any other numeric value or 8-byte string segment uses 9 bytes of disk space.

  • ZCOMPRESSED
    Data is compressed with the "deflate" compression algorithm specified in RFC 1951 (the same algorithm used by gzip). Files written with this compression level cannot be read by PSPP 0.8.1 or earlier or by SPSS 20 or earlier.

COMPRESSED is the default compression level. The SET command can change this default.

The PERMISSIONS subcommand specifies operating system permissions for the new system file. WRITEABLE, the default, creates the file with read and write permission. READONLY creates the file for read-only access.

By default, all the variables in the active dataset dictionary are written to the system file. The DROP subcommand can be used to specify a list of variables not to be written. In contrast, KEEP specifies variables to be written, with all variables not specified not written.

Normally variables are saved to a system file under the same names they have in the active dataset. Use the RENAME subcommand to change these names. Specify, within parentheses, a list of variable names followed by an equals sign (=) and the names that they should be renamed to. Multiple parenthesized groups of variable names can be included on a single RENAME subcommand. Variables' names may be swapped using a RENAME subcommand of the form /RENAME=(A B=B A).

Alternate syntax for the RENAME subcommand allows the parentheses to be eliminated. When this is done, only a single variable may be renamed at once. For instance, /RENAME=A=B. This alternate syntax is discouraged.

DROP, KEEP, and RENAME are performed in left-to-right order. They each may be present any number of times. SAVE never modifies the active dataset. DROP, KEEP, and RENAME only affect the system file written to disk.

The VERSION subcommand specifies the version of the file format. Valid versions are 2 and 3. The default version is 3. In version 2 system files, variable names longer than 8 bytes are truncated. The two versions are otherwise identical.

The NAMES and MAP subcommands are currently ignored.

SAVE causes the data to be read. It is a procedure.

SAVE DATA COLLECTION

SAVE DATA COLLECTION
        /OUTFILE={'FILE_NAME',FILE_HANDLE}
        /METADATA={'FILE_NAME',FILE_HANDLE}
        /{UNCOMPRESSED,COMPRESSED,ZCOMPRESSED}
        /PERMISSIONS={WRITEABLE,READONLY}
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /VERSION=VERSION
        /RENAME=(SRC_NAMES=TARGET_NAMES)...
        /NAMES
        /MAP

Like SAVE, SAVE DATA COLLECTION writes the dictionary and data in the active dataset to a system file. In addition, it writes metadata to an additional XML metadata file.

OUTFILE is required. Specify the system file to be written as a string file name or a file handle. handle](../language/files/file-handles.md).

METADATA is also required. Specify the metadata file to be written as a string file name or a file handle. Metadata files customarily use a .mdd extension.

The current implementation of this command is experimental. It only outputs an approximation of the metadata file format. Please report bugs.

Other subcommands are optional. They have the same meanings as in the SAVE command.

SAVE DATA COLLECTION causes the data to be read. It is a procedure.

SAVE TRANSLATE

SAVE TRANSLATE
        /OUTFILE={'FILE_NAME',FILE_HANDLE}
        /TYPE={CSV,TAB}
        [/REPLACE]
        [/MISSING={IGNORE,RECODE}]

        [/DROP=VAR_LIST]
        [/KEEP=VAR_LIST]
        [/RENAME=(SRC_NAMES=TARGET_NAMES)...]
        [/UNSELECTED={RETAIN,DELETE}]
        [/MAP]

        ...additional subcommands depending on TYPE...

The SAVE TRANSLATE command is used to save data into various formats understood by other applications.

The OUTFILE and TYPE subcommands are mandatory. OUTFILE specifies the file to be written, as a string file name or a file handle. TYPE determines the handle](../language/files/file-handles.md). TYPE determines the type of the file or source to read. It must be one of the following:

  • CSV
    Comma-separated value format,

  • TAB
    Tab-delimited format.

By default, SAVE TRANSLATE does not overwrite an existing file. Use REPLACE to force an existing file to be overwritten.

With MISSING=IGNORE, the default, SAVE TRANSLATE treats user-missing values as if they were not missing. Specify MISSING=RECODE to output numeric user-missing values like system-missing values and string user-missing values as all spaces.

By default, all the variables in the active dataset dictionary are saved to the system file, but DROP or KEEP can select a subset of variable to save. The RENAME subcommand can also be used to change the names under which variables are saved; because they are used only in the output, these names do not have to conform to the usual PSPP variable naming rules. UNSELECTED determines whether cases filtered out by the FILTER command are written to the output file. These subcommands have the same syntax and meaning as on the SAVE command.

Each supported file type has additional subcommands, explained in separate sections below.

SAVE TRANSLATE causes the data to be read. It is a procedure.

Comma- and Tab-Separated Data Files

SAVE TRANSLATE
        /OUTFILE={'FILE_NAME',FILE_HANDLE}
        /TYPE=CSV
        [/REPLACE]
        [/MISSING={IGNORE,RECODE}]

        [/DROP=VAR_LIST]
        [/KEEP=VAR_LIST]
        [/RENAME=(SRC_NAMES=TARGET_NAMES)...]
        [/UNSELECTED={RETAIN,DELETE}]

        [/FIELDNAMES]
        [/CELLS={VALUES,LABELS}]
        [/TEXTOPTIONS DELIMITER='DELIMITER']
        [/TEXTOPTIONS QUALIFIER='QUALIFIER']
        [/TEXTOPTIONS DECIMAL={DOT,COMMA}]
        [/TEXTOPTIONS FORMAT={PLAIN,VARIABLE}]

The SAVE TRANSLATE command with TYPE=CSV or TYPE=TAB writes data in a comma- or tab-separated value format similar to that described by RFC 4180. Each variable becomes one output column, and each case becomes one line of output. If FIELDNAMES is specified, an additional line at the top of the output file lists variable names.

The CELLS and TEXTOPTIONS FORMAT settings determine how values are written to the output file:

  • CELLS=VALUES FORMAT=PLAIN (the default settings)
    Writes variables to the output in "plain" formats that ignore the details of variable formats. Numeric values are written as plain decimal numbers with enough digits to indicate their exact values in machine representation. Numeric values include e followed by an exponent if the exponent value would be less than -4 or greater than 16. Dates are written in MM/DD/YYYY format and times in HH:MM:SS format. WKDAY and MONTH values are written as decimal numbers.

    Numeric values use, by default, the decimal point character set with SET DECIMAL. Use DECIMAL=DOT or DECIMAL=COMMA to force a particular decimal point character.

  • CELLS=VALUES FORMAT=VARIABLE
    Writes variables using their print formats. Leading and trailing spaces are removed from numeric values, and trailing spaces are removed from string values.

  • CELLS=LABEL FORMAT=PLAIN
    CELLS=LABEL FORMAT=VARIABLE
    Writes value labels where they exist, and otherwise writes the values themselves as described above.

    Regardless of CELLS and TEXTOPTIONS FORMAT, numeric system-missing values are output as a single space.

    For TYPE=TAB, tab characters delimit values. For TYPE=CSV, the TEXTOPTIONS DELIMITER and DECIMAL settings determine the character that separate values within a line. If DELIMITER is specified, then the specified string separate values. If DELIMITER is not specified, then the default is a comma with DECIMAL=DOT or a semicolon with DECIMAL=COMMA. If DECIMAL is not given either, it is inferred from the decimal point character set with SET DECIMAL.

    The TEXTOPTIONS QUALIFIER setting specifies a character that is output before and after a value that contains the delimiter character or the qualifier character. The default is a double quote ("). A qualifier character that appears within a value is doubled.

SYSFILE INFO

SYSFILE INFO FILE='FILE_NAME' [ENCODING='ENCODING'].

SYSFILE INFO reads the dictionary in an SPSS system file, SPSS/PC+ system file, or SPSS portable file, and displays the information in its dictionary.

Specify a file name or file handle. SYSFILE INFO reads that file and displays information on its dictionary.

PSPP automatically detects the encoding of string data in the file, when possible. The character encoding of old SPSS system files cannot always be guessed correctly, and SPSS/PC+ system files do not include any indication of their encoding. Specify the ENCODING subcommand with an IANA character set name as its string argument to override the default, or specify ENCODING='DETECT' to analyze and report possibly valid encodings for the system file. The ENCODING subcommand is a PSPP extension.

SYSFILE INFO does not affect the current active dataset.

XEXPORT

XEXPORT
        /OUTFILE='FILE_NAME'
        /DIGITS=N
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /RENAME=(SRC_NAMES=TARGET_NAMES)...
        /TYPE={COMM,TAPE}
        /MAP

The XEXPORT transformation writes the active dataset dictionary and data to a specified portable file.

This transformation is a PSPP extension.

It is similar to the EXPORT procedure, with two differences:

  • XEXPORT is a transformation, not a procedure. It is executed when the data is read by a procedure or procedure-like command.

  • XEXPORT does not support the UNSELECTED subcommand.

See EXPORT for more information.

XSAVE

XSAVE
        /OUTFILE='FILE_NAME'
        /{UNCOMPRESSED,COMPRESSED,ZCOMPRESSED}
        /PERMISSIONS={WRITEABLE,READONLY}
        /DROP=VAR_LIST
        /KEEP=VAR_LIST
        /VERSION=VERSION
        /RENAME=(SRC_NAMES=TARGET_NAMES)...
        /NAMES
        /MAP

The XSAVE transformation writes the active dataset's dictionary and data to a system file. It is similar to the SAVE procedure, with two differences:

  • XSAVE is a transformation, not a procedure. It is executed when the data is read by a procedure or procedure-like command.

  • XSAVE does not support the UNSELECTED subcommand.

See SAVE for more information.

Combining Data Files

This chapter describes commands that allow data from system files, portable files, and open datasets to be combined to form a new active dataset. These commands can combine data files in the following ways:

  • ADD FILES interleaves or appends the cases from each input file. It is used with input files that have variables in common, but distinct sets of cases.

  • MATCH FILES adds the data together in cases that match across multiple input files. It is used with input files that have cases in common, but different information about each case.

  • UPDATE updates a master data file from data in a set of transaction files. Each case in a transaction data file modifies a matching case in the primary data file, or it adds a new case if no matching case can be found.

These commands share the majority of their syntax, described below. Each command's documentation explains its additional syntax.

Common Syntax

Per input file:
        /FILE={*,'FILE_NAME'}
        [/RENAME=(SRC_NAMES=TARGET_NAMES)...]
        [/IN=VAR_NAME]
        [/SORT]

Once per command:
        /BY VAR_LIST[({D|A})] [VAR_LIST[({D|A}]]...
        [/DROP=VAR_LIST]
        [/KEEP=VAR_LIST]
        [/FIRST=VAR_NAME]
        [/LAST=VAR_NAME]
        [/MAP]

Each of these commands reads two or more input files and combines them. The command's output becomes the new active dataset. None of the commands actually change the input files. Therefore, if you want the changes to become permanent, you must explicitly save them using an appropriate procedure or transformation.

The syntax of each command begins with a specification of the files to be read as input. For each input file, specify FILE with a system file or portable file's name as a string, a dataset or file handle name, or an asterisk (*) to use the active dataset as input. Use of portable files on FILE is a PSPP extension.

At least two FILE subcommands must be specified. If the active dataset is used as an input source, then TEMPORARY must not be in effect.

Each FILE subcommand may be followed by any number of RENAME subcommands that specify a parenthesized group or groups of variable names as they appear in the input file, followed by those variables' new names, separated by an equals sign (=), e.g. /RENAME=(OLD1=NEW1)(OLD2=NEW2). To rename a single variable, the parentheses may be omitted: /RENAME=OLD=NEW. Within a parenthesized group, variables are renamed simultaneously, so that /RENAME=(A B=B A) exchanges the names of variables A and B. Otherwise, renaming occurs in left-to-right order.

Each FILE subcommand may optionally be followed by a single IN subcommand, which creates a numeric variable with the specified name and format F1.0. The IN variable takes value 1 in an output case if the given input file contributed to that output case, and 0 otherwise. The DROP, KEEP, and RENAME subcommands have no effect on IN variables.

If BY is used (see below), the SORT keyword must be specified after a FILE if that input file is not already sorted on the BY variables. When SORT is specified, PSPP sorts the input file's data on the BY variables before it applies it to the command. When SORT is used, BY is required. SORT is a PSPP extension.

PSPP merges the dictionaries of all of the input files to form the dictionary of the new active dataset, like so:

  • The variables in the new active dataset are the union of all the input files' variables, matched based on their name. When a single input file contains a variable with a given name, the output file will contain exactly that variable. When more than one input file contains a variable with a given name, those variables must all have the same type (numeric or string) and, for string variables, the same width. Variables are matched after renaming with the RENAME subcommand. Thus, RENAME can be used to resolve conflicts.

  • The variable label for each output variable is taken from the first specified input file that has a variable label for that variable, and similarly for value labels and missing values.

  • The file label of the new active dataset is that of the first specified FILE that has a file label.

  • The documents in the new active dataset are the concatenation of all the input files' documents, in the order in which the FILE subcommands are specified.

  • If all of the input files are weighted on the same variable, then the new active dataset is weighted on that variable. Otherwise, the new active dataset is not weighted.

The remaining subcommands apply to the output file as a whole, rather than to individual input files. They must be specified at the end of the command specification, following all of the FILE and related subcommands. The most important of these subcommands is BY, which specifies a set of one or more variables that may be used to find corresponding cases in each of the input files. The variables specified on BY must be present in all of the input files. Furthermore, if any of the input files are not sorted on the BY variables, then SORT must be specified for those input files.

The variables listed on BY may include (A) or (D) annotations to specify ascending or descending sort order. See SORT CASES, for more details on this notation. Adding (A) or (D) to the BY subcommand specification is a PSPP extension.

The DROP subcommand can be used to specify a list of variables to exclude from the output. By contrast, the KEEP subcommand can be used to specify variables to include in the output; all variables not listed are dropped. DROP and KEEP are executed in left-to-right order and may be repeated any number of times. DROP and KEEP do not affect variables created by the IN, FIRST, and LAST subcommands, which are always included in the new active dataset, but they can be used to drop BY variables.

The FIRST and LAST subcommands are optional. They may only be specified on MATCH FILES and ADD FILES, and only when BY is used. FIRST and LIST each adds a numeric variable to the new active dataset, with the name given as the subcommand's argument and F1.0 print and write formats. The value of the FIRST variable is 1 in the first output case with a given set of values for the BY variables, and 0 in other cases. Similarly, the LAST variable is 1 in the last case with a given of BY values, and 0 in other cases.

When any of these commands creates an output case, variables that are only in files that are not present for the current case are set to the system-missing value for numeric variables or spaces for string variables.

These commands may combine any number of files, limited only by the machine's memory.

ADD FILES

ADD FILES

Per input file:
        /FILE={*,'FILE_NAME'}
        [/RENAME=(SRC_NAMES=TARGET_NAMES)...]
        [/IN=VAR_NAME]
        [/SORT]

Once per command:
        [/BY VAR_LIST[({D|A})] [VAR_LIST[({D|A})]...]]
        [/DROP=VAR_LIST]
        [/KEEP=VAR_LIST]
        [/FIRST=VAR_NAME]
        [/LAST=VAR_NAME]
        [/MAP]

ADD FILES adds cases from multiple input files. The output, which replaces the active dataset, consists all of the cases in all of the input files.

ADD FILES shares the bulk of its syntax with other PSPP commands for combining multiple data files (see Common Syntax for details).

When BY is not used, the output of ADD FILES consists of all the cases from the first input file specified, followed by all the cases from the second file specified, and so on. When BY is used, the output is additionally sorted on the BY variables.

When ADD FILES creates an output case, variables that are not part of the input file from which the case was drawn are set to the system-missing value for numeric variables or spaces for string variables.

MATCH FILES

MATCH FILES

Per input file:
        /{FILE,TABLE}={*,'FILE_NAME'}
        [/RENAME=(SRC_NAMES=TARGET_NAMES)...]
        [/IN=VAR_NAME]
        [/SORT]

Once per command:
        /BY VAR_LIST[({D|A}] [VAR_LIST[({D|A})]...]
        [/DROP=VAR_LIST]
        [/KEEP=VAR_LIST]
        [/FIRST=VAR_NAME]
        [/LAST=VAR_NAME]
        [/MAP]

MATCH FILES merges sets of corresponding cases in multiple input files into single cases in the output, combining their data.

MATCH FILES shares the bulk of its syntax with other PSPP commands for combining multiple data files (see Common Syntax for details).

How MATCH FILES matches up cases from the input files depends on whether BY is specified:

  • If BY is not used, MATCH FILES combines the first case from each input file to produce the first output case, then the second case from each input file for the second output case, and so on. If some input files have fewer cases than others, then the shorter files do not contribute to cases output after their input has been exhausted.

  • If BY is used, MATCH FILES combines cases from each input file that have identical values for the BY variables.

    When BY is used, TABLE subcommands may be used to introduce "table lookup files". TABLE has same syntax as FILE, and the RENAME, IN, and SORT subcommands may follow a TABLE in the same way as FILE. Regardless of the number of TABLEs, at least one FILE must specified. Table lookup files are treated in the same way as other input files for most purposes and, in particular, table lookup files must be sorted on the BY variables or the SORT subcommand must be specified for that TABLE.

    Cases in table lookup files are not consumed after they have been used once. This means that data in table lookup files can correspond to any number of cases in FILE input files. Table lookup files are analogous to lookup tables in traditional relational database systems.

    If a table lookup file contains more than one case with a given set of BY variables, only the first case is used.

When MATCH FILES creates an output case, variables that are only in files that are not present for the current case are set to the system-missing value for numeric variables or spaces for string variables.

UPDATE

UPDATE

Per input file:
        /FILE={*,'FILE_NAME'}
        [/RENAME=(SRC_NAMES=TARGET_NAMES)...]
        [/IN=VAR_NAME]
        [/SORT]

Once per command:
        /BY VAR_LIST[({D|A})] [VAR_LIST[({D|A})]]...
        [/DROP=VAR_LIST]
        [/KEEP=VAR_LIST]
        [/MAP]

UPDATE updates a "master file" by applying modifications from one or more "transaction files".

UPDATE shares the bulk of its syntax with other PSPP commands for combining multiple data files (see Common Syntax for details).

At least two FILE subcommands must be specified. The first FILE subcommand names the master file, and the rest name transaction files. Every input file must either be sorted on the variables named on the BY subcommand, or the SORT subcommand must be used just after the FILE subcommand for that input file.

UPDATE uses the variables specified on the BY subcommand, which is required, to attempt to match each case in a transaction file with a case in the master file:

  • When a match is found, then the values of the variables present in the transaction file replace those variables' values in the new active file. If there are matching cases in more than more transaction file, PSPP applies the replacements from the first transaction file, then from the second transaction file, and so on. Similarly, if a single transaction file has cases with duplicate BY values, then those are applied in order to the master file.

    When a variable in a transaction file has a missing value or when a string variable's value is all blanks, that value is never used to update the master file.

  • If a case in the master file has no matching case in any transaction file, then it is copied unchanged to the output.

  • If a case in a transaction file has no matching case in the master file, then it causes a new case to be added to the output, initialized from the values in the transaction file.

Manipulating Variables

Every value in a dataset is associated with a variable. Variables describe variable. Variables describe what the values represent and properties of those values, such as the format in which they should be displayed, whether they are numeric or alphabetic and how missing values should be represented. There are several utility commands for examining and adjusting variables.

ADD VALUE LABELS

ADD VALUE LABELS has the same syntax and purpose as VALUE LABELS, but it does not clear value labels from the variables before adding the ones specified.

ADD VALUE LABELS
        /VAR_LIST VALUE 'LABEL' [VALUE 'LABEL']...

DELETE VARIABLES

DELETE VARIABLES deletes the specified variables from the dictionary.

DELETE VARIABLES VAR_LIST.

DELETE VARIABLES should not be used after defining transformations but before executing a procedure. If it is anyhow, it causes the data to be read. If it is used while TEMPORARY is in effect, it causes the temporary transformations to become permanent.

DELETE VARIABLES may not be used to delete all variables from the dictionary; use NEW FILE instead.

DISPLAY

The DISPLAY command displays information about the variables in the active dataset. A variety of different forms of information can be requested. By default, all variables in the active dataset are displayed. However you can select variables of interest using the /VARIABLES subcommand.

DISPLAY [SORTED] NAMES [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] INDEX [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] LABELS [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] VARIABLES [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] DICTIONARY [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] SCRATCH [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] ATTRIBUTES [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] @ATTRIBUTES [[/VARIABLES=]VAR_LIST].
DISPLAY [SORTED] VECTORS.

The following keywords primarily cause information about variables to be displayed. With these keywords, by default information is displayed about all variable in the active dataset, in the order that variables occur in the active dataset dictionary. The SORTED keyword causes output to be sorted alphabetically by variable name.

  • NAMES
    The variables' names are displayed.

  • INDEX
    The variables' names are displayed along with a value describing their position within the active dataset dictionary.

  • LABELS
    Variable names, positions, and variable labels are displayed.

  • VARIABLES
    Variable names, positions, print and write formats, and missing values are displayed.

  • DICTIONARY
    Variable names, positions, print and write formats, missing values, variable labels, and value labels are displayed.

  • SCRATCH
    Displays Variablen ames, for scratch variables only. variables](../language/datasets/scratch-variables.md) only.

  • ATTRIBUTES
    Datafile and variable attributes are displayed, except attributes whose names begin with @ or $@.

  • @ATTRIBUTES
    All datafile and variable attributes, even those whose names begin with @ or $@.

With the VECTOR keyword, DISPLAY lists all the currently declared vectors. If the SORTED keyword is given, the vectors are listed in alphabetical order; otherwise, they are listed in textual order of definition within the PSPP syntax file.

For related commands, see DISPLAY DOCUMENTS and DISPLAY FILE LABEL.

FORMATS

FORMATS VAR_LIST (FMT_SPEC) [VAR_LIST (FMT_SPEC)]....

FORMATS set both print and write formats for the specified variables to the specified output format. format](../language/datasets/formats/index.md).

Specify a list of variables followed by a format specification in parentheses. The print and write formats of the specified variables will be changed. All of the variables listed together must have the same type and, for string variables, the same width.

Additional lists of variables and formats may be included following the first one.

FORMATS takes effect immediately. It is not affected by conditional and looping structures such as DO IF or LOOP.

LEAVE

LEAVE prevents the specified variables from being reinitialized whenever a new case is processed.

LEAVE VAR_LIST.

Normally, when a data file is processed, every variable in the active dataset is initialized to the system-missing value or spaces at the beginning of processing for each case. When a variable has been specified on LEAVE, this is not the case. Instead, that variable is initialized to 0 (not system-missing) or spaces for the first case. After that, it retains its value between cases.

This becomes useful for counters. For instance, in the example below the variable SUM maintains a running total of the values in the ITEM variable.

DATA LIST /ITEM 1-3.
COMPUTE SUM=SUM+ITEM.
PRINT /ITEM SUM.
LEAVE SUM
BEGIN DATA.
123
404
555
999
END DATA.

Partial output from this example:

123   123.00
404   527.00
555  1082.00
999  2081.00

It is best to use LEAVE command immediately before invoking a procedure command, because the left status of variables is reset by certain transformations—for instance, COMPUTE and IF. Left status is also reset by all procedure invocations.

MISSING VALUES

In many situations, the data available for analysis is incomplete, so that a placeholder must be used to indicate that the value is unknown. One way that missing values are represented, for numeric data, is the "system-missing value". "system-missing value". Another, more flexible way is through "user-missing values" which are determined on a per variable basis.

The MISSING VALUES command sets user-missing values for variables.

MISSING VALUES VAR_LIST (MISSING_VALUES).

where MISSING_VALUES takes one of the following forms:
        NUM1
        NUM1, NUM2
        NUM1, NUM2, NUM3
        NUM1 THRU NUM2
        NUM1 THRU NUM2, NUM3
        STRING1
        STRING1, STRING2
        STRING1, STRING2, STRING3
As part of a range, `LO` or `LOWEST` may take the place of NUM1;
`HI` or `HIGHEST` may take the place of NUM2.

MISSING VALUES sets user-missing values for numeric and string variables. Long string variables may have missing values, but characters after the first 8 bytes of the missing value must be spaces.

Specify a list of variables, followed by a list of their user-missing values in parentheses. Up to three discrete values may be given, or, for numeric variables only, a range of values optionally accompanied by a single discrete value. Ranges may be open-ended on one end, indicated through the use of the keyword LO or LOWEST or HI or HIGHEST.

The MISSING VALUES command takes effect immediately. It is not affected by conditional and looping constructs such as DO IF or LOOP.

MRSETS

MRSETS creates, modifies, deletes, and displays multiple response sets. A multiple response set is a set of variables that represent multiple responses to a survey question.

Multiple responses are represented in one of the two following ways:

  • A "multiple dichotomy set" is analogous to a survey question with a set of checkboxes. Each variable in the set is treated in a Boolean fashion: one value (the "counted value") means that the box was checked, and any other value means that it was not.

  • A "multiple category set" represents a survey question where the respondent is instructed to list up to N choices. Each variable represents one of the responses.

MRSETS
    /MDGROUP NAME=NAME VARIABLES=VAR_LIST VALUE=VALUE
     [CATEGORYLABELS={VARLABELS,COUNTEDVALUES}]
     [{LABEL='LABEL',LABELSOURCE=VARLABEL}]

    /MCGROUP NAME=NAME VARIABLES=VAR_LIST [LABEL='LABEL']

    /DELETE NAME={[NAMES],ALL}

    /DISPLAY NAME={[NAMES],ALL}

Any number of subcommands may be specified in any order.

The MDGROUP subcommand creates a new multiple dichotomy set or replaces an existing multiple response set. The NAME, VARIABLES, and VALUE specifications are required. The others are optional:

  • NAME specifies the name used in syntax for the new multiple dichotomy set. The name must begin with $; it must otherwise follow the rules for identifiers. follow the rules for identifiers.

  • VARIABLES specifies the variables that belong to the set. At least two variables must be specified. The variables must be all string or all numeric.

  • VALUE specifies the counted value. If the variables are numeric, the value must be an integer. If the variables are strings, then the value must be a string that is no longer than the shortest of the variables in the set (ignoring trailing spaces).

  • CATEGORYLABELS optionally specifies the source of the labels for each category in the set:

    − VARLABELS, the default, uses variable labels or, for variables without variable labels, variable names. PSPP warns if two variables have the same variable label, since these categories cannot be distinguished in output.

    − COUNTEDVALUES instead uses each variable's value label for the counted value. PSPP warns if two variables have the same value label for the counted value or if one of the variables lacks a value label, since such categories cannot be distinguished in output.

  • LABEL optionally specifies a label for the multiple response set. If neither LABEL nor LABELSOURCE=VARLABEL is specified, the set is unlabeled.

  • LABELSOURCE=VARLABEL draws the multiple response set's label from the first variable label among the variables in the set; if none of the variables has a label, the name of the first variable is used. LABELSOURCE=VARLABEL must be used with CATEGORYLABELS=COUNTEDVALUES. It is mutually exclusive with LABEL.

The MCGROUP subcommand creates a new multiple category set or replaces an existing multiple response set. The NAME and VARIABLES specifications are required, and LABEL is optional. Their meanings are as described above in MDGROUP. PSPP warns if two variables in the set have different value labels for a single value, since each of the variables in the set should have the same possible categories.

The DELETE subcommand deletes multiple response groups. A list of groups may be named within a set of required square brackets, or ALL may be used to delete all groups.

The DISPLAY subcommand displays information about defined multiple response sets. Its syntax is the same as the DELETE subcommand.

Multiple response sets are saved to and read from system files by, e.g., the SAVE and GET command. Otherwise, multiple response sets are currently used only by third party software.

NUMERIC

NUMERIC explicitly declares new numeric variables, optionally setting their output formats.

NUMERIC VAR_LIST [(FMT_SPEC)] [/VAR_LIST [(FMT_SPEC)]]...

Specify the names of the new numeric variables as VAR_LIST. If you wish to set the variables' output formats, follow their names by an output format in an output format in parentheses; otherwise, the default is F8.2.

Variables created with NUMERIC are initialized to the system-missing value.

PRINT FORMATS

PRINT FORMATS VAR_LIST (FMT_SPEC) [VAR_LIST (FMT_SPEC)]....

PRINT FORMATS sets the print formats for the specified variables to the specified format specification.

It has the same syntax as FORMATS, but PRINT FORMATS sets only print formats, not write formats.

RENAME VARIABLES

RENAME VARIABLES changes the names of variables in the active dataset.

RENAME VARIABLES (OLD_NAMES=NEW_NAMES)... .

Specify lists of the old variable names and new variable names, separated by an equals sign (=), within parentheses. There must be the same number of old and new variable names. Each old variable is renamed to the corresponding new variable name. Multiple parenthesized groups of variables may be specified. When the old and new variable names contain only a single variable name, the parentheses are optional.

RENAME VARIABLES takes effect immediately. It does not cause the data to be read.

RENAME VARIABLES may not be specified following TEMPORARY.

SORT VARIABLES

SORT VARIABLES reorders the variables in the active dataset's dictionary according to a chosen sort key.

SORT VARIABLES [BY]
    (NAME | TYPE | FORMAT | LABEL | VALUES | MISSING | MEASURE
     | ROLE | COLUMNS | ALIGNMENT | ATTRIBUTE NAME)
    [(D)].

The main specification is one of the following identifiers, which determines how the variables are sorted:

  • NAME
    Sorts the variables according to their names, in a case-insensitive fashion. However, when variable names differ only in a number at the end, they are sorted numerically. For example, VAR5 is sorted before VAR400 even though 4 precedes 5.

  • TYPE
    Sorts numeric variables before string variables, and shorter string variables before longer ones.

  • FORMAT
    Groups variables by print format; within a format, sorts narrower formats before wider ones; with the same format and width, sorts fewer decimal places before more decimal places. See PRINT FORMATS.

  • LABEL
    Sorts variables without a variable label before those with one. See VARIABLE LABELS.

  • VALUES
    Sorts variables without value labels before those with some. See VALUE LABELS.

  • MISSING
    Sorts variables without missing values before those with some. See MISSING VALUES.

  • MEASURE
    Sorts nominal variables first, followed by ordinal variables, followed by scale variables. See VARIABLE LEVEL.

  • ROLE
    Groups variables according to their role. See VARIABLE ROLE.

  • COLUMNS
    Sorts variables in ascending display width. See VARIABLE WIDTH.

  • ALIGNMENT
    Sorts variables according to their alignment, first left-aligned, then right-aligned, then centered. See VARIABLE ALIGNMENT.

  • ATTRIBUTE NAME
    Sorts variables according to the first value of their NAME attribute. Variables without attributes are sorted first. See VARIABLE ATTRIBUTE.

Only one sort criterion can be specified. The sort is "stable," so to sort on multiple criteria one may perform multiple sorts. For example, the following will sort primarily based on alignment, with variables that have the same alignment ordered based on display width:

SORT VARIABLES BY COLUMNS.
SORT VARIABLES BY ALIGNMENT.

Specify (D) to reverse the sort order.

STRING

STRING creates new string variables.

STRING VAR_LIST (FMT_SPEC) [/VAR_LIST (FMT_SPEC)] [...].

Specify a list of names for the variable you want to create, followed by the desired output format in parentheses. format](../language/datasets/formats/index.html) in parentheses. Variable widths are implicitly derived from the specified output formats. The created variables will be initialized to spaces.

If you want to create several variables with distinct output formats, you can either use two or more separate STRING commands, or you can specify further variable list and format specification pairs, each separated from the previous by a slash (/).

The following example is one way to create three string variables; Two of the variables have format A24 and the other A80:

STRING firstname lastname (A24) / address (A80).

Here is another way to achieve the same result:

STRING firstname lastname (A24).
STRING address (A80).

... and here is yet another way:

STRING firstname (A24).
STRING lastname (A24).
STRING address (A80).

VALUE LABELS

The values of a variable can be associated with explanatory text strings. In this way, a short value can stand for a longer, more descriptive label.

Both numeric and string variables can be given labels. For string variables, the values are case-sensitive, so that, for example, a capitalized value and its lowercase variant would have to be labeled separately if both are present in the data.

VALUE LABELS
        /VAR_LIST VALUE 'LABEL' [VALUE 'LABEL']...

VALUE LABELS allows values of variables to be associated with labels.

To set up value labels for one or more variables, specify the variable names after a slash (/), followed by a list of values and their associated labels, separated by spaces.

Value labels in output are normally broken into lines automatically. Put \n in a label string to force a line break at that point. The label may still be broken into lines at additional points.

Before VALUE LABELS is executed, any existing value labels are cleared from the variables specified. Use ADD VALUE LABELS to add value labels without clearing those already present.

VARIABLE ALIGNMENT

VARIABLE ALIGNMENT sets the alignment of variables for display editing purposes. It does not affect the display of variables in PSPP output.

VARIABLE ALIGNMENT
        VAR_LIST ( LEFT | RIGHT | CENTER )
        [ /VAR_LIST ( LEFT | RIGHT | CENTER ) ]
        .
        .
        .
        [ /VAR_LIST ( LEFT | RIGHT | CENTER ) ]

VARIABLE ATTRIBUTE

VARIABLE ATTRIBUTE adds, modifies, or removes user-defined attributes associated with variables in the active dataset. Custom variable attributes are not interpreted by PSPP, but they are saved as part of system files and may be used by other software that reads them.

VARIABLE ATTRIBUTE
         VARIABLES=VAR_LIST
         ATTRIBUTE=NAME('VALUE') [NAME('VALUE')]...
         ATTRIBUTE=NAME[INDEX]('VALUE') [NAME[INDEX]('VALUE')]...
         DELETE=NAME [NAME]...
         DELETE=NAME[INDEX] [NAME[INDEX]]...

The required VARIABLES subcommand must come first. Specify the variables to which the following ATTRIBUTE or DELETE subcommand should apply.

Use the ATTRIBUTE subcommand to add or modify custom variable attributes. Specify the name of the attribute as an identifier, followed by the desired identifier, followed by the desired value, in parentheses, as a quoted string. The specified attributes are then added or modified in the variables specified on VARIABLES. Attribute names that begin with $ are reserved for PSPP's internal use, and attribute names that begin with @ or $@ are not displayed by most PSPP commands that display other attributes. Other attribute names are not treated specially.

Attributes may also be organized into arrays. To assign to an array element, add an integer array index enclosed in square brackets ([ and ]) between the attribute name and value. Array indexes start at 1, not 0. An attribute array that has a single element (number 1) is not distinguished from a non-array attribute.

Use the DELETE subcommand to delete an attribute from the variable specified on VARIABLES. Specify an attribute name by itself to delete an entire attribute, including all array elements for attribute arrays. Specify an attribute name followed by an array index in square brackets to delete a single element of an attribute array. In the latter case, all the array elements numbered higher than the deleted element are shifted down, filling the vacated position.

To associate custom attributes with the entire active dataset, instead of with particular variables, use DATAFILE ATTRIBUTE instead.

VARIABLE ATTRIBUTE takes effect immediately. It is not affected by conditional and looping structures such as DO IF or LOOP.

VARIABLE LABELS

Each variable can have a "label" to supplement its name. Whereas a variable name is a concise, easy-to-type mnemonic for the variable, a label may be longer and more descriptive.

VARIABLE LABELS
        VARIABLE 'LABEL'
        [VARIABLE 'LABEL']...

VARIABLE LABELS associates explanatory names with variables. This name, called a "variable label", is displayed by statistical procedures.

Specify each variable followed by its label as a quoted string. Variable-label pairs may be separated by an optional slash /.

If a listed variable already has a label, the new one replaces it. Specifying an empty string as the label, e.g. '', removes a label.

VARIABLE LEVEL

VARIABLE LEVEL variables ({SCALE | NOMINAL | ORDINAL})...

VARIABLE LEVEL sets the measurement level of the listed variables. level](../language/datasets/variables.md) of the listed variables.

VARIABLE ROLE

VARIABLE ROLE
        /ROLE VAR_LIST
        [/ROLE VAR_LIST]...

VARIABLE ROLE sets the intended role of a variable for use in dialog boxes in graphical user interfaces. Each ROLE specifies one of the following roles for the variables that follow it:

  • INPUT
    An input variable, such as an independent variable.

  • TARGET
    An output variable, such as an dependent variable.

  • BOTH
    A variable used for input and output.

  • NONE
    No role assigned. (This is a variable's default role.)

  • PARTITION
    Used to break the data into groups for testing.

  • SPLIT
    No meaning except for certain third party software. (This role's meaning is unrelated to SPLIT FILE.)

The PSPPIRE GUI does not yet use variable roles.

VARIABLE WIDTH

VARIABLE WIDTH
        VAR_LIST (width)
        [ /VAR_LIST (width) ]
        .
        .
        .
        [ /VAR_LIST (width) ]

VARIABLE WIDTH sets the column width of variables for display editing purposes. It does not affect the display of variables in the PSPP output.

VECTOR

Two possible syntaxes:
        VECTOR VEC_NAME=VAR_LIST.
        VECTOR VEC_NAME_LIST(COUNT [FORMAT]).

VECTOR allows a group of variables to be accessed as if they were consecutive members of an array with a vector(index) notation.

To make a vector out of a set of existing variables, specify a name for the vector followed by an equals sign (=) and the variables to put in the vector. The variables must be all numeric or all string, and string variables must have the same width.

To make a vector and create variables at the same time, specify one or more vector names followed by a count in parentheses. This will create variables named VEC1 through VEC<count>. By default, the new variables are numeric with format F8.2, but an alternate format may be specified inside the parentheses before or after the count and separated from it by white space or a comma. With a string format such as A8, the variables will be string variables; with a numeric format, they will be numeric. Variable names including the suffixes may not exceed 64 characters in length, and none of the variables may exist prior to VECTOR.

Vectors created with VECTOR disappear after any procedure or procedure-like command is executed. The variables contained in the vectors remain, unless they are scratch variables. variables](../language/datasets/scratch-variables.md).

Variables within a vector may be referenced in expressions using vector(index) syntax.

WRITE FORMATS

WRITE FORMATS VAR_LIST (FMT_SPEC) [VAR_LIST (FMT_SPEC)]....

WRITE FORMATS sets the write formats for the specified variables to the specified format specification. It has the same syntax as FORMATS, but WRITE FORMATS sets only write formats, not print formats.

Transforming Data

The PSPP procedures in this chapter manipulate data and prepare the active dataset for later analyses. They do not produce output.

AGGREGATE

AGGREGATE
        [OUTFILE={*,'FILE_NAME',FILE_HANDLE} [MODE={REPLACE,ADDVARIABLES}]]
        [/MISSING=COLUMNWISE]
        [/PRESORTED]
        [/DOCUMENT]
        [/BREAK=VAR_LIST]
        /DEST_VAR['LABEL']...=AGR_FUNC(SRC_VARS[, ARGS]...)...

AGGREGATE summarizes groups of cases into single cases. It divides cases into groups that have the same values for one or more variables called "break variables". Several functions are available for summarizing case contents.

The AGGREGATE syntax consists of subcommands to control its behavior, all of which are optional, followed by one or more destination variable assigments, each of which uses an aggregation function to define how it is calculated.

The OUTFILE subcommand, which must be first, names the destination for AGGREGATE output. It may name a system file by file name or file handle, a dataset by its name, or * to replace the active dataset. AGGREGATE writes its output to this file.

With OUTFILE=* only, MODE may be specified immediately afterward with the value ADDVARIABLES or REPLACE:

  • With REPLACE, the default, the active dataset is replaced by a new dataset which contains just the break variables and the destination varibles. The new file contains as many cases as there are unique combinations of the break variables.

  • With ADDVARIABLES, the destination variables are added to those in the existing active dataset. Cases that have the same combination of values in their break variables receive identical values for the destination variables. The number of cases in the active dataset remains unchanged. The data must be sorted on the break variables, that is, ADDVARIABLES implies PRESORTED

Without OUTFILE, AGGREGATE acts as if OUTFILE=* MODE=ADDVARIABLES were specified.

By default, AGGREGATE first sorts the data on the break variables. If the active dataset is already sorted or grouped by the break variables, specify PRESORTED to save time. With MODE=ADDVARIABLES, the data must be pre-sorted.

Specify DOCUMENT to copy the documents from the active dataset into the aggregate file. Otherwise, the aggregate file does not contain any documents, even if the aggregate file replaces the active dataset.

Normally, AGGREGATE produces a non-missing value whenever there is enough non-missing data for the aggregation function in use, that is, just one non-missing value or, for the SD and SD. aggregation functions, two non-missing values. Specify /MISSING=COLUMNWISE to make AGGREGATE output a missing value when one or more of the input values are missing.

The BREAK subcommand is optionally but usually present. On BREAK, list the variables used to divide the active dataset into groups to be summarized.

AGGREGATE is particular about the order of subcommands. OUTFILE must be first, followed by MISSING. PRESORTED and DOCUMENT follow MISSING, in either order, followed by BREAK, then followed by aggregation variable specifications.

At least one set of aggregation variables is required. Each set comprises a list of aggregation variables, an equals sign (=), the name of an aggregation function (see the list below), and a list of source variables in parentheses. A few aggregation functions do not accept source variables, and some aggregation functions expect additional arguments after the source variable names.

AGGREGATE typically creates aggregation variables with no variable label, value labels, or missing values. Their default print and write formats depend on the aggregation function used, with details given in the table below. A variable label for an aggregation variable may be specified just after the variable's name in the aggregation variable list.

Each set must have exactly as many source variables as aggregation variables. Each aggregation variable receives the results of applying the specified aggregation function to the corresponding source variable.

The following aggregation functions may be applied only to numeric variables:

  • MEAN(VAR_NAME...)
    Arithmetic mean. Limited to numeric values. The default format is F8.2.

  • MEDIAN(VAR_NAME...)
    The median value. Limited to numeric values. The default format is F8.2.

  • SD(VAR_NAME...)
    Standard deviation of the mean. Limited to numeric values. The default format is F8.2.

  • SUM(VAR_NAME...)
    Sum. Limited to numeric values. The default format is F8.2.

    These aggregation functions may be applied to numeric and string variables:

  • CGT(VAR_NAME..., VALUE)
    CLT(VAR_NAME..., VALUE)
    CIN(VAR_NAME..., LOW, HIGH)
    COUT(VAR_NAME..., LOW, HIGH)
    Total weight of cases greater than or less than VALUE or inside or outside the closed range [LOW,HIGH], respectively. The default format is F5.3.

  • FGT(VAR_NAME..., VALUE)
    FLT(VAR_NAME..., VALUE)
    FIN(VAR_NAME..., LOW, HIGH)
    FOUT(VAR_NAME..., LOW, HIGH)
    Fraction of values greater than or less than VALUE or inside or outside the closed range [LOW,HIGH], respectively. The default format is F5.3.

  • FIRST(VAR_NAME...)
    LAST(VAR_NAME...)
    First or last non-missing value, respectively, in break group. The aggregation variable receives the complete dictionary information from the source variable. The sort performed by AGGREGATE (and by SORT CASES) is stable. This means that the first (or last) case with particular values for the break variables before sorting is also the first (or last) case in that break group after sorting.

  • MIN(VAR_NAME...)
    MAX(VAR_NAME...)
    Minimum or maximum value, respectively. The aggregation variable receives the complete dictionary information from the source variable.

  • N(VAR_NAME...)
    NMISS(VAR_NAME...)
    Total weight of non-missing or missing values, respectively. The default format is F7.0 if weighting is not enabled, F8.2 if it is (see WEIGHT).

  • NU(VAR_NAME...)
    NUMISS(VAR_NAME...)
    Count of non-missing or missing values, respectively, ignoring case weights. The default format is F7.0.

  • PGT(VAR_NAME..., VALUE)
    PLT(VAR_NAME..., VALUE)
    PIN(VAR_NAME..., LOW, HIGH)
    POUT(VAR_NAME..., LOW, HIGH)
    Percentage between 0 and 100 of values greater than or less than VALUE or inside or outside the closed range [LOW,HIGH], respectively. The default format is F5.1.

These aggregation functions do not accept source variables:

  • N
    Total weight of cases aggregated to form this group. The default format is F7.0 if weighting is not enabled, F8.2 if it is (see WEIGHT).

  • NU
    Count of cases aggregated to form this group, ignoring case weights. The default format is F7.0.

Aggregation functions compare string values in terms of Unicode character codes.

The aggregation functions listed above exclude all user-missing values from calculations. To include user-missing values, insert a period (.) at the end of the function name. (e.g. SUM.). (Be aware that specifying such a function as the last token on a line causes the period to be interpreted as the end of the command.)

AGGREGATE both ignores and cancels the current SPLIT FILE settings.

Example

The personnel.sav dataset provides the occupations and salaries of many individuals. For many purposes however such detailed information is not interesting, but often the aggregated statistics of each occupation are of interest. Here, the AGGREGATE command is used to calculate the mean, the median and the standard deviation of each occupation.

GET FILE="personnel.sav".
AGGREGATE OUTFILE=* MODE=REPLACE
        /BREAK=occupation
        /occ_mean_salary=MEAN(salary)
        /occ_median_salary=MEDIAN(salary)
        /occ_std_dev_salary=SD(salary).
LIST.

Since we chose the MODE=REPLACE option, cases for the individual persons are no longer present. They have each been replaced by a single case per aggregated value.

                                Data List
┌──────────────────┬───────────────┬─────────────────┬──────────────────┐
│    occupation    │occ_mean_salary│occ_median_salary│occ_std_dev_salary│
├──────────────────┼───────────────┼─────────────────┼──────────────────┤
│Artist            │       37836.18│         34712.50│           7631.48│
│Baker             │       45075.20│         45075.20│           4411.21│
│Barrister         │       39504.00│         39504.00│                 .│
│Carpenter         │       39349.11│         36190.04│           7453.40│
│Cleaner           │       41142.50│         39647.49│          14378.98│
│Cook              │       40357.79│         43194.00│          11064.51│
│Manager           │       46452.14│         45657.56│           6901.69│
│Mathematician     │       34531.06│         34763.06│           5267.68│
│Painter           │       45063.55│         45063.55│          15159.67│
│Payload Specialist│       34355.72│         34355.72│                 .│
│Plumber           │       40413.91│         40410.00│           4726.05│
│Scientist         │       36687.07│         36803.83│          10873.54│
│Scrientist        │       42530.65│         42530.65│                 .│
│Tailor            │       34586.79│         34586.79│           3728.98│
└──────────────────┴───────────────┴─────────────────┴──────────────────┘

Some values for the standard deviation are blank because there is only one case with the respective occupation.

AUTORECODE

AUTORECODE VARIABLES=SRC_VARS INTO DEST_VARS
        [ /DESCENDING ]
        [ /PRINT ]
        [ /GROUP ]
        [ /BLANK = {VALID, MISSING} ]

The AUTORECODE procedure considers the N values that a variable takes on and maps them onto values 1...N on a new numeric variable.

Subcommand VARIABLES is the only required subcommand and must come first. Specify VARIABLES, an equals sign (=), a list of source variables, INTO, and a list of target variables. There must the same number of source and target variables. The target variables must not already exist.

AUTORECODE ordinarily assigns each increasing non-missing value of a source variable (for a string, this is based on character code comparisons) to consecutive values of its target variable. For example, the smallest non-missing value of the source variable is recoded to value 1, the next smallest to 2, and so on. If the source variable has user-missing values, they are recoded to consecutive values just above the non-missing values. For example, if a source variables has seven distinct non-missing values, then the smallest missing value would be recoded to 8, the next smallest to 9, and so on.

Use DESCENDING to reverse the sort order for non-missing values, so that the largest non-missing value is recoded to 1, the second-largest to 2, and so on. Even with DESCENDING, user-missing values are still recoded in ascending order just above the non-missing values.

The system-missing value is always recoded into the system-missing variable in target variables.

If a source value has a value label, then that value label is retained for the new value in the target variable. Otherwise, the source value itself becomes each new value's label.

Variable labels are copied from the source to target variables.

PRINT is currently ignored.

The GROUP subcommand is relevant only if more than one variable is to be recoded. It causes a single mapping between source and target values to be used, instead of one map per variable. With GROUP, user-missing values are taken from the first source variable that has any user-missing values.

If /BLANK=MISSING is given, then string variables which contain only whitespace are recoded as SYSMIS. If /BLANK=VALID is specified then they are allocated a value like any other. /BLANK is not relevant to numeric values. /BLANK=VALID is the default.

AUTORECODE is a procedure. It causes the data to be read.

Example

In the file personnel.sav, the variable occupation is a string variable. Except for data of a purely commentary nature, string variables are generally a bad idea. One reason is that data entry errors are easily overlooked. This has happened in personnel.sav; one entry which should read "Scientist" has been mistyped as "Scrientist". The syntax below shows how to correct this error in the DO IF clause1, which then uses AUTORECODE to create a new numeric variable which takes recoded values of occupation. Finally, we remove the old variable and rename the new variable to the name of the old variable:

get file='personnel.sav'.

* Correct a typing error in the original file.
do if occupation = "Scrientist".
 compute occupation = "Scientist".
end if.

autorecode
   variables = occupation into occ
   /blank = missing.

* Delete the old variable.
delete variables occupation.

* Rename the new variable to the old variable's name.
rename variables (occ = occupation).

* Inspect the new variable.
display dictionary /variables=occupation.

Notice, in the output below, how the new variable has been automatically allocated value labels which correspond to the strings of the old variable. This means that in future analyses the descriptive strings are reported instead of the numeric values.

                                   Variables
+----------+--------+--------------+-----+-----+---------+----------+---------+
|          |        |  Measurement |     |     |         |   Print  |  Write  |
|Name      |Position|     Level    | Role|Width|Alignment|  Format  |  Format |
+----------+--------+--------------+-----+-----+---------+----------+---------+
|occupation|       6|Unknown       |Input|    8|Right    |F2.0      |F2.0     |
+----------+--------+--------------+-----+-----+---------+----------+---------+

            Value Labels
+---------------+------------------+
|Variable Value |       Label      |
+---------------+------------------+
|occupation 1   |Artist            |
|           2   |Baker             |
|           3   |Barrister         |
|           4   |Carpenter         |
|           5   |Cleaner           |
|           6   |Cook              |
|           7   |Manager           |
|           8   |Mathematician     |
|           9   |Painter           |
|           10  |Payload Specialist|
|           11  |Plumber           |
|           12  |Scientist         |
|           13  |Tailor            |
+---------------+------------------+

  1. One must use care when correcting such data input errors rather than simply marking them as missing. For example, if an occupation has been entered "Barister", did the person mean "Barrister" or "Barista"? ↩

COMPUTE

COMPUTE VARIABLE = EXPRESSION.
   or
COMPUTE vector(INDEX) = EXPRESSION.

COMPUTE assigns the value of an expression to a target variable. For each case, the expression is evaluated and its value assigned to the target variable. Numeric and string variables may be assigned. When a string expression's width differs from the target variable's width, the string result of the expression is truncated or padded with spaces on the right as necessary. The expression and variable types must match.

For numeric variables only, the target variable need not already exist. Numeric variables created by COMPUTE are assigned an F8.2 output format. String variables must be declared before they can be used as targets for COMPUTE.

The target variable may be specified as an element of a vector. In this case, an expression INDEX must be specified in parentheses following the vector name. The expression INDEX must evaluate to a numeric value that, after rounding down to the nearest integer, is a valid index for the named vector.

Using COMPUTE to assign to a variable specified on LEAVE resets the variable's left state. Therefore, LEAVE should be specified following COMPUTE, not before.

COMPUTE is a transformation. It does not cause the active dataset to be read.

When COMPUTE is specified following TEMPORARY, the LAG function may not be used.

Example

The dataset physiology.sav contains the height and weight of persons. For some purposes, neither height nor weight alone is of interest. Epidemiologists are often more interested in the "body mass index" which can sometimes be used as a predictor for clinical conditions. The body mass index is defined as the weight of the person in kilograms divided by the square of the person's height in metres.1

get file='physiology.sav'.

* height is in mm so we must divide by 1000 to get metres.
compute bmi = weight / (height/1000)**2.
variable label bmi "Body Mass Index".

descriptives /weight height bmi.

This syntax shows how you can use COMPUTE to generate a new variable called bmi and have every case's value calculated from the existing values of weight and height. It also shows how you can add a label to this new variable, so that a more descriptive label appears in subsequent analyses, and this can be seen in the output from the DESCRIPTIVES command, below.

The expression which follows the = sign can be as complicated as necessary. See Expressions for a full description of the language accepted.

                  Descriptive Statistics
┌─────────────────────┬──┬───────┬───────┬───────┬───────┐
│                     │ N│  Mean │Std Dev│Minimum│Maximum│
├─────────────────────┼──┼───────┼───────┼───────┼───────┤
│Weight in kilograms  │40│  72.12│  26.70│  ─55.6│   92.1│
│Height in millimeters│40│1677.12│ 262.87│    179│   1903│
│Body Mass Index      │40│  67.46│ 274.08│ ─21.62│1756.82│
│Valid N (listwise)   │40│       │       │       │       │
│Missing N (listwise) │ 0│       │       │       │       │
└─────────────────────┴──┴───────┴───────┴───────┴───────┘

  1. Since BMI is a quantity with a ratio scale and has units, the term "index" is a misnomer, but that is what it is called. ↩

FLIP

FLIP /VARIABLES=VAR_LIST /NEWNAMES=VAR_NAME.

FLIP transposes rows and columns in the active dataset. It causes cases to be swapped with variables, and vice versa.

All variables in the transposed active dataset are numeric. String variables take on the system-missing value in the transposed file.

N subcommands are required. If specified, the VARIABLES subcommand selects variables to be transformed into cases, and variables not specified are discarded. If the VARIABLES subcommand is omitted, all variables are selected for transposition.

The variables specified by NEWNAMES, which must be a string variable, is used to give names to the variables created by FLIP. Only the first 8 characters of the variable are used. If NEWNAMES is not specified then the default is a variable named CASE_LBL, if it exists. If it does not then the variables created by FLIP are named VAR000 through VAR999, then VAR1000, VAR1001, and so on.

When a NEWNAMES variable is available, the names must be canonicalized before becoming variable names. Invalid characters are replaced by letter V in the first position, or by _ in subsequent positions. If the name thus generated is not unique, then numeric extensions are added, starting with 1, until a unique name is found or there are no remaining possibilities. If the latter occurs then the FLIP operation aborts.

The resultant dictionary contains a CASE_LBL variable, a string variable of width 8, which stores the names of the variables in the dictionary before the transposition. Variables names longer than 8 characters are truncated. If FLIP is called again on this dataset, the CASE_LBL variable can be passed to the NEWNAMES subcommand to recreate the original variable names.

FLIP honors N OF CASES. It ignores TEMPORARY, so that "temporary" transformations become permanent.

Example

In the syntax below, data has been entered using DATA LIST such that the first variable in the dataset is a string variable containing a description of the other data for the case. Clearly this is not a convenient arrangement for performing statistical analyses, so it would have been better to think a little more carefully about how the data should have been arranged. However often the data is provided by some third party source, and you have no control over the form. Fortunately, we can use FLIP to exchange the variables and cases in the active dataset.

data list notable list /heading (a16) v1 v2 v3 v4 v5 v6
begin data.
date-of-birth 1970 1989 2001 1966 1976 1982
sex 1 0 0 1 0 1
score 10 10 9 3 8 9
end data.

echo 'Before FLIP:'.
display variables.
list.

flip /variables = all /newnames = heading.

echo 'After FLIP:'.
display variables.
list.

As you can see in the results below, before the FLIP command has run there are seven variables (six containing data and one for the heading) and three cases. Afterwards there are four variables (one per case, plus the CASE_LBL variable) and six cases. You can delete the CASE_LBL variable (see DELETE VARIABLES) if you don't need it.

Before FLIP:

                  Variables
┌───────┬────────┬────────────┬────────────┐
│Name   │Position│Print Format│Write Format│
├───────┼────────┼────────────┼────────────┤
│heading│       1│A16         │A16         │
│v1     │       2│F8.2        │F8.2        │
│v2     │       3│F8.2        │F8.2        │
│v3     │       4│F8.2        │F8.2        │
│v4     │       5│F8.2        │F8.2        │
│v5     │       6│F8.2        │F8.2        │
│v6     │       7│F8.2        │F8.2        │
└───────┴────────┴────────────┴────────────┘

                           Data List
┌─────────────┬───────┬───────┬───────┬───────┬───────┬───────┐
│   heading   │   v1  │   v2  │   v3  │   v4  │   v5  │   v6  │
├─────────────┼───────┼───────┼───────┼───────┼───────┼───────┤
│date─of─birth│1970.00│1989.00│2001.00│1966.00│1976.00│1982.00│
│sex          │   1.00│    .00│    .00│   1.00│    .00│   1.00│
│score        │  10.00│  10.00│   9.00│   3.00│   8.00│   9.00│
└─────────────┴───────┴───────┴───────┴───────┴───────┴───────┘

After FLIP:

                     Variables
┌─────────────┬────────┬────────────┬────────────┐
│Name         │Position│Print Format│Write Format│
├─────────────┼────────┼────────────┼────────────┤
│CASE_LBL     │       1│A8          │A8          │
│date_of_birth│       2│F8.2        │F8.2        │
│sex          │       3│F8.2        │F8.2        │
│score        │       4│F8.2        │F8.2        │
└─────────────┴────────┴────────────┴────────────┘

             Data List
┌────────┬─────────────┬────┬─────┐
│CASE_LBL│date_of_birth│ sex│score│
├────────┼─────────────┼────┼─────┤
│v1      │      1970.00│1.00│10.00│
│v2      │      1989.00│ .00│10.00│
│v3      │      2001.00│ .00│ 9.00│
│v4      │      1966.00│1.00│ 3.00│
│v5      │      1976.00│ .00│ 8.00│
│v6      │      1982.00│1.00│ 9.00│
└────────┴─────────────┴────┴─────┘

IF

  IF CONDITION VARIABLE=EXPRESSION.
or
  IF CONDITION vector(INDEX)=EXPRESSION.

The IF transformation evaluates a test expression and, if it is true, assigns the value of a target expression to a target variable.

Specify a boolean-valued test expression to be tested following the expression to be tested following the IF keyword. The test expression is evaluated for each case:

  • If it is true, then the target expression is evaluated and assigned to the specified variable.

  • If it is false or missing, nothing is done.

Numeric and string variables may be assigned. When a string expression's width differs from the target variable's width, the string result is truncated or padded with spaces on the right as necessary. The expression and variable types must match.

The target variable may be specified as an element of a vector. In this case, a vector index expression must be specified in parentheses following the vector name. The index expression must evaluate to a numeric value that, after rounding down to the nearest integer, is a valid index for the named vector.

Using IF to assign to a variable specified on LEAVE resets the variable's left state. Therefore, use LEAVE after IF, not before.

When IF follows TEMPORARY, the LAG function may not be used.

RECODE

The RECODE command is used to transform existing values into other, user specified values. The general form is:

RECODE SRC_VARS
        (SRC_VALUE SRC_VALUE ... = DEST_VALUE)
        (SRC_VALUE SRC_VALUE ... = DEST_VALUE)
        (SRC_VALUE SRC_VALUE ... = DEST_VALUE) ...
         [INTO DEST_VARS].

Following the RECODE keyword itself comes SRC_VARS, a list of variables whose values are to be transformed. These variables must all string or all numeric variables.

After the list of source variables, there should be one or more "mappings". Each mapping is enclosed in parentheses, and contains the source values and a destination value separated by a single =. The source values are used to specify the values in the dataset which need to change, and the destination value specifies the new value to which they should be changed. Each SRC_VALUE may take one of the following forms:

  • NUMBER (numeric source variables only)
    Matches a number.

  • STRING (string source variables only)
    Matches a string enclosed in single or double quotes.

  • NUM1 THRU NUM2 (numeric source variables only)
    Matches all values in the range between NUM1 and NUM2, including both endpoints of the range. NUM1 should be less than NUM2. Open-ended ranges may be specified using LO or LOWEST for NUM1 or HI or HIGHEST for NUM2.

  • MISSING
    Matches system missing and user missing values.

  • SYSMIS (numeric source variables only)
    Match system-missing values.

  • ELSE
    Matches any values that are not matched by any other SRC_VALUE. This should appear only as the last mapping in the command.

After the source variables comes an = and then the DEST_VALUE, which may take any of the following forms:

  • NUMBER (numeric destination variables only)
    A literal numeric value to which the source values should be changed.

  • STRING (string destination variables only)
    A literal string value (enclosed in quotation marks) to which the source values should be changed. This implies the destination variable must be a string variable.

  • SYSMIS (numeric destination variables only)
    The keyword SYSMIS changes the value to the system missing value. This implies the destination variable must be numeric.

  • COPY
    The special keyword COPY means that the source value should not be modified, but copied directly to the destination value. This is meaningful only if INTO DEST_VARS is specified.

Mappings are considered from left to right. Therefore, if a value is matched by a SRC_VALUE from more than one mapping, the first (leftmost) mapping which matches is considered. Any subsequent matches are ignored.

The clause INTO DEST_VARS is optional. The behaviour of the command is slightly different depending on whether it appears or not:

  • Without INTO DEST_VARS, then values are recoded "in place". This means that the recoded values are written back to the source variables from whence the original values came. In this case, the DEST_VALUE for every mapping must imply a value which has the same type as the SRC_VALUE. For example, if the source value is a string value, it is not permissible for DEST_VALUE to be SYSMIS or another forms which implies a numeric result. It is also not permissible for DEST_VALUE to be longer than the width of the source variable.

    The following example recodes two numeric variables x and y in place. 0 becomes 99, the values 1 to 10 inclusive are unchanged, values 1000 and higher are recoded to the system-missing value, and all other values are changed to 999:

    RECODE x y
        (0 = 99)
        (1 THRU 10 = COPY)
        (1000 THRU HIGHEST = SYSMIS)
        (ELSE = 999).

  • With INTO DEST_VARS, recoded values are written into the variables specified in DEST_VARS, which must therefore contain a list of valid variable names. The number of variables in DEST_VARS must be the same as the number of variables in SRC_VARS and the respective order of the variables in DEST_VARS corresponds to the order of SRC_VARS. That is to say, the recoded value whose original value came from the Nth variable in SRC_VARS is placed into the Nth variable in DEST_VARS. The source variables are unchanged. If any mapping implies a string as its destination value, then the respective destination variable must already exist, or have been declared using STRING or another transformation. Numeric variables however are automatically created if they don't already exist.

    The following example deals with two source variables, a and b which contain string values. Hence there are two destination variables v1 and v2. Any cases where a or b contain the values apple, pear or pomegranate result in v1 or v2 being filled with the string fruit whilst cases with tomato, lettuce or carrot result in vegetable. Other values produce the result unknown:

    STRING v1 (A20).
STRING v2 (A20).

RECODE a b
        ("apple" "pear" "pomegranate" = "fruit")
        ("tomato" "lettuce" "carrot" = "vegetable")
        (ELSE = "unknown")
        INTO v1 v2.

There is one special mapping, not mentioned above. If the source variable is a string variable then a mapping may be specified as (CONVERT). This mapping, if it appears must be the last mapping given and the INTO DEST_VARS clause must also be given and must not refer to a string variable. CONVERT causes a number specified as a string to be converted to a numeric value. For example it converts the string "3" into the numeric value 3 (note that it does not convert three into 3). If the string cannot be parsed as a number, then the system-missing value is assigned instead. In the following example, cases where the value of x (a string variable) is the empty string, are recoded to 999 and all others are converted to the numeric equivalent of the input value. The results are placed into the numeric variable y:

RECODE x ("" = 999) (CONVERT) INTO y.

It is possible to specify multiple recodings on a single command. Introduce additional recodings with a slash (/) to separate them from the previous recodings:

RECODE
    a (2 = 22) (ELSE = 99)
   /b (1 = 3) INTO z.

Here we have two recodings. The first affects the source variable a and recodes in-place the value 2 into 22 and all other values to 99. The second recoding copies the values of b into the variable z, changing any instances of 1 into 3.

SORT CASES

SORT CASES BY VAR_LIST[({D|A}] [ VAR_LIST[({D|A}] ] ...

SORT CASES sorts the active dataset by the values of one or more variables.

Specify BY and a list of variables to sort by. By default, variables are sorted in ascending order. To override sort order, specify (D) or (DOWN) after a list of variables to get descending order, or (A) or (UP) for ascending order. These apply to all the listed variables up until the preceding (A), (D), (UP) or (DOWN).

SORT CASES performs a stable sort, meaning that records with equal values of the sort variables have the same relative order before and after sorting. Thus, re-sorting an already sorted file does not affect the ordering of cases.

SORT CASES is a procedure. It causes the data to be read.

SORT CASES attempts to sort the entire active dataset in main memory. If workspace is exhausted, it falls back to a merge sort algorithm which creates numerous temporary files.

SORT CASES may not be specified following TEMPORARY.

Example

In the syntax below, the data from the file physiology.sav is sorted by two variables, viz sex in descending order and temperature in ascending order.

get file='physiology.sav'.
sort cases by sex (D) temperature(A).
list.

In the output below, you can see that all the cases with a sex of 1 (female) appear before those with a sex of 0 (male). This is because they have been sorted in descending order. Within each sex, the data is sorted on the temperature variable, this time in ascending order.

           Data List
┌───┬──────┬──────┬───────────┐
│sex│height│weight│temperature│
├───┼──────┼──────┼───────────┤
│  1│  1606│  56.1│      34.56│
│  1│   179│  56.3│      35.15│
│  1│  1609│  55.4│      35.46│
│  1│  1606│  56.0│      36.06│
│  1│  1607│  56.3│      36.26│
│  1│  1604│  56.0│      36.57│
│  1│  1604│  56.6│      36.81│
│  1│  1606│  56.3│      36.88│
│  1│  1604│  57.8│      37.32│
│  1│  1598│  55.6│      37.37│
│  1│  1607│  55.9│      37.84│
│  1│  1605│  54.5│      37.86│
│  1│  1603│  56.1│      38.80│
│  1│  1604│  58.1│      38.85│
│  1│  1605│  57.7│      38.98│
│  1│  1709│  55.6│      39.45│
│  1│  1604│ -55.6│      39.72│
│  1│  1601│  55.9│      39.90│
│  0│  1799│  90.3│      32.59│
│  0│  1799│  89.0│      33.61│
│  0│  1799│  90.6│      34.04│
│  0│  1801│  90.5│      34.42│
│  0│  1802│  87.7│      35.03│
│  0│  1793│  90.1│      35.11│
│  0│  1801│  92.1│      35.98│
│  0│  1800│  89.5│      36.10│
│  0│  1645│  92.1│      36.68│
│  0│  1698│  90.2│      36.94│
│  0│  1800│  89.6│      37.02│
│  0│  1800│  88.9│      37.03│
│  0│  1801│  88.9│      37.12│
│  0│  1799│  90.4│      37.33│
│  0│  1903│  91.5│      37.52│
│  0│  1799│  90.9│      37.53│
│  0│  1800│  91.0│      37.60│
│  0│  1799│  90.4│      37.68│
│  0│  1801│  91.7│      38.98│
│  0│  1801│  90.9│      39.03│
│  0│  1799│  89.3│      39.77│
│  0│  1884│  88.6│      39.97│
└───┴──────┴──────┴───────────┘

SORT CASES affects only the active file. It does not have any effect upon the physiology.sav file itself. For that, you would have to rewrite the file using the SAVE command.

Selecting Data

This chapter documents PSPP commands that temporarily or permanently select data records from the active dataset for analysis.

FILTER

FILTER BY VAR_NAME.
FILTER OFF.

FILTER allows a boolean-valued variable to be used to select cases from the data stream for processing.

To set up filtering, specify BY and a variable name. Keyword BY is optional but recommended. Cases which have a zero or system- or user-missing value are excluded from analysis, but not deleted from the data stream. Cases with other values are analyzed. To filter based on a different condition, use transformations such as COMPUTE or RECODE to compute a filter variable of the required form, then specify that variable on FILTER.

FILTER OFF turns off case filtering.

Filtering takes place immediately before cases pass to a procedure for analysis. Only one filter variable may be active at a time. Normally, case filtering continues until it is explicitly turned off with FILTER OFF. However, if FILTER is placed after TEMPORARY, it filters only the next procedure or procedure-like command.

N OF CASES

N [OF CASES] NUM_OF_CASES [ESTIMATED].

N OF CASES limits the number of cases processed by any procedures that follow it in the command stream. N OF CASES 100, for example, tells PSPP to disregard all cases after the first 100.

When N OF CASES is specified after TEMPORARY, it affects only the next procedure. Otherwise, cases beyond the limit specified are not processed by any later procedure.

If the limit specified on N OF CASES is greater than the number of cases in the active dataset, it has no effect.

When N OF CASES is used along with SAMPLE or SELECT IF, the case limit is applied to the cases obtained after sampling or case selection, regardless of how N OF CASES is placed relative to SAMPLE or SELECT IF in the command file. Thus, the commands N OF CASES 100 and SAMPLE .5 both randomly sample approximately half of the active dataset's cases, then select the first 100 of those sampled, regardless of their order in the command file.

N OF CASES with the ESTIMATED keyword gives an estimated number of cases before DATA LIST or another command to read in data. ESTIMATED never limits the number of cases processed by procedures. PSPP currently does not use case count estimates.

SAMPLE

SAMPLE NUM1 [FROM NUM2].

SAMPLE randomly samples a proportion of the cases in the active file. Unless it follows TEMPORARY, it permanently removes cases from the active dataset.

The proportion to sample may be expressed as a single number between 0 and 1. If N is the number of currently-selected cases in the active dataset, then SAMPLE K. will select approximately K×N cases.

The proportion to sample can also be specified in the style SAMPLE M FROM N. With this style, cases are selected as follows:

  1. If N is the number of currently-selected cases in the active dataset, exactly M cases are selected.

  2. If N is greater than the number of currently-selected cases in the active dataset, an equivalent proportion of cases are selected.

  3. If N is less than the number of currently-selected cases in the active, exactly M cases are selected from the first N cases in the active dataset.

SAMPLE and SELECT IF are performed in the order specified by the syntax file.

SAMPLE is always performed before N OF CASES, regardless of ordering in the syntax file.

The same values for SAMPLE may result in different samples. To obtain the same sample, use the SET command to set the random number seed to the same value before each SAMPLE. Different samples may still result when the file is processed on systems with different machine types or PSPP versions. By default, the random number seed is based on the system time.

SELECT IF

SELECT IF EXPRESSION.

SELECT IF selects cases for analysis based on the value of EXPRESSION. Cases not selected are permanently eliminated from the active dataset, unless TEMPORARY is in effect.

Specify a boolean expression. If expression](../language/expressions/index.md#boolean-values). If the expression is true for a particular case, the case is analyzed. If the expression is false or missing, then the case is deleted from the data stream.

Place SELECT IF early in the command file. Cases that are deleted early can be processed more efficiently in time and space. Once cases have been deleted from the active dataset using SELECT IF they cannot be re-instated. If you want to be able to re-instate cases, then use FILTER instead.

When SELECT IF is specified following TEMPORARY, the LAG the LAG function may not be used.

Example

A shop steward is interested in the salaries of younger personnel in a firm. The file personnel.sav provides the salaries of all the workers and their dates of birth. The syntax below shows how SELECT IF can be used to limit analysis only to those persons born after December 31, 1999.

get file = 'personnel.sav'.

echo 'Salaries of all personnel'.
descriptives salary.

echo 'Salaries of personnel born after December 31 1999'.
select if dob > date.dmy (31,12,1999).
descriptives salary.

From the output shown below, one can see that there are 56 persons listed in the dataset, and 17 of them were born after December 31, 1999.

Salaries of all personnel

               Descriptive Statistics
┌────────────────────────┬──┬────────┬───────┬───────┬───────┐
│                        │ N│  Mean  │Std Dev│Minimum│Maximum│
├────────────────────────┼──┼────────┼───────┼───────┼───────┤
│Annual salary before tax│56│40028.97│8721.17│$23,451│$57,044│
│Valid N (listwise)      │56│        │       │       │       │
│Missing N (listwise)    │ 0│        │       │       │       │
└────────────────────────┴──┴────────┴───────┴───────┴───────┘

Salaries of personnel born after December 31 1999

               Descriptive Statistics
┌────────────────────────┬──┬────────┬───────┬───────┬───────┐
│                        │ N│  Mean  │Std Dev│Minimum│Maximum│
├────────────────────────┼──┼────────┼───────┼───────┼───────┤
│Annual salary before tax│17│31828.59│4454.80│$23,451│$39,504│
│Valid N (listwise)      │17│        │       │       │       │
│Missing N (listwise)    │ 0│        │       │       │       │
└────────────────────────┴──┴────────┴───────┴───────┴───────┘

Note that the personnel.sav file from which the data were read is unaffected. The transformation affects only the active file.

SPLIT FILE

SPLIT FILE [{LAYERED, SEPARATE}] BY VAR_LIST.
SPLIT FILE OFF.

SPLIT FILE allows multiple sets of data present in one data file to be analyzed separately using single statistical procedure commands.

Specify a list of variable names to analyze multiple sets of data separately. Groups of adjacent cases having the same values for these variables are analyzed by statistical procedure commands as one group. An independent analysis is carried out for each group of cases, and the variable values for the group are printed along with the analysis.

When a list of variable names is specified, one of the keywords LAYERED or SEPARATE may also be specified. With LAYERED, which is the default, the separate analyses for each group are presented together in a single table. With SEPARATE, each analysis is presented in a separate table. Not all procedures honor the distinction.

Groups are formed only by adjacent cases. To create a split using a variable where like values are not adjacent in the working file, first sort the data by that variable.

Specify OFF to disable SPLIT FILE and resume analysis of the entire active dataset as a single group of data.

When SPLIT FILE is specified after TEMPORARY, it affects only the next procedure.

Example

The file horticulture.sav contains data describing the yield of a number of horticultural specimens which have been subjected to various treatments. If we wanted to investigate linear statistics of the yeild, one way to do this is using DESCRIPTIVES. However, it is reasonable to expect the mean to be different depending on the treatment. So we might want to perform three separate procedures -- one for each treatment.1 The following syntax shows how this can be done automatically using the SPLIT FILE command.

get file='horticulture.sav'.

* Ensure cases are sorted before splitting.
sort cases by treatment.

split file by treatment.

* Run descriptives on the yield variable
descriptives /variable = yield.

In the following output, you can see that the table of descriptive statistics appears 3 times—once for each value of treatment. In this example N, the number of observations are identical in all splits. This is because that experiment was deliberately designed that way. However in general one can expect a different N for each split.

    Split Values
┌─────────┬───────┐
│Variable │ Value │
├─────────┼───────┤
│treatment│control│
└─────────┴───────┘

            Descriptive Statistics
┌────────────────────┬──┬─────┬───────┬───────┬───────┐
│                    │ N│ Mean│Std Dev│Minimum│Maximum│
├────────────────────┼──┼─────┼───────┼───────┼───────┤
│yield               │30│51.23│   8.28│  37.86│  68.59│
│Valid N (listwise)  │30│     │       │       │       │
│Missing N (listwise)│ 0│     │       │       │       │
└────────────────────┴──┴─────┴───────┴───────┴───────┘

 Split Values
┌─────────┬────────────┐
│Variable │    Value   │
├─────────┼────────────┤
│treatment│conventional│
└─────────┴────────────┘

            Descriptive Statistics
┌────────────────────┬──┬─────┬───────┬───────┬───────┐
│                    │ N│ Mean│Std Dev│Minimum│Maximum│
├────────────────────┼──┼─────┼───────┼───────┼───────┤
│yield               │30│53.57│   8.92│  36.30│  70.66│
│Valid N (listwise)  │30│     │       │       │       │
│Missing N (listwise)│ 0│     │       │       │       │
└────────────────────┴──┴─────┴───────┴───────┴───────┘

 Split Values
┌─────────┬───────────┐
│Variable │   Value   │
├─────────┼───────────┤
│treatment│traditional│
└─────────┴───────────┘

            Descriptive Statistics
┌────────────────────┬──┬─────┬───────┬───────┬───────┐
│                    │ N│ Mean│Std Dev│Minimum│Maximum│
├────────────────────┼──┼─────┼───────┼───────┼───────┤
│yield               │30│56.87│   8.88│  39.08│  75.93│
│Valid N (listwise)  │30│     │       │       │       │
│Missing N (listwise)│ 0│     │       │       │       │
└────────────────────┴──┴─────┴───────┴───────┴───────┘

Example 13.3: The results of running DESCRIPTIVES with an active split

Unless TEMPORARY was used, after a split has been defined for a dataset it remains active until explicitly disabled.

  1. There are other, possibly better, ways to achieve a similar result using the MEANS or EXAMINE commands. ↩

TEMPORARY

TEMPORARY.

TEMPORARY is used to make the effects of transformations following its execution temporary. These transformations affect only the execution of the next procedure or procedure-like command. Their effects are not be saved to the active dataset.

The only specification on TEMPORARY is the command name.

TEMPORARY may not appear within a DO IF or LOOP construct. It may appear only once between procedures and procedure-like commands.

Scratch variables cannot be used following TEMPORARY.

Example

In the syntax below, there are two COMPUTE transformation. One of them immediately follows a TEMPORARY command, and therefore affects only the next procedure, which in this case is the first DESCRIPTIVES command.

data list notable /x 1-2.
begin data.
 2
 4
10
15
20
24
end data.

compute x=x/2.

temporary.
compute x=x+3.

descriptives x.
descriptives x.

The data read by the first DESCRIPTIVES procedure are 4, 5, 8, 10.5, 13, 15. The data read by the second DESCRIPTIVES procedure are 1, 2, 5, 7.5, 10, 12. This is because the second COMPUTE transformation has no effect on the second DESCRIPTIVES procedure. You can check these figures in the following output.

                Descriptive Statistics
┌────────────────────┬─┬────┬───────┬───────┬───────┐
│                    │N│Mean│Std Dev│Minimum│Maximum│
├────────────────────┼─┼────┼───────┼───────┼───────┤
│x                   │6│9.25│   4.38│      4│     15│
│Valid N (listwise)  │6│    │       │       │       │
│Missing N (listwise)│0│    │       │       │       │
└────────────────────┴─┴────┴───────┴───────┴───────┘

           Descriptive Statistics
┌────────────────────┬─┬────┬───────┬───────┬───────┐
│                    │N│Mean│Std Dev│Minimum│Maximum│
├────────────────────┼─┼────┼───────┼───────┼───────┤
│x                   │6│6.25│   4.38│      1│     12│
│Valid N (listwise)  │6│    │       │       │       │
│Missing N (listwise)│0│    │       │       │       │
└────────────────────┴─┴────┴───────┴───────┴───────┘

WEIGHT

WEIGHT BY VAR_NAME.
WEIGHT OFF.

WEIGHT assigns cases varying weights, changing the frequency distribution of the active dataset. Execution of WEIGHT is delayed until data have been read.

If a variable name is specified, WEIGHT causes the values of that variable to be used as weighting factors for subsequent statistical procedures. Use of keyword BY is optional but recommended. Weighting variables must be numeric. Scratch variables may not be variables](../language/datasets/scratch-variables.md) may not be used for weighting.

When OFF is specified, subsequent statistical procedures weight all cases equally.

A positive integer weighting factor W on a case yields the same statistical output as would replicating the case W times. A weighting factor of 0 is treated for statistical purposes as if the case did not exist in the input. Weighting values need not be integers, but negative and system-missing values for the weighting variable are interpreted as weighting factors of 0. User-missing values are not treated specially.

When WEIGHT is specified after TEMPORARY, it affects only the next procedure.

WEIGHT does not cause cases in the active dataset to be replicated in memory.

Example

One could define a dataset containing an inventory of stock items. It would be reasonable to use a string variable for a description of the item, and a numeric variable for the number in stock, like in the syntax below.

data list notable list /item (a16) quantity (f8.0).
begin   data
nuts    345
screws  10034
washers 32012
bolts   876
end data.

echo 'Unweighted frequency table'.
frequencies /variables = item /format=dfreq.

weight by quantity.

echo 'Weighted frequency table'.
frequencies /variables = item /format=dfreq.

One analysis which most surely would be of interest is the relative amounts or each item in stock. However without setting a weight variable, FREQUENCIES does not tell us what we want to know, since there is only one case for each stock item. The output below shows the difference between the weighted and unweighted frequency tables.

Unweighted frequency table

                          item
┌─────────────┬─────────┬───────┬─────────────┬──────────────────┐
│             │Frequency│Percent│Valid Percent│Cumulative Percent│
├─────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Valid bolts  │        1│  25.0%│        25.0%│             25.0%│
│      nuts   │        1│  25.0%│        25.0%│             50.0%│
│      screws │        1│  25.0%│        25.0%│             75.0%│
│      washers│        1│  25.0%│        25.0%│            100.0%│
├─────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Total        │        4│ 100.0%│             │                  │
└─────────────┴─────────┴───────┴─────────────┴──────────────────┘

Weighted frequency table

                          item
┌─────────────┬─────────┬───────┬─────────────┬──────────────────┐
│             │Frequency│Percent│Valid Percent│Cumulative Percent│
├─────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Valid washers│    32012│  74.0%│        74.0%│             74.0%│
│      screws │    10034│  23.2%│        23.2%│             97.2%│
│      bolts  │      876│   2.0%│         2.0%│             99.2%│
│      nuts   │      345│    .8%│          .8%│            100.0%│
├─────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Total        │    43267│ 100.0%│             │                  │
└─────────────┴─────────┴───────┴─────────────┴──────────────────┘

Conditionals and Loops

This chapter documents PSPP commands used for conditional execution, looping, and flow of control.

BREAK

BREAK.

BREAK terminates execution of the innermost currently executing LOOP construct.

BREAK is allowed only inside LOOP...END LOOP.

DEFINE…!ENDDEFINE

Overview

DEFINE macro_name([argument[/argument]...])
...body...
!ENDDEFINE.

Each argument takes the following form:

{!arg_name= | !POSITIONAL}
[!DEFAULT(default)]
[!NOEXPAND]
{!TOKENS(count) | !CHAREND('token') | !ENCLOSE('start' | 'end') | !CMDEND}

The following directives may be used within body:

!OFFEXPAND
!ONEXPAND

The following functions may be used within the body:

!BLANKS(count)
!CONCAT(arg...)
!EVAL(arg)
!HEAD(arg)
!INDEX(haystack, needle)
!LENGTH(arg)
!NULL
!QUOTE(arg)
!SUBSTR(arg, start[, count])
!TAIL(arg)
!UNQUOTE(arg)
!UPCASE(arg)

The body may also include the following constructs:

!IF (condition) !THEN true-expansion !ENDIF
!IF (condition) !THEN true-expansion !ELSE false-expansion !ENDIF

!DO !var = start !TO end [!BY step]
  body
!DOEND
!DO !var !IN (expression)
  body
!DOEND

!LET !var = expression

Introduction

The DEFINE command creates a "macro", which is a name for a fragment of PSPP syntax called the macro's "body". Following the DEFINE command, syntax may "call" the macro by name any number of times. Each call substitutes, or "expands", the macro's body in place of the call, as if the body had been written in its place.

The following syntax defines a macro named !vars that expands to the variable names v1 v2 v3. The macro's name begins with !, which is optional for macro names. The () following the macro name are required:

DEFINE !vars()
v1 v2 v3
!ENDDEFINE.

Here are two ways that !vars might be called given the preceding definition:

DESCRIPTIVES !vars.
FREQUENCIES /VARIABLES=!vars.

With macro expansion, the above calls are equivalent to the following:

DESCRIPTIVES v1 v2 v3.
FREQUENCIES /VARIABLES=v1 v2 v3.

The !vars macro expands to a fixed body. Macros may have more sophisticated contents:

  • Macro "arguments" that are substituted into the body whenever they are named. The values of a macro's arguments are specified each time it is called.

  • Macro "functions", expanded when the macro is called.

  • !IF constructs, for conditional expansion.

  • Two forms of !DO construct, for looping over a numerical range or a collection of tokens.

  • !LET constructs, for assigning to macro variables.

Many identifiers associated with macros begin with !, a character not normally allowed in identifiers. These identifiers are reserved only for use with macros, which helps keep them from being confused with other kinds of identifiers.

The following sections provide more details on macro syntax and semantics.

Macro Bodies

As previously shown, a macro body may contain a fragment of a PSPP command (such as a variable name). A macro body may also contain full PSPP commands. In the latter case, the macro body should also contain the command terminators.

Most PSPP commands may occur within a macro. The DEFINE command itself is one exception, because the inner !ENDDEFINE ends the outer macro definition. For compatibility, BEGIN DATA...END DATA. should not be used within a macro.

The body of a macro may call another macro. The following shows one way that could work:

DEFINE !commands()
DESCRIPTIVES !vars.
FREQUENCIES /VARIABLES=!vars.
!ENDDEFINE.

* Initially define the 'vars' macro to analyze v1...v3.
DEFINE !vars() v1 v2 v3 !ENDDEFINE.
!commands

* Redefine 'vars' macro to analyze different variables.
DEFINE !vars() v4 v5 !ENDDEFINE.
!commands

The !commands macro would be easier to use if it took the variables to analyze as an argument rather than through another macro. The following section shows how to do that.

Macro Arguments

This section explains how to use macro arguments. As an initial example, the following syntax defines a macro named !analyze that takes all the syntax up to the first command terminator as an argument:

DEFINE !analyze(!POSITIONAL !CMDEND)
DESCRIPTIVES !1.
FREQUENCIES /VARIABLES=!1.
!ENDDEFINE.

When !analyze is called, it expands to a pair of analysis commands with each !1 in the body replaced by the argument. That is, these calls:

!analyze v1 v2 v3.
!analyze v4 v5.

act like the following:

DESCRIPTIVES v1 v2 v3.
FREQUENCIES /VARIABLES=v1 v2 v3.
DESCRIPTIVES v4 v5.
FREQUENCIES /VARIABLES=v4 v5.

Macros may take any number of arguments, described within the parentheses in the DEFINE command. Arguments come in two varieties based on how their values are specified when the macro is called:

  • A "positional" argument has a required value that follows the macro's name. Use the !POSITIONAL keyword to declare a positional argument.

    When a macro is called, the positional argument values appear in the same order as their definitions, before any keyword argument values.

    References to a positional argument in a macro body are numbered: !1 is the first positional argument, !2 the second, and so on. In addition, !* expands to all of the positional arguments' values, separated by spaces.

    The following example uses a positional argument:

    DEFINE !analyze(!POSITIONAL !CMDEND)
DESCRIPTIVES !1.
FREQUENCIES /VARIABLES=!1.
!ENDDEFINE.

!analyze v1 v2 v3.
!analyze v4 v5.

  • A "keyword" argument has a name. In the macro call, its value is specified with the syntax name=value. The names allow keyword argument values to take any order in the call.

    In declaration and calls, a keyword argument's name may not begin with !, but references to it in the macro body do start with a leading !.

    The following example uses a keyword argument that defaults to ALL if the argument is not assigned a value:

    DEFINE !analyze_kw(vars=!DEFAULT(ALL) !CMDEND)
DESCRIPTIVES !vars.
FREQUENCIES /VARIABLES=!vars.
!ENDDEFINE.

!analyze_kw vars=v1 v2 v3.  /* Analyze specified variables.
!analyze_kw.                /* Analyze all variables.

If a macro has both positional and keyword arguments, then the positional arguments must come first in the DEFINE command, and their values also come first in macro calls. A keyword argument may be omitted by leaving its keyword out of the call, and a positional argument may be omitted by putting a command terminator where it would appear. (The latter case also omits any following positional arguments and all keyword arguments, if there are any.) When an argument is omitted, a default value is used: either the value specified in !DEFAULT(value), or an empty value otherwise.

Each argument declaration specifies the form of its value:

  • !TOKENS(count)
    Exactly count tokens, e.g. !TOKENS(1) for a single token. Each identifier, number, quoted string, operator, or punctuator is a token (see Tokens for details).

    The following variant of !analyze_kw accepts only a single variable name (or ALL) as its argument:

    DEFINE !analyze_one_var(!POSITIONAL !TOKENS(1))
DESCRIPTIVES !1.
FREQUENCIES /VARIABLES=!1.
!ENDDEFINE.

!analyze_one_var v1.

  • !CHAREND('TOKEN')
    Any number of tokens up to TOKEN, which should be an operator or punctuator token such as / or +. The TOKEN does not become part of the value.

    With the following variant of !analyze_kw, the variables must be following by /:

    DEFINE !analyze_parens(vars=!CHARNED('/'))
DESCRIPTIVES !vars.
FREQUENCIES /VARIABLES=!vars.
!ENDDEFINE.

!analyze_parens vars=v1 v2 v3/.

  • !ENCLOSE('START','END')
    Any number of tokens enclosed between START and END, which should each be operator or punctuator tokens. For example, use !ENCLOSE('(',')') for a value enclosed within parentheses. (Such a value could never have right parentheses inside it, even paired with left parentheses.) The start and end tokens are not part of the value.

    With the following variant of !analyze_kw, the variables must be specified within parentheses:

    DEFINE !analyze_parens(vars=!ENCLOSE('(',')'))
DESCRIPTIVES !vars.
FREQUENCIES /VARIABLES=!vars.
!ENDDEFINE.

!analyze_parens vars=(v1 v2 v3).

  • !CMDEND
    Any number of tokens up to the end of the command. This should be used only for the last positional parameter, since it consumes all of the tokens in the command calling the macro.

    The following variant of !analyze_kw takes all the variable names up to the end of the command as its argument:

    DEFINE !analyze_kw(vars=!CMDEND)
DESCRIPTIVES !vars.
FREQUENCIES /VARIABLES=!vars.
!ENDDEFINE.

!analyze_kw vars=v1 v2 v3.

By default, when an argument's value contains a macro call, the call is expanded each time the argument appears in the macro's body. The !NOEXPAND keyword in an argument declaration suppresses this expansion.

Controlling Macro Expansion

Multiple factors control whether macro calls are expanded in different situations. At the highest level, SET MEXPAND controls whether macro calls are expanded. By default, it is enabled. SET MEXPAND, for details.

A macro body may contain macro calls. By default, these are expanded. If a macro body contains !OFFEXPAND or !ONEXPAND directives, then !OFFEXPAND disables expansion of macro calls until the following !ONEXPAND.

A macro argument's value may contain a macro call. These macro calls are expanded, unless the argument was declared with the !NOEXPAND keyword.

The argument to a macro function is a special context that does not expand macro calls. For example, if !vars is the name of a macro, then !LENGTH(!vars) expands to 5, as does !LENGTH(!1) if positional argument 1 has value !vars. To expand macros in these cases, use the !EVAL macro function, e.g. !LENGTH(!EVAL(!vars)) or !LENGTH(!EVAL(!1)).

These rules apply to macro calls, not to uses within a macro body of macro functions, macro arguments, and macro variables created by !DO or !LET, which are always expanded.

SET MEXPAND may appear within the body of a macro, but it will not affect expansion of the macro that it appears in. Use !OFFEXPAND and !ONEXPAND instead.

Macro Functions

Macro bodies may manipulate syntax using macro functions. Macro functions accept tokens as arguments and expand to sequences of characters.

The arguments to macro functions have a restricted form. They may only be a single token (such as an identifier or a string), a macro argument, or a call to a macro function. Thus, the following are valid macro arguments:

  • x
  • 5.0
  • x
  • !1
  • "5 + 6"
  • !CONCAT(x,y)

and the following are not (because they are each multiple tokens):

  • x y
  • 5+6

Macro functions expand to sequences of characters. When these character strings are processed further as character strings, e.g. with !LENGTH, any character string is valid. When they are interpreted as PSPP syntax, e.g. when the expansion becomes part of a command, they need to be valid for that purpose. For example, !UNQUOTE("It's") will yield an error if the expansion It's becomes part of a PSPP command, because it contains unbalanced single quotes, but !LENGTH(!UNQUOTE("It's")) expands to 4.

The following macro functions are available.

  • !BLANKS(count)
    Expands to COUNT unquoted spaces, where COUNT is a nonnegative integer. Outside quotes, any positive number of spaces are equivalent; for a quoted string of spaces, use !QUOTE(!BLANKS(COUNT)).

    In the examples below, _ stands in for a space to make the results visible.

    !BLANKS(0)                  ↦ empty
!BLANKS(1)                  ↦ _
!BLANKS(2)                  ↦ __
!QUOTE(!BLANKS(5))          ↦ '_____'
    CallExpansion
    !BLANKS(0)(empty)
    !BLANKS(1)_
    !BLANKS(2)__
    `!QUOTE(!BLANKS(5))'_____'

  • !CONCAT(arg...)
    Expands to the concatenation of all of the arguments. Before concatenation, each quoted string argument is unquoted, as if !UNQUOTE were applied. This allows for "token pasting", combining two (or more) tokens into a single one:

    CallExpansion
    !CONCAT(x, y)xy
    !CONCAT('x', 'y')xy
    !CONCAT(12, 34)1234
    !CONCAT(!NULL, 123)123

    !CONCAT is often used for constructing a series of similar variable names from a prefix followed by a number and perhaps a suffix. For example:

    CallExpansion
    !CONCAT(x, 0)x0
    !CONCAT(x, 0, y)x0y

    An identifier token must begin with a letter (or # or @), which means that attempting to use a number as the first part of an identifier will produce a pair of distinct tokens rather than a single one. For example:

    CallExpansion
    !CONCAT(0, x)0 x
    !CONCAT(0, x, y)0 xy

  • !EVAL(arg)
    Expands macro calls in ARG. This is especially useful if ARG is the name of a macro or a macro argument that expands to one, because arguments to macro functions are not expanded by default (see Controlling Macro Expansion).

    The following examples assume that !vars is a macro that expands to a b c:

    CallExpansion
    !varsa b c
    !QUOTE(!vars)'!vars'
    !EVAL(!vars)a b c
    !QUOTE(!EVAL(!vars))'a b c'

    These examples additionally assume that argument !1 has value !vars:

    CallExpansion
    !1a b c
    !QUOTE(!1)'!vars'
    !EVAL(!1)a b c
    !QUOTE(!EVAL(!1))'a b c'

  • !HEAD(arg)
    !TAIL(arg)
    !HEAD expands to just the first token in an unquoted version of ARG, and !TAIL to all the tokens after the first.

    CallExpansion
    !HEAD('a b c')a
    !HEAD('a')a
    !HEAD(!NULL)(empty)
    !HEAD('')(empty)
    !TAIL('a b c')b c
    !TAIL('a')(empty)
    !TAIL(!NULL)(empty)
    !TAIL('')(empty)

  • !INDEX(haystack, needle)
    Looks for NEEDLE in HAYSTACK. If it is present, expands to the 1-based index of its first occurrence; if not, expands to 0.

    CallExpansion
    !INDEX(banana, an)2
    !INDEX(banana, nan)3
    !INDEX(banana, apple)0
    !INDEX("banana", nan)4
    !INDEX("banana", "nan")0
    !INDEX(!UNQUOTE("banana"), !UNQUOTE("nan"))3

  • !LENGTH(arg)
    Expands to a number token representing the number of characters in ARG.

    CallExpansion
    !LENGTH(123)3
    !LENGTH(123.00)6
    !LENGTH( 123 )3
    !LENGTH("123")5
    !LENGTH(xyzzy)5
    !LENGTH("xyzzy")7
    !LENGTH("xy""zzy")9
    !LENGTH(!UNQUOTE("xyzzy"))5
    !LENGTH(!UNQUOTE("xy""zzy"))6
    !LENGTH(!1)5 (if !1 is a b c)
    !LENGTH(!1)0 (if !1 is empty)
    !LENGTH(!NULL)0

  • !NULL
    Expands to an empty character sequence.

    CallExpansion
    !NULL(empty)
    !QUOTE(!NULL)''

  • !QUOTE(arg)
    !UNQUOTE(arg)
    The !QUOTE function expands to its argument surrounded by apostrophes, doubling any apostrophes inside the argument to make sure that it is valid PSPP syntax for a string. If the argument was already a quoted string, !QUOTE expands to it unchanged.

    Given a quoted string argument, the !UNQUOTED function expands to the string's contents, with the quotes removed and any doubled quote marks reduced to singletons. If the argument was not a quoted string, !UNQUOTE expands to the argument unchanged.

    CallExpansion
    !QUOTE(123.0)'123.0'
    !QUOTE( 123 )'123'
    !QUOTE('a b c')'a b c'
    !QUOTE("a b c")"a b c"
    !QUOTE(!1)'a ''b'' c' (if !1 is a 'b' c)
    !UNQUOTE(123.0)123.0
    !UNQUOTE( 123 )123
    !UNQUOTE('a b c')a b c
    !UNQUOTE("a b c")a b c
    !UNQUOTE(!1)a 'b' c (if !1 is a 'b' c)
    !QUOTE(!UNQUOTE(123.0))'123.0'
    !QUOTE(!UNQUOTE( 123 ))'123'
    !QUOTE(!UNQUOTE('a b c'))'a b c'
    !QUOTE(!UNQUOTE("a b c"))'a b c'
    !QUOTE(!UNQUOTE(!1))'a ''b'' c' (if !1 is a 'b' c)

  • !SUBSTR(arg, start[, count])
    Expands to a substring of ARG starting from 1-based position START. If COUNT is given, it limits the number of characters in the expansion; if it is omitted, then the expansion extends to the end of ARG.

    |Call|Expansion|
|:-----|:--------|
|`!SUBSTR(banana, 3)`|`nana`|
|`!SUBSTR(banana, 3, 3)`|`nan`|
|`!SUBSTR("banana", 1, 3)`|error (`"ba` is not a valid token)|
|`!SUBSTR(!UNQUOTE("banana"), 3)`|`nana`|
|`!SUBSTR("banana", 3, 3)`|`ana`|
|`!SUBSTR(banana, 3, 0)`|(empty)|
|`!SUBSTR(banana, 3, 10)`|`nana`|
|`!SUBSTR(banana, 10, 3)`|(empty)|

  • !UPCASE(arg)
    Expands to an unquoted version of ARG with all letters converted to uppercase.

    CallExpansion
    !UPCASE(freckle)FRECKLE
    !UPCASE('freckle')FRECKLE
    !UPCASE('a b c')A B C
    !UPCASE('A B C')A B C

Macro Expressions

Macro expressions are used in conditional expansion and loops, which are described in the following sections. A macro expression may use the following operators, listed in descending order of operator precedence:

  • ()
    Parentheses override the default operator precedence.

  • !EQ !NE !GT !LT !GE !LE = ~= <> > < >= <=
    Relational operators compare their operands and yield a Boolean result, either 0 for false or 1 for true.

    These operators always compare their operands as strings. This can be surprising when the strings are numbers because, e.g., 1 < 1.0 and 10 < 2 both evaluate to 1 (true).

    Comparisons are case sensitive, so that a = A evaluates to 0 (false).

  • !NOT ~
    !AND &
    !OR |
    Logical operators interpret their operands as Boolean values, where quoted or unquoted 0 is false and anything else is true, and yield a Boolean result, either 0 for false or 1 for true.

Macro expressions do not include any arithmetic operators.

An operand in an expression may be a single token (including a macro argument name) or a macro function invocation. Either way, the expression evaluator unquotes the operand, so that 1 = '1' is true.

Macro Conditional Expansion

The !IF construct may be used inside a macro body to allow for conditional expansion. It takes the following forms:

!IF (EXPRESSION) !THEN TRUE-EXPANSION !IFEND
!IF (EXPRESSION) !THEN TRUE-EXPANSION !ELSE FALSE-EXPANSION !IFEND

When EXPRESSION evaluates to true, the macro processor expands TRUE-EXPANSION; otherwise, it expands FALSE-EXPANSION, if it is present. The macro processor considers quoted or unquoted 0 to be false, and anything else to be true.

Macro Loops

The body of a macro may include two forms of loops: loops over numerical ranges and loops over tokens. Both forms expand a "loop body" multiple times, each time setting a named "loop variable" to a different value. The loop body typically expands the loop variable at least once.

The MITERATE setting limits the number of iterations in a loop. This is a safety measure to ensure that macro expansion terminates. PSPP issues a warning when the MITERATE limit is exceeded.

Loops Over Ranges

!DO !VAR = START !TO END [!BY STEP]
  BODY
!DOEND

A loop over a numerical range has the form shown above. START, END, and STEP (if included) must be expressions with numeric values. The macro processor accepts both integers and real numbers. The macro processor expands BODY for each numeric value from START to END, inclusive.

The default value for STEP is 1. If STEP is positive and FIRST > LAST, or if STEP is negative and FIRST < LAST, then the macro processor doesn't expand the body at all. STEP may not be zero.

Loops Over Tokens

!DO !VAR !IN (EXPRESSION)
  BODY
!DOEND

A loop over tokens takes the form shown above. The macro processor evaluates EXPRESSION and expands BODY once per token in the result, substituting the token for !VAR each time it appears.

Macro Variable Assignment

The !LET construct evaluates an expression and assigns the result to a macro variable. It may create a new macro variable or change the value of one created by a previous !LET or !DO, but it may not change the value of a macro argument. !LET has the following form:

!LET !VAR = EXPRESSION

If EXPRESSION is more than one token, it must be enclosed in parentheses.

Macro Settings

the SET command controls some macro behavior. This section describes these settings.

Any SET command that changes these settings within a macro body only takes effect following the macro. This is because PSPP expands a macro's entire body at once, so that SET inside the body only executes afterwards.

The MEXPAND setting controls whether macros will be expanded at all. By default, macro expansion is on. To avoid expansion of macros called within a macro body, use !OFFEXPAND and !ONEXPAND.

When MPRINT is turned on, PSPP outputs an expansion of each macro called. This feature can be useful for debugging macro definitions. For reading the expanded version, keep in mind that macro expansion removes comments and standardizes white space.

MNEST limits the depth of expansion of macro calls, that is, the nesting level of macro expansion. The default is 50. This is mainly useful to avoid infinite expansion in the case of a macro that calls itself.

MITERATE limits the number of iterations in a !DO construct. The default is 1000.

Additional Notes

Calling Macros from Macros

If the body of macro A includes a call to macro B, the call can use macro arguments (including !*) and macro variables as part of arguments to B. For !TOKENS arguments, the argument or variable name counts as one token regardless of the number that it expands into; for !CHAREND and !ENCLOSE arguments, the delimiters come only from the call, not the expansions; and !CMDEND ends at the calling command, not any end of command within an argument or variable.

Macro functions are not supported as part of the arguments in a macro call. To get the same effect, use !LET to define a macro variable, then pass the macro variable to the macro.

When macro A calls macro B, the order of their DEFINE commands doesn't matter, as long as macro B has been defined when A is called.

Command Terminators

Macros and command terminators require care. Macros honor the syntax differences between interactive and batch syntax, which means that the interpretation of a macro can vary depending on the syntax mode in use. We assume here that interactive mode is in use, in which . at the end of a line is the primary way to end a command.

The DEFINE command needs to end with . following the !ENDDEFINE. The macro body may contain . if it is intended to expand to whole commands, but using . within a macro body that expands to just syntax fragments (such as a list of variables) will cause syntax errors.

Macro directives such as !IF and !DO do not end with ..

Expansion Contexts

PSPP does not expand macros within comments, whether introduced within a line by /* or as a separate COMMENT or * command. (SPSS does expand macros in COMMENT and *.)

Macros do not expand within quoted strings.

Macros are expanded in the TITLE and SUBTITLE commands as long as their arguments are not quoted strings.

PRESERVE and RESTORE

Some macro bodies might use the SET command to change certain settings. When this is the case, consider using the PRESERVE and RESTORE commands to save and then restore these settings.

DO IF…END IF

DO IF condition.
        ...
[ELSE IF condition.
        ...
]...
[ELSE.
        ...]
END IF.

DO IF allows one of several sets of transformations to be executed, depending on user-specified conditions.

If the specified boolean expression evaluates as true, then the block of code following DO IF is executed. If it evaluates as missing, then none of the code blocks is executed. If it is false, then the boolean expression on the first ELSE IF, if present, is tested in turn, with the same rules applied. If all expressions evaluate to false, then the ELSE code block is executed, if it is present.

When DO IF or ELSE IF is specified following TEMPORARY, the LAG function may not be used.

DO REPEAT…END REPEAT

DO REPEAT dummy_name=expansion....
        ...
END REPEAT [PRINT].

expansion takes one of the following forms:
        var_list
        num_or_range...
        'string'...
        ALL

num_or_range takes one of the following forms:
        number
        num1 TO num2

DO REPEAT repeats a block of code, textually substituting different variables, numbers, or strings into the block with each repetition.

Specify a dummy variable name followed by an equals sign (=) and the list of replacements. Replacements can be a list of existing or new variables, numbers, strings, or ALL to specify all existing variables. When numbers are specified, runs of increasing integers may be indicated as NUM1 TO NUM2, so that 1 TO 5 is short for 1 2 3 4 5.

Multiple dummy variables can be specified. Each variable must have the same number of replacements.

The code within DO REPEAT is repeated as many times as there are replacements for each variable. The first time, the first value for each dummy variable is substituted; the second time, the second value for each dummy variable is substituted; and so on.

Dummy variable substitutions work like macros. They take place anywhere in a line that the dummy variable name occurs. This includes command and subcommand names, so command and subcommand names that appear in the code block should not be used as dummy variable identifiers. Dummy variable substitutions do not occur inside quoted strings, comments, unquoted strings (such as the text on the TITLE or DOCUMENT command), or inside BEGIN DATA...END DATA.

Substitution occurs only on whole words, so that, for example, a dummy variable PRINT would not be substituted into the word PRINTOUT.

New variable names used as replacements are not automatically created as variables, but only if used in the code block in a context that would create them, e.g. on a NUMERIC or STRING command or on the left side of a COMPUTE assignment.

Any command may appear within DO REPEAT, including nested DO REPEAT commands. If INCLUDE or INSERT appears within DO REPEAT, the substitutions do not apply to the included file.

If PRINT is specified on END REPEAT, the commands after substitutions are made should be printed to the listing file, prefixed by a plus sign (+). This feature is not yet implemented.

LOOP…END LOOP

LOOP [INDEX_VAR=START TO END [BY INCR]] [IF CONDITION].
        ...
END LOOP [IF CONDITION].

LOOP iterates a group of commands. A number of termination options are offered.

Specify INDEX_VAR to make that variable count from one value to another by a particular increment. INDEX_VAR must be a pre-existing numeric variable. START, END, and INCR are numeric expressions. expressions.

During the first iteration, INDEX_VAR is set to the value of START. During each successive iteration, INDEX_VAR is increased by the value of INCR. If END > START, then the loop terminates when INDEX_VAR > END; otherwise it terminates when INDEX_VAR < END. If INCR is not specified then it defaults to +1 or -1 as appropriate.

If END > START and INCR < 0, or if END < START and INCR > 0, then the loop is never executed. INDEX_VAR is nevertheless set to the value of start.

Modifying INDEX_VAR within the loop is allowed, but it has no effect on the value of INDEX_VAR in the next iteration.

Specify a boolean expression for the condition on LOOP to cause the loop to be executed only if the condition is true. If the condition is false or missing before the loop contents are executed the first time, the loop contents are not executed at all.

If index and condition clauses are both present on LOOP, the index variable is always set before the condition is evaluated. Thus, a condition that makes use of the index variable will always see the index value to be used in the next execution of the body.

Specify a boolean expression for the condition on END LOOP to cause the loop to terminate if the condition is true after the enclosed code block is executed. The condition is evaluated at the end of the loop, not at the beginning, so that the body of a loop with only a condition on END LOOP will always execute at least once.

If the index clause is not present, then the global MXLOOPS setting, which defaults to 40, limits the number of iterations.

BREAK also terminates LOOP execution.

Loop index variables are by default reset to system-missing from one case to another, not left, unless a scratch variable is used as index. When loops are nested, this is usually undesired behavior, which can be corrected with LEAVE or by using a scratch variable as the loop index.

When LOOP or END LOOP is specified following TEMPORARY, the LAG function may not be used.

Statistics

This chapter documents the statistical procedures that PSPP supports.

#DESCRIPTIVES

DESCRIPTIVES
        /VARIABLES=VAR_LIST
        /MISSING={VARIABLE,LISTWISE} {INCLUDE,NOINCLUDE}
        /FORMAT={LABELS,NOLABELS} {NOINDEX,INDEX} {LINE,SERIAL}
        /SAVE
        /STATISTICS={ALL,MEAN,SEMEAN,STDDEV,VARIANCE,KURTOSIS,
                     SKEWNESS,RANGE,MINIMUM,MAXIMUM,SUM,DEFAULT,
                     SESKEWNESS,SEKURTOSIS}
        /SORT={NONE,MEAN,SEMEAN,STDDEV,VARIANCE,KURTOSIS,SKEWNESS,
               RANGE,MINIMUM,MAXIMUM,SUM,SESKEWNESS,SEKURTOSIS,NAME}
              {A,D}

The DESCRIPTIVES procedure reads the active dataset and outputs linear descriptive statistics requested by the user. It can also compute Z-scores.

The VARIABLES subcommand, which is required, specifies the list of variables to be analyzed. Keyword VARIABLES is optional.

All other subcommands are optional:

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set, then user-missing values are included in the calculations. If NOINCLUDE is set, which is the default, user-missing values are excluded. If VARIABLE is set, then missing values are excluded on a variable by variable basis; if LISTWISE is set, then the entire case is excluded whenever any value in that case has a system-missing or, if INCLUDE is set, user-missing value.

The FORMAT subcommand has no effect. It is accepted for backward compatibility.

The SAVE subcommand causes DESCRIPTIVES to calculate Z scores for all the specified variables. The Z scores are saved to new variables. Variable names are generated by trying first the original variable name with Z prepended and truncated to a maximum of 8 characters, then the names ZSC000 through ZSC999, STDZ00 through STDZ09, ZZZZ00 through ZZZZ09, ZQZQ00 through ZQZQ09, in that order. Z-score variable names may also be specified explicitly on VARIABLES in the variable list by enclosing them in parentheses after each variable. When Z scores are calculated, PSPP ignores TEMPORARY, treating temporary transformations as permanent.

The STATISTICS subcommand specifies the statistics to be displayed:

  • ALL
    All of the statistics below.
  • MEAN
    Arithmetic mean.
  • SEMEAN
    Standard error of the mean.
  • STDDEV
    Standard deviation.
  • VARIANCE
    Variance.
  • KURTOSIS
    Kurtosis and standard error of the kurtosis.
  • SKEWNESS
    Skewness and standard error of the skewness.
  • RANGE
    Range.
  • MINIMUM
    Minimum value.
  • MAXIMUM
    Maximum value.
  • SUM
    Sum.
  • DEFAULT
    Mean, standard deviation of the mean, minimum, maximum.
  • SEKURTOSIS
    Standard error of the kurtosis.
  • SESKEWNESS
    Standard error of the skewness.

The SORT subcommand specifies how the statistics should be sorted. Most of the possible values should be self-explanatory. NAME causes the statistics to be sorted by name. By default, the statistics are listed in the order that they are specified on the VARIABLES subcommand. The A and D settings request an ascending or descending sort order, respectively.

Example

The physiology.sav file contains various physiological data for a sample of persons. Running the DESCRIPTIVES command on the variables height and temperature with the default options allows one to see simple linear statistics for these two variables. In the example below, these variables are specfied on the VARIABLES subcommand and the SAVE option has been used, to request that Z scores be calculated.

After the command completes, this example runs DESCRIPTIVES again, this time on the zheight and ztemperature variables, which are the two normalized (Z-score) variables generated by the first DESCRIPTIVES command.

get file='physiology.sav'.

descriptives
        /variables = height temperature
        /save.

descriptives
        /variables = zheight ztemperature.

In the output below, we can see that there are 40 valid data for each of the variables and no missing values. The mean average of the height and temperature is 16677.12 and 37.02 respectively. The descriptive statistics for temperature seem reasonable. However there is a very high standard deviation for height and a suspiciously low minimum. This is due to a data entry error in the data.

In the second Descriptive Statistics output, one can see that the mean and standard deviation of both Z score variables is 0 and 1 respectively. All Z score statistics should have these properties since they are normalized versions of the original scores.

              Mapping of Variables to Z-scores
┌────────────────────────────────────────────┬────────────┐
│                   Source                   │   Target   │
├────────────────────────────────────────────┼────────────┤
│Height in millimeters                       │Zheight     │
│Internal body temperature in degrees Celcius│Ztemperature│
└────────────────────────────────────────────┴────────────┘

                             Descriptive Statistics
┌──────────────────────────────────────────┬──┬───────┬───────┬───────┬───────┐
│                                          │ N│  Mean │Std Dev│Minimum│Maximum│
├──────────────────────────────────────────┼──┼───────┼───────┼───────┼───────┤
│Height in millimeters                     │40│1677.12│ 262.87│    179│   1903│
│Internal body temperature in degrees      │40│  37.02│   1.82│  32.59│  39.97│
│Celcius                                   │  │       │       │       │       │
│Valid N (listwise)                        │40│       │       │       │       │
│Missing N (listwise)                      │ 0│       │       │       │       │
└──────────────────────────────────────────┴──┴───────┴───────┴───────┴───────┘

                             Descriptive Statistics
┌─────────────────────────────────────────┬──┬─────────┬──────┬───────┬───────┐
│                                         │  │         │  Std │       │       │
│                                         │ N│   Mean  │  Dev │Minimum│Maximum│
├─────────────────────────────────────────┼──┼─────────┼──────┼───────┼───────┤
│Z─score of Height in millimeters         │40│1.93E─015│  1.00│  ─5.70│    .86│
│Z─score of Internal body temperature in  │40│1.37E─015│  1.00│  ─2.44│   1.62│
│degrees Celcius                          │  │         │      │       │       │
│Valid N (listwise)                       │40│         │      │       │       │
│Missing N (listwise)                     │ 0│         │      │       │       │
└─────────────────────────────────────────┴──┴─────────┴──────┴───────┴───────┘

FREQUENCIES

FREQUENCIES
        /VARIABLES=VAR_LIST
        /FORMAT={TABLE,NOTABLE,LIMIT(LIMIT)}
                {AVALUE,DVALUE,AFREQ,DFREQ}
        /MISSING={EXCLUDE,INCLUDE}
        /STATISTICS={DEFAULT,MEAN,SEMEAN,MEDIAN,MODE,STDDEV,VARIANCE,
                     KURTOSIS,SKEWNESS,RANGE,MINIMUM,MAXIMUM,SUM,
                     SESKEWNESS,SEKURTOSIS,ALL,NONE}
        /NTILES=NTILES
        /PERCENTILES=percent...
        /HISTOGRAM=[MINIMUM(X_MIN)] [MAXIMUM(X_MAX)]
                   [{FREQ[(Y_MAX)],PERCENT[(Y_MAX)]}] [{NONORMAL,NORMAL}]
        /PIECHART=[MINIMUM(X_MIN)] [MAXIMUM(X_MAX)]
                  [{FREQ,PERCENT}] [{NOMISSING,MISSING}]
        /BARCHART=[MINIMUM(X_MIN)] [MAXIMUM(X_MAX)]
                  [{FREQ,PERCENT}]
        /ORDER={ANALYSIS,VARIABLE}


(These options are not currently implemented.)
        /HBAR=...
        /GROUPED=...

The FREQUENCIES procedure outputs frequency tables for specified variables. FREQUENCIES can also calculate and display descriptive statistics (including median and mode) and percentiles, and various graphical representations of the frequency distribution.

The VARIABLES subcommand is the only required subcommand. Specify the variables to be analyzed.

The FORMAT subcommand controls the output format. It has several possible settings:

  • TABLE, the default, causes a frequency table to be output for every variable specified. NOTABLE prevents them from being output. LIMIT with a numeric argument causes them to be output except when there are more than the specified number of values in the table.

  • Normally frequency tables are sorted in ascending order by value. This is AVALUE. DVALUE tables are sorted in descending order by value. AFREQ and DFREQ tables are sorted in ascending and descending order, respectively, by frequency count.

The MISSING subcommand controls the handling of user-missing values. When EXCLUDE, the default, is set, user-missing values are not included in frequency tables or statistics. When INCLUDE is set, user-missing are included. System-missing values are never included in statistics, but are listed in frequency tables.

The available STATISTICS are the same as available in DESCRIPTIVES, with the addition of MEDIAN, the data's median value, and MODE, the mode. (If there are multiple modes, the smallest value is reported.) By default, the mean, standard deviation of the mean, minimum, and maximum are reported for each variable.

PERCENTILES causes the specified percentiles to be reported. The percentiles should be presented at a list of numbers between 0 and 100 inclusive. The NTILES subcommand causes the percentiles to be reported at the boundaries of the data set divided into the specified number of ranges. For instance, /NTILES=4 would cause quartiles to be reported.

The HISTOGRAM subcommand causes the output to include a histogram for each specified numeric variable. The X axis by default ranges from the minimum to the maximum value observed in the data, but the MINIMUM and MAXIMUM keywords can set an explicit range.1 Histograms are not created for string variables.

Specify NORMAL to superimpose a normal curve on the histogram.

The PIECHART subcommand adds a pie chart for each variable to the data. Each slice represents one value, with the size of the slice proportional to the value's frequency. By default, all non-missing values are given slices. The MINIMUM and MAXIMUM keywords can be used to limit the displayed slices to a given range of values. The keyword NOMISSING causes missing values to be omitted from the piechart. This is the default. If instead, MISSING is specified, then the pie chart includes a single slice representing all system missing and user-missing cases.

The BARCHART subcommand produces a bar chart for each variable. The MINIMUM and MAXIMUM keywords can be used to omit categories whose counts which lie outside the specified limits. The FREQ option (default) causes the ordinate to display the frequency of each category, whereas the PERCENT option displays relative percentages.

The FREQ and PERCENT options on HISTOGRAM and PIECHART are accepted but not currently honoured.

The ORDER subcommand is accepted but ignored.

Example

The syntax below runs a frequency analysis on the sex and occupation variables from the personnel.sav file. This is useful to get an general idea of the way in which these nominal variables are distributed.

get file='personnel.sav'.

frequencies /variables = sex occupation
            /statistics = none.

If you are using the graphic user interface, the dialog box is set up such that by default, several statistics are calculated. Some are not particularly useful for categorical variables, so you may want to disable those.

From the output, shown below, it is evident that there are 33 males, 21 females and 2 persons for whom their sex has not been entered.

One can also see how many of each occupation there are in the data. When dealing with string variables used as nominal values, running a frequency analysis is useful to detect data input entries. Notice that one occupation value has been mistyped as "Scrientist". This entry should be corrected, or marked as missing before using the data.

                                sex
┌──────────────┬─────────┬───────┬─────────────┬──────────────────┐
│              │Frequency│Percent│Valid Percent│Cumulative Percent│
├──────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Valid   Male  │       33│  58.9%│        61.1%│             61.1%│
│        Female│       21│  37.5%│        38.9%│            100.0%│
├──────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Missing .     │        2│   3.6%│             │                  │
├──────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Total         │       56│ 100.0%│             │                  │
└──────────────┴─────────┴───────┴─────────────┴──────────────────┘

                                  occupation
┌────────────────────────┬─────────┬───────┬─────────────┬──────────────────┐
│                        │Frequency│Percent│Valid Percent│Cumulative Percent│
├────────────────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Valid Artist            │        8│  14.3%│        14.3%│             14.3%│
│      Baker             │        2│   3.6%│         3.6%│             17.9%│
│      Barrister         │        1│   1.8%│         1.8%│             19.6%│
│      Carpenter         │        4│   7.1%│         7.1%│             26.8%│
│      Cleaner           │        4│   7.1%│         7.1%│             33.9%│
│      Cook              │        7│  12.5%│        12.5%│             46.4%│
│      Manager           │        8│  14.3%│        14.3%│             60.7%│
│      Mathematician     │        4│   7.1%│         7.1%│             67.9%│
│      Painter           │        2│   3.6%│         3.6%│             71.4%│
│      Payload Specialist│        1│   1.8%│         1.8%│             73.2%│
│      Plumber           │        5│   8.9%│         8.9%│             82.1%│
│      Scientist         │        7│  12.5%│        12.5%│             94.6%│
│      Scrientist        │        1│   1.8%│         1.8%│             96.4%│
│      Tailor            │        2│   3.6%│         3.6%│            100.0%│
├────────────────────────┼─────────┼───────┼─────────────┼──────────────────┤
│Total                   │       56│ 100.0%│             │                  │
└────────────────────────┴─────────┴───────┴─────────────┴──────────────────┘

  1. The number of bins is chosen according to the Freedman-Diaconis rule: $$2 \times IQR(x)n^{-1/3}$$ where \(IQR(x)\) is the interquartile range of \(x\) and \(n\) is the number of samples. (EXAMINE uses a different algorithm to determine bin sizes.) ↩

#EXAMINE

EXAMINE
        VARIABLES= VAR1 [VAR2] ... [VARN]
           [BY FACTOR1 [BY SUBFACTOR1]
             [ FACTOR2 [BY SUBFACTOR2]]
             ...
             [ FACTOR3 [BY SUBFACTOR3]]
            ]
        /STATISTICS={DESCRIPTIVES, EXTREME[(N)], ALL, NONE}
        /PLOT={BOXPLOT, NPPLOT, HISTOGRAM, SPREADLEVEL[(T)], ALL, NONE}
        /CINTERVAL P
        /COMPARE={GROUPS,VARIABLES}
        /ID=IDENTITY_VARIABLE
        /{TOTAL,NOTOTAL}
        /PERCENTILE=[PERCENTILES]={HAVERAGE, WAVERAGE, ROUND, AEMPIRICAL, EMPIRICAL }
        /MISSING={LISTWISE, PAIRWISE} [{EXCLUDE, INCLUDE}]
       [{NOREPORT,REPORT}]

EXAMINE is used to perform exploratory data analysis. In particular, it is useful for testing how closely a distribution follows a normal distribution, and for finding outliers and extreme values.

The VARIABLES subcommand is mandatory. It specifies the dependent variables and optionally variables to use as factors for the analysis. Variables listed before the first BY keyword (if any) are the dependent variables. The dependent variables may optionally be followed by a list of factors which tell PSPP how to break down the analysis for each dependent variable.

Following the dependent variables, factors may be specified. The factors (if desired) should be preceded by a single BY keyword. The format for each factor is FACTORVAR [BY SUBFACTORVAR]. Each unique combination of the values of FACTORVAR and SUBFACTORVAR divide the dataset into "cells". Statistics are calculated for each cell and for the entire dataset (unless NOTOTAL is given).

The STATISTICS subcommand specifies which statistics to show. DESCRIPTIVES produces a table showing some parametric and non-parametrics statistics. EXTREME produces a table showing the extremities of each cell. A number in parentheses determines how many upper and lower extremities to show. The default number is 5.

The subcommands TOTAL and NOTOTAL are mutually exclusive. If TOTAL appears, then statistics for the entire dataset as well as for each cell are produced. If NOTOTAL appears, then statistics are produced only for the cells (unless no factor variables have been given). These subcommands have no effect if there have been no factor variables specified.

The PLOT subcommand specifies which plots are to be produced if any. Available plots are HISTOGRAM, NPPLOT, BOXPLOT and SPREADLEVEL. The first three can be used to visualise how closely each cell conforms to a normal distribution, whilst the spread vs. level plot can be useful to visualise how the variance differs between factors. Boxplots show you the outliers and extreme values.1

The SPREADLEVEL plot displays the interquartile range versus the median. It takes an optional parameter T, which specifies how the data should be transformed prior to plotting. The given value T is a power to which the data are raised. For example, if T is given as 2, then the square of the data is used. Zero, however is a special value. If T is 0 or is omitted, then data are transformed by taking its natural logarithm instead of raising to the power of T.

When one or more plots are requested, EXAMINE also performs the Shapiro-Wilk test for each category. There are however a number of provisos:

  • All weight values must be integer.
  • The cumulative weight value must be in the range [3, 5000].

The COMPARE subcommand is only relevant if producing boxplots, and it is only useful there is more than one dependent variable and at least one factor. If /COMPARE=GROUPS is specified, then one plot per dependent variable is produced, each of which contain boxplots for all the cells. If /COMPARE=VARIABLES is specified, then one plot per cell is produced, each containing one boxplot per dependent variable. If the /COMPARE subcommand is omitted, then PSPP behaves as if /COMPARE=GROUPS were given.

The ID subcommand is relevant only if /PLOT=BOXPLOT or /STATISTICS=EXTREME has been given. If given, it should provide the name of a variable which is to be used to labels extreme values and outliers. Numeric or string variables are permissible. If the ID subcommand is not given, then the case number is used for labelling.

The CINTERVAL subcommand specifies the confidence interval to use in calculation of the descriptives command. The default is 95%.

The PERCENTILES subcommand specifies which percentiles are to be calculated, and which algorithm to use for calculating them. The default is to calculate the 5, 10, 25, 50, 75, 90, 95 percentiles using the HAVERAGE algorithm.

The TOTAL and NOTOTAL subcommands are mutually exclusive. If NOTOTAL is given and factors have been specified in the VARIABLES subcommand, then statistics for the unfactored dependent variables are produced in addition to the factored variables. If there are no factors specified then TOTAL and NOTOTAL have no effect.

The following example generates descriptive statistics and histograms for two variables score1 and score2. Two factors are given: gender and gender BY culture. Therefore, the descriptives and histograms are generated for each distinct value of gender and for each distinct combination of the values of gender and race. Since the NOTOTAL keyword is given, statistics and histograms for score1 and score2 covering the whole dataset are not produced.

EXAMINE score1 score2 BY
        gender
        gender BY culture
        /STATISTICS = DESCRIPTIVES
        /PLOT = HISTOGRAM
        /NOTOTAL.

Here is a second example showing how EXAMINE may be used to find extremities.

EXAMINE height weight BY
        gender
        /STATISTICS = EXTREME (3)
        /PLOT = BOXPLOT
        /COMPARE = GROUPS
        /ID = name.

In this example, we look at the height and weight of a sample of individuals and how they differ between male and female. A table showing the 3 largest and the 3 smallest values of height and weight for each gender, and for the whole dataset as are shown. In addition, the /PLOT subcommand requests boxplots. Because /COMPARE = GROUPS was specified, boxplots for male and female are shown in juxtaposed in the same graphic, allowing us to easily see the difference between the genders. Since the variable name was specified on the ID subcommand, values of the name variable are used to label the extreme values.

⚠️ If you specify many dependent variables or factor variables for which there are many distinct values, then EXAMINE will produce a very large quantity of output.

  1. HISTOGRAM uses Sturges' rule to determine the number of bins, as approximately \(1 + \log2(n)\), where \(n\) is the number of samples. (FREQUENCIES uses a different algorithm to find the bin size.) ↩

GRAPH

GRAPH
        /HISTOGRAM [(NORMAL)]= VAR
        /SCATTERPLOT [(BIVARIATE)] = VAR1 WITH VAR2 [BY VAR3]
        /BAR = {SUMMARY-FUNCTION(VAR1) | COUNT-FUNCTION} BY VAR2 [BY VAR3]
        [ /MISSING={LISTWISE, VARIABLE} [{EXCLUDE, INCLUDE}] ]
       [{NOREPORT,REPORT}]

GRAPH produces a graphical plots of data. Only one of the subcommands HISTOGRAM, BAR or SCATTERPLOT can be specified, i.e. only one plot can be produced per call of GRAPH. The MISSING is optional.

Scatterplot

The subcommand SCATTERPLOT produces an xy plot of the data. GRAPH uses VAR3, if specified, to determine the colours and/or markers for the plot. The following is an example for producing a scatterplot.

GRAPH
        /SCATTERPLOT = height WITH weight BY gender.

This example produces a scatterplot where height is plotted versus weight. Depending on the value of gender, the colour of the datapoint is different. With this plot it is possible to analyze gender differences for height versus weight relation.

Histogram

The subcommand HISTOGRAM produces a histogram. Only one variable is allowed for the histogram plot. The keyword NORMAL may be specified in parentheses, to indicate that the ideal normal curve should be superimposed over the histogram. For an alternative method to produce histograms, see EXAMINE. The following example produces a histogram plot for the variable weight.

GRAPH
        /HISTOGRAM = weight.

Bar Chart

The subcommand BAR produces a bar chart. This subcommand requires that a COUNT-FUNCTION be specified (with no arguments) or a SUMMARY-FUNCTION with a variable VAR1 in parentheses. Following the summary or count function, the keyword BY should be specified and then a catagorical variable, VAR2. The values of VAR2 determine the labels of the bars to be plotted. A second categorical variable VAR3 may be specified, in which case a clustered (grouped) bar chart is produced.

Valid count functions are:

  • COUNT
    The weighted counts of the cases in each category.
  • PCT
    The weighted counts of the cases in each category expressed as a percentage of the total weights of the cases.
  • CUFREQ
    The cumulative weighted counts of the cases in each category.
  • CUPCT
    The cumulative weighted counts of the cases in each category expressed as a percentage of the total weights of the cases.

The summary function is applied to VAR1 across all cases in each category. The recognised summary functions are:

  • SUM
    The sum.
  • MEAN
    The arithmetic mean.
  • MAXIMUM
    The maximum value.
  • MINIMUM
    The minimum value.

The following examples assume a dataset which is the results of a survey. Each respondent has indicated annual income, their sex and city of residence. One could create a bar chart showing how the mean income varies between of residents of different cities, thus:

GRAPH  /BAR  = MEAN(INCOME) BY CITY.

This can be extended to also indicate how income in each city differs between the sexes.

GRAPH  /BAR  = MEAN(INCOME) BY CITY BY SEX.

One might also want to see how many respondents there are from each city. This can be achieved as follows:

GRAPH  /BAR  = COUNT BY CITY.

The FREQUENCIES and CROSSTABS commands can also produce bar charts.

CORRELATIONS

CORRELATIONS
     /VARIABLES = VAR_LIST [ WITH VAR_LIST ]
     [
      .
      .
      .
      /VARIABLES = VAR_LIST [ WITH VAR_LIST ]
      /VARIABLES = VAR_LIST [ WITH VAR_LIST ]
     ]

     [ /PRINT={TWOTAIL, ONETAIL} {SIG, NOSIG} ]
     [ /STATISTICS=DESCRIPTIVES XPROD ALL]
     [ /MISSING={PAIRWISE, LISTWISE} {INCLUDE, EXCLUDE} ]

The CORRELATIONS procedure produces tables of the Pearson correlation coefficient for a set of variables. The significance of the coefficients are also given.

At least one VARIABLES subcommand is required. If you specify the WITH keyword, then a non-square correlation table is produced. The variables preceding WITH, are used as the rows of the table, and the variables following WITH are used as the columns of the table. If no WITH subcommand is specified, then CORRELATIONS produces a square, symmetrical table using all variables.

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set, then user-missing values are included in the calculations, but system-missing values are not. If EXCLUDE is set, which is the default, user-missing values are excluded as well as system-missing values.

If LISTWISE is set, then the entire case is excluded from analysis whenever any variable specified in any /VARIABLES subcommand contains a missing value. If PAIRWISE is set, then a case is considered missing only if either of the values for the particular coefficient are missing. The default is PAIRWISE.

The PRINT subcommand is used to control how the reported significance values are printed. If the TWOTAIL option is used, then a two-tailed test of significance is printed. If the ONETAIL option is given, then a one-tailed test is used. The default is TWOTAIL.

If the NOSIG option is specified, then correlation coefficients with significance less than 0.05 are highlighted. If SIG is specified, then no highlighting is performed. This is the default.

The STATISTICS subcommand requests additional statistics to be displayed. The keyword DESCRIPTIVES requests that the mean, number of non-missing cases, and the non-biased estimator of the standard deviation are displayed. These statistics are displayed in a separated table, for all the variables listed in any /VARIABLES subcommand. The XPROD keyword requests cross-product deviations and covariance estimators to be displayed for each pair of variables. The keyword ALL is the union of DESCRIPTIVES and XPROD.

CROSSTABS

CROSSTABS
        /TABLES=VAR_LIST BY VAR_LIST [BY VAR_LIST]...
        /MISSING={TABLE,INCLUDE,REPORT}
        /FORMAT={TABLES,NOTABLES}
                {AVALUE,DVALUE}
        /CELLS={COUNT,ROW,COLUMN,TOTAL,EXPECTED,RESIDUAL,SRESIDUAL,
                ASRESIDUAL,ALL,NONE}
        /COUNT={ASIS,CASE,CELL}
               {ROUND,TRUNCATE}
        /STATISTICS={CHISQ,PHI,CC,LAMBDA,UC,BTAU,CTAU,RISK,GAMMA,D,
                     KAPPA,ETA,CORR,ALL,NONE}
        /BARCHART

(Integer mode.)
        /VARIABLES=VAR_LIST (LOW,HIGH)...

The CROSSTABS procedure displays crosstabulation tables requested by the user. It can calculate several statistics for each cell in the crosstabulation tables. In addition, a number of statistics can be calculated for each table itself.

The TABLES subcommand is used to specify the tables to be reported. Any number of dimensions is permitted, and any number of variables per dimension is allowed. The TABLES subcommand may be repeated as many times as needed. This is the only required subcommand in "general mode".

Occasionally, one may want to invoke a special mode called "integer mode". Normally, in general mode, PSPP automatically determines what values occur in the data. In integer mode, the user specifies the range of values that the data assumes. To invoke this mode, specify the VARIABLES subcommand, giving a range of data values in parentheses for each variable to be used on the TABLES subcommand. Data values inside the range are truncated to the nearest integer, then assigned to that value. If values occur outside this range, they are discarded. When it is present, the VARIABLES subcommand must precede the TABLES subcommand.

In general mode, numeric and string variables may be specified on TABLES. In integer mode, only numeric variables are allowed.

The MISSING subcommand determines the handling of user-missing values. When set to TABLE, the default, missing values are dropped on a table by table basis. When set to INCLUDE, user-missing values are included in tables and statistics. When set to REPORT, which is allowed only in integer mode, user-missing values are included in tables but marked with a footnote and excluded from statistical calculations.

The FORMAT subcommand controls the characteristics of the crosstabulation tables to be displayed. It has a number of possible settings:

  • TABLES, the default, causes crosstabulation tables to be output.

  • NOTABLES, which is equivalent to CELLS=NONE, suppresses them.

  • AVALUE, the default, causes values to be sorted in ascending order. DVALUE asserts a descending sort order.

The CELLS subcommand controls the contents of each cell in the displayed crosstabulation table. The possible settings are:

  • COUNT
    Frequency count.
  • ROW
    Row percent.
  • COLUMN
    Column percent.
  • TOTAL
    Table percent.
  • EXPECTED
    Expected value.
  • RESIDUAL
    Residual.
  • SRESIDUAL
    Standardized residual.
  • ASRESIDUAL
    Adjusted standardized residual.
  • ALL
    All of the above.
  • NONE
    Suppress cells entirely.

/CELLS without any settings specified requests COUNT, ROW, COLUMN, and TOTAL. If CELLS is not specified at all then only COUNT is selected.

By default, crosstabulation and statistics use raw case weights, without rounding. Use the /COUNT subcommand to perform rounding: CASE rounds the weights of individual weights as cases are read, CELL rounds the weights of cells within each crosstabulation table after it has been constructed, and ASIS explicitly specifies the default non-rounding behavior. When rounding is requested, ROUND, the default, rounds to the nearest integer and TRUNCATE rounds toward zero.

The STATISTICS subcommand selects statistics for computation:

  • CHISQ
    Pearson chi-square, likelihood ratio, Fisher's exact test, continuity correction, linear-by-linear association.
  • PHI
    Phi.
  • CC
    Contingency coefficient.
  • LAMBDA
    Lambda.
  • UC
    Uncertainty coefficient.
  • BTAU
    Tau-b.
  • CTAU
    Tau-c.
  • RISK
    Risk estimate.
  • GAMMA
    Gamma.
  • D
    Somers' D.
  • KAPPA
    Cohen's Kappa.
  • ETA
    Eta.
  • CORR
    Spearman correlation, Pearson's r.
  • ALL
    All of the above.
  • NONE
    No statistics.

Selected statistics are only calculated when appropriate for the statistic. Certain statistics require tables of a particular size, and some statistics are calculated only in integer mode.

/STATISTICS without any settings selects CHISQ. If the STATISTICS subcommand is not given, no statistics are calculated.

The /BARCHART subcommand produces a clustered bar chart for the first two variables on each table. If a table has more than two variables, the counts for the third and subsequent levels are aggregated and the chart is produced as if there were only two variables.

Currently the implementation of CROSSTABS has the following limitations:

  • Significance of some symmetric and directional measures is not calculated.
  • Asymptotic standard error is not calculated for Goodman and Kruskal's tau or symmetric Somers' d.
  • Approximate T is not calculated for symmetric uncertainty coefficient.

Fixes for any of these deficiencies would be welcomed.

Example

A researcher wishes to know if, in an industry, a person's sex is related to the person's occupation. To investigate this, she has determined that the personnel.sav is a representative, randomly selected sample of persons. The researcher's null hypothesis is that a person's sex has no relation to a person's occupation. She uses a chi-squared test of independence to investigate the hypothesis.

get file="personnel.sav".

crosstabs
   /tables= occupation by sex
   /cells = count expected
   /statistics=chisq.

The syntax above conducts a chi-squared test of independence. The line /tables = occupation by sex indicates that occupation and sex are the variables to be tabulated.

As shown in the output below, CROSSTABS generates a contingency table containing the observed count and the expected count of each sex and each occupation. The expected count is the count which would be observed if the null hypothesis were true.

The significance of the Pearson Chi-Square value is very much larger than the normally accepted value of 0.05 and so one cannot reject the null hypothesis. Thus the researcher must conclude that a person's sex has no relation to the person's occupation.

                      Summary
┌────────────────┬───────────────────────────────┐
│                │             Cases             │
│                ├──────────┬─────────┬──────────┤
│                │   Valid  │ Missing │   Total  │
│                ├──┬───────┼─┬───────┼──┬───────┤
│                │ N│Percent│N│Percent│ N│Percent│
├────────────────┼──┼───────┼─┼───────┼──┼───────┤
│occupation × sex│54│  96.4%│2│   3.6%│56│ 100.0%│
└────────────────┴──┴───────┴─┴───────┴──┴───────┘

                     occupation × sex
┌──────────────────────────────────────┬───────────┬─────┐
│                                      │    sex    │     │
│                                      ├────┬──────┤     │
│                                      │Male│Female│Total│
├──────────────────────────────────────┼────┼──────┼─────┤
│occupation Artist             Count   │   2│     6│    8│
│                              Expected│4.89│  3.11│  .15│
│          ────────────────────────────┼────┼──────┼─────┤
│           Baker              Count   │   1│     1│    2│
│                              Expected│1.22│   .78│  .04│
│          ────────────────────────────┼────┼──────┼─────┤
│           Barrister          Count   │   0│     1│    1│
│                              Expected│ .61│   .39│  .02│
│          ────────────────────────────┼────┼──────┼─────┤
│           Carpenter          Count   │   3│     1│    4│
│                              Expected│2.44│  1.56│  .07│
│          ────────────────────────────┼────┼──────┼─────┤
│           Cleaner            Count   │   4│     0│    4│
│                              Expected│2.44│  1.56│  .07│
│          ────────────────────────────┼────┼──────┼─────┤
│           Cook               Count   │   3│     2│    5│
│                              Expected│3.06│  1.94│  .09│
│          ────────────────────────────┼────┼──────┼─────┤
│           Manager            Count   │   4│     4│    8│
│                              Expected│4.89│  3.11│  .15│
│          ────────────────────────────┼────┼──────┼─────┤
│           Mathematician      Count   │   3│     1│    4│
│                              Expected│2.44│  1.56│  .07│
│          ────────────────────────────┼────┼──────┼─────┤
│           Painter            Count   │   1│     1│    2│
│                              Expected│1.22│   .78│  .04│
│          ────────────────────────────┼────┼──────┼─────┤
│           Payload Specialist Count   │   1│     0│    1│
│                              Expected│ .61│   .39│  .02│
│          ────────────────────────────┼────┼──────┼─────┤
│           Plumber            Count   │   5│     0│    5│
│                              Expected│3.06│  1.94│  .09│
│          ────────────────────────────┼────┼──────┼─────┤
│           Scientist          Count   │   5│     2│    7│
│                              Expected│4.28│  2.72│  .13│
│          ────────────────────────────┼────┼──────┼─────┤
│           Scrientist         Count   │   0│     1│    1│
│                              Expected│ .61│   .39│  .02│
│          ────────────────────────────┼────┼──────┼─────┤
│           Tailor             Count   │   1│     1│    2│
│                              Expected│1.22│   .78│  .04│
├──────────────────────────────────────┼────┼──────┼─────┤
│Total                         Count   │  33│    21│   54│
│                              Expected│ .61│   .39│ 1.00│
└──────────────────────────────────────┴────┴──────┴─────┘

                    Chi─Square Tests
┌──────────────────┬─────┬──┬──────────────────────────┐
│                  │Value│df│Asymptotic Sig. (2─tailed)│
├──────────────────┼─────┼──┼──────────────────────────┤
│Pearson Chi─Square│15.59│13│                      .272│
│Likelihood Ratio  │19.66│13│                      .104│
│N of Valid Cases  │   54│  │                          │
└──────────────────┴─────┴──┴──────────────────────────┘

CTABLES

CTABLES has the following overall syntax. At least one TABLE subcommand is required:

CTABLES
  ...global subcommands...
  [/TABLE axis [BY axis [BY axis]]
   ...per-table subcommands...]...

where each axis may be empty or take one of the following forms:

variable
variable [{C | S}]
axis + axis
axis > axis
(axis)
axis [summary [string] [format]]

The following subcommands precede the first TABLE subcommand and apply to all of the output tables. All of these subcommands are optional:

/FORMAT
    [MINCOLWIDTH={DEFAULT | width}]
    [MAXCOLWIDTH={DEFAULT | width}]
    [UNITS={POINTS | INCHES | CM}]
    [EMPTY={ZERO | BLANK | string}]
    [MISSING=string]
/VLABELS
    VARIABLES=variables
    DISPLAY={DEFAULT | NAME | LABEL | BOTH | NONE}
/SMISSING {VARIABLE | LISTWISE}
/PCOMPUTE &postcompute=EXPR(expression)
/PPROPERTIES &postcompute...
    [LABEL=string]
    [FORMAT=[summary format]...]
    [HIDESOURCECATS={NO | YES}
/WEIGHT VARIABLE=variable
/HIDESMALLCOUNTS COUNT=count

The following subcommands follow TABLE and apply only to the previous TABLE. All of these subcommands are optional:

/SLABELS
    [POSITION={COLUMN | ROW | LAYER}]
    [VISIBLE={YES | NO}]
/CLABELS {AUTO | {ROWLABELS|COLLABELS}={OPPOSITE|LAYER}}
/CATEGORIES VARIABLES=variables
    {[value, value...]
   | [ORDER={A | D}]
     [KEY={VALUE | LABEL | summary(variable)}]
     [MISSING={EXCLUDE | INCLUDE}]}
    [TOTAL={NO | YES} [LABEL=string] [POSITION={AFTER | BEFORE}]]
    [EMPTY={INCLUDE | EXCLUDE}]
/TITLES
    [TITLE=string...]
    [CAPTION=string...]
    [CORNER=string...]

The CTABLES (aka "custom tables") command produces multi-dimensional tables from categorical and scale data. It offers many options for data summarization and formatting.

This section's examples use data from the 2008 (USA) National Survey of Drinking and Driving Attitudes and Behaviors, a public domain data set from the (USA) National Highway Traffic Administration and available at https://data.transportation.gov. PSPP includes this data set, with a modified dictionary, as examples/nhtsa.sav.

Basics

The only required subcommand is TABLE, which specifies the variables to include along each axis:

     /TABLE rows [BY columns [BY layers]]

In TABLE, each of ROWS, COLUMNS, and LAYERS is either empty or an axis expression that specifies one or more variables. At least one must specify an axis expression.

Categorical Variables

An axis expression that names a categorical variable divides the data into cells according to the values of that variable. When all the variables named on TABLE are categorical, by default each cell displays the number of cases that it contains, so specifying a single variable yields a frequency table, much like the output of the FREQUENCIES command:

     CTABLES /TABLE=ageGroup.

         Custom Tables
┌───────────────────────┬─────┐
│                       │Count│
├───────────────────────┼─────┤
│Age group 15 or younger│    0│
│          16 to 25     │ 1099│
│          26 to 35     │  967│
│          36 to 45     │ 1037│
│          46 to 55     │ 1175│
│          56 to 65     │ 1247│
│          66 or older  │ 1474│
└───────────────────────┴─────┘

Specifying a row and a column categorical variable yields a crosstabulation, much like the output of the CROSSTABS command:

CTABLES /TABLE=ageGroup BY gender.

             Custom Tables
┌───────────────────────┬────────────┐
│                       │S3a. GENDER:│
│                       ├─────┬──────┤
│                       │ Male│Female│
│                       ├─────┼──────┤
│                       │Count│ Count│
├───────────────────────┼─────┼──────┤
│Age group 15 or younger│    0│     0│
│          16 to 25     │  594│   505│
│          26 to 35     │  476│   491│
│          36 to 45     │  489│   548│
│          46 to 55     │  526│   649│
│          56 to 65     │  516│   731│
│          66 or older  │  531│   943│
└───────────────────────┴─────┴──────┘

The > "nesting" operator nests multiple variables on a single axis, e.g.:

CTABLES /TABLE likelihoodOfBeingStoppedByPolice BY ageGroup > gender.

                                 Custom Tables
┌─────────────────────────────────┬───────────────────────────────────────────┐
│                                 │  86. In the past year, have you hosted a  │
│                                 │  social event or party where alcohol was  │
│                                 │             served to adults?             │
│                                 ├─────────────────────┬─────────────────────┤
│                                 │         Yes         │          No         │
│                                 ├─────────────────────┼─────────────────────┤
│                                 │        Count        │        Count        │
├─────────────────────────────────┼─────────────────────┼─────────────────────┤
│Age    15 or      S3a.     Male  │                    0│                    0│
│group  younger    GENDER:  Female│                    0│                    0│
│      ───────────────────────────┼─────────────────────┼─────────────────────┤
│       16 to 25   S3a.     Male  │                  208│                  386│
│                  GENDER:  Female│                  202│                  303│
│      ───────────────────────────┼─────────────────────┼─────────────────────┤
│       26 to 35   S3a.     Male  │                  225│                  251│
│                  GENDER:  Female│                  242│                  249│
│      ───────────────────────────┼─────────────────────┼─────────────────────┤
│       36 to 45   S3a.     Male  │                  223│                  266│
│                  GENDER:  Female│                  240│                  307│
│      ───────────────────────────┼─────────────────────┼─────────────────────┤
│       46 to 55   S3a.     Male  │                  201│                  325│
│                  GENDER:  Female│                  282│                  366│
│      ───────────────────────────┼─────────────────────┼─────────────────────┤
│       56 to 65   S3a.     Male  │                  196│                  320│
│                  GENDER:  Female│                  279│                  452│
│      ───────────────────────────┼─────────────────────┼─────────────────────┤
│       66 or      S3a.     Male  │                  162│                  367│
│       older      GENDER:  Female│                  243│                  700│
└─────────────────────────────────┴─────────────────────┴─────────────────────┘

The + "stacking" operator allows a single output table to include multiple data analyses. With +, CTABLES divides the output table into multiple "sections", each of which includes an analysis of the full data set. For example, the following command separately tabulates age group and driving frequency by gender:

CTABLES /TABLE ageGroup + freqOfDriving BY gender.

                                 Custom Tables
┌────────────────────────────────────────────────────────────────┬────────────┐
│                                                                │S3a. GENDER:│
│                                                                ├─────┬──────┤
│                                                                │ Male│Female│
│                                                                ├─────┼──────┤
│                                                                │Count│ Count│
├────────────────────────────────────────────────────────────────┼─────┼──────┤
│Age group                                    15 or younger      │    0│     0│
│                                             16 to 25           │  594│   505│
│                                             26 to 35           │  476│   491│
│                                             36 to 45           │  489│   548│
│                                             46 to 55           │  526│   649│
│                                             56 to 65           │  516│   731│
│                                             66 or older        │  531│   943│
├────────────────────────────────────────────────────────────────┼─────┼──────┤
│ 1. How often do you usually drive a car or  Every day          │ 2305│  2362│
│other motor vehicle?                         Several days a week│  440│   834│
│                                             Once a week or less│  125│   236│
│                                             Only certain times │   58│    72│
│                                             a year             │     │      │
│                                             Never              │  192│   348│
└────────────────────────────────────────────────────────────────┴─────┴──────┘

When + and > are used together, > binds more tightly. Use parentheses to override operator precedence. Thus:

CTABLES /TABLE hasConsideredReduction + hasBeenCriticized > gender.
CTABLES /TABLE (hasConsideredReduction + hasBeenCriticized) > gender.

                                 Custom Tables
┌───────────────────────────────────────────────────────────────────────┬─────┐
│                                                                       │Count│
├───────────────────────────────────────────────────────────────────────┼─────┤
│26. During the last 12 months, has there been a    Yes                 │  513│
│time when you felt you should cut down on your    ─────────────────────┼─────┤
│drinking?                                          No                  │ 3710│
├───────────────────────────────────────────────────────────────────────┼─────┤
│27. During the last 12 months, has there been a    Yes S3a.      Male  │  135│
│time when people criticized your drinking?             GENDER:   Female│   49│
│                                                  ─────────────────────┼─────┤
│                                                   No  S3a.      Male  │ 1916│
│                                                       GENDER:   Female│ 2126│
└───────────────────────────────────────────────────────────────────────┴─────┘

                                 Custom Tables
┌───────────────────────────────────────────────────────────────────────┬─────┐
│                                                                       │Count│
├───────────────────────────────────────────────────────────────────────┼─────┤
│26. During the last 12 months, has there been a    Yes S3a.      Male  │  333│
│time when you felt you should cut down on your         GENDER:   Female│  180│
│drinking?                                         ─────────────────────┼─────┤
│                                                   No  S3a.      Male  │ 1719│
│                                                       GENDER:   Female│ 1991│
├───────────────────────────────────────────────────────────────────────┼─────┤
│27. During the last 12 months, has there been a    Yes S3a.      Male  │  135│
│time when people criticized your drinking?             GENDER:   Female│   49│
│                                                  ─────────────────────┼─────┤
│                                                   No  S3a.      Male  │ 1916│
│                                                       GENDER:   Female│ 2126│
└───────────────────────────────────────────────────────────────────────┴─────┘

Scalar Variables

For a categorical variable, CTABLES divides the table into a cell per category. For a scalar variable, CTABLES instead calculates a summary measure, by default the mean, of the values that fall into a cell. For example, if the only variable specified is a scalar variable, then the output is a single cell that holds the mean of all of the data:

CTABLES /TABLE age.

          Custom Tables
┌──────────────────────────┬────┐
│                          │Mean│
├──────────────────────────┼────┤
│D1. AGE: What is your age?│  48│
└──────────────────────────┴────┘

A scalar variable may nest with categorical variables. The following example shows the mean age of survey respondents across gender and language groups:

CTABLES /TABLE gender > age BY region.

Custom Tables
┌─────────────────────────────────────┬───────────────────────────────────────┐
│                                     │Was this interview conducted in English│
│                                     │              or Spanish?              │
│                                     ├───────────────────┬───────────────────┤
│                                     │      English      │      Spanish      │
│                                     ├───────────────────┼───────────────────┤
│                                     │        Mean       │        Mean       │
├─────────────────────────────────────┼───────────────────┼───────────────────┤
│D1. AGE: What is   S3a.        Male  │                 46│                 37│
│your age?          GENDER:     Female│                 51│                 39│
└─────────────────────────────────────┴───────────────────┴───────────────────┘

The order of nesting of scalar and categorical variables affects table labeling, but it does not affect the data displayed in the table. The following example shows how the output changes when the nesting order of the scalar and categorical variable are interchanged:

CTABLES /TABLE age > gender BY region.

                                 Custom Tables
┌─────────────────────────────────────┬───────────────────────────────────────┐
│                                     │Was this interview conducted in English│
│                                     │              or Spanish?              │
│                                     ├───────────────────┬───────────────────┤
│                                     │      English      │      Spanish      │
│                                     ├───────────────────┼───────────────────┤
│                                     │        Mean       │        Mean       │
├─────────────────────────────────────┼───────────────────┼───────────────────┤
│S3a.       Male   D1. AGE: What is   │                 46│                 37│
│GENDER:           your age?          │                   │                   │
│          ───────────────────────────┼───────────────────┼───────────────────┤
│           Female D1. AGE: What is   │                 51│                 39│
│                  your age?          │                   │                   │
└─────────────────────────────────────┴───────────────────┴───────────────────┘

Only a single scalar variable may appear in each section; that is, a scalar variable may not nest inside a scalar variable directly or indirectly. Scalar variables may only appear on one axis within TABLE.

Overriding Measurement Level

By default, CTABLES uses a variable's measurement level to decide whether to treat it as categorical or scalar. Variables assigned the nominal or ordinal measurement level are treated as categorical, and scalar variables are treated as scalar.

When PSPP reads data from a file in an external format, such as a text file, variables' measurement levels are often unknown. If CTABLES runs when a variable has an unknown measurement level, it makes an initial pass through the data to guess measurement levels. Use the VARIABLE LEVEL command to set or change a variable's measurement level.

To treat a variable as categorical or scalar only for one use on CTABLES, add [C] or [S], respectively, after the variable name. The following example shows the output when variable monthDaysMin1drink is analyzed as scalar (the default for its measurement level) and as categorical:

CTABLES
    /TABLE monthDaysMin1drink BY gender
    /TABLE monthDaysMin1drink [C] BY gender.

                                 Custom Tables
┌────────────────────────────────────────────────────────────────┬────────────┐
│                                                                │S3a. GENDER:│
│                                                                ├────┬───────┤
│                                                                │Male│ Female│
│                                                                ├────┼───────┤
│                                                                │Mean│  Mean │
├────────────────────────────────────────────────────────────────┼────┼───────┤
│20. On how many of the thirty days in this typical month did you│   7│      5│
│have one or more alcoholic beverages to drink?                  │    │       │
└────────────────────────────────────────────────────────────────┴────┴───────┘

                                 Custom Tables
┌────────────────────────────────────────────────────────────────┬────────────┐
│                                                                │S3a. GENDER:│
│                                                                ├─────┬──────┤
│                                                                │ Male│Female│
│                                                                ├─────┼──────┤
│                                                                │Count│ Count│
├────────────────────────────────────────────────────────────────┼─────┼──────┤
│20. On how many of the thirty days in this typical month None   │  152│   258│
│did you have one or more alcoholic beverages to drink?   1      │  403│   653│
│                                                         2      │  284│   324│
│                                                         3      │  169│   215│
│                                                         4      │  178│   143│
│                                                         5      │  107│   106│
│                                                         6      │   67│    59│
│                                                         7      │   31│    11│
│                                                         8      │  101│    74│
│                                                         9      │    6│     4│
│                                                         10     │   95│    75│
│                                                         11     │    4│     0│
│                                                         12     │   58│    33│
│                                                         13     │    3│     2│
│                                                         14     │   13│     3│
│                                                         15     │   79│    58│
│                                                         16     │   10│     6│
│                                                         17     │    4│     2│
│                                                         18     │    5│     4│
│                                                         19     │    2│     0│
│                                                         20     │  105│    47│
│                                                         21     │    2│     0│
│                                                         22     │    3│     3│
│                                                         23     │    0│     3│
│                                                         24     │    3│     0│
│                                                         25     │   35│    25│
│                                                         26     │    1│     1│
│                                                         27     │    3│     3│
│                                                         28     │   13│     8│
│                                                         29     │    3│     3│
│                                                         Every  │  104│    43│
│                                                         day    │     │      │
└────────────────────────────────────────────────────────────────┴─────┴──────┘

Data Summarization

The CTABLES command allows the user to control how the data are summarized with "summary specifications", syntax that lists one or more summary function names, optionally separated by commas, and which are enclosed in square brackets following a variable name on the TABLE subcommand. When all the variables are categorical, summary specifications can be given for the innermost nested variables on any one axis. When a scalar variable is present, only the scalar variable may have summary specifications.

The following example includes a summary specification for column and row percentages for categorical variables, and mean and median for a scalar variable:

CTABLES
    /TABLE=age [MEAN, MEDIAN] BY gender
    /TABLE=ageGroup [COLPCT, ROWPCT] BY gender.

                    Custom Tables
┌──────────────────────────┬───────────────────────┐
│                          │      S3a. GENDER:     │
│                          ├───────────┬───────────┤
│                          │    Male   │   Female  │
│                          ├────┬──────┼────┬──────┤
│                          │Mean│Median│Mean│Median│
├──────────────────────────┼────┼──────┼────┼──────┤
│D1. AGE: What is your age?│  46│    45│  50│    52│
└──────────────────────────┴────┴──────┴────┴──────┘

                     Custom Tables
┌───────────────────────┬─────────────────────────────┐
│                       │         S3a. GENDER:        │
│                       ├──────────────┬──────────────┤
│                       │     Male     │    Female    │
│                       ├────────┬─────┼────────┬─────┤
│                       │Column %│Row %│Column %│Row %│
├───────────────────────┼────────┼─────┼────────┼─────┤
│Age group 15 or younger│     .0%│    .│     .0%│    .│
│          16 to 25     │   19.0%│54.0%│   13.1%│46.0%│
│          26 to 35     │   15.2%│49.2%│   12.7%│50.8%│
│          36 to 45     │   15.6%│47.2%│   14.2%│52.8%│
│          46 to 55     │   16.8%│44.8%│   16.8%│55.2%│
│          56 to 65     │   16.5%│41.4%│   18.9%│58.6%│
│          66 or older  │   17.0%│36.0%│   24.4%│64.0%│
└───────────────────────┴────────┴─────┴────────┴─────┘

A summary specification may override the default label and format by appending a string or format specification or both (in that order) to the summary function name. For example:

CTABLES /TABLE=ageGroup [COLPCT 'Gender %' PCT5.0,
                         ROWPCT 'Age Group %' PCT5.0]
               BY gender.

                           Custom Tables
┌───────────────────────┬─────────────────────────────────────────┐
│                       │               S3a. GENDER:              │
│                       ├────────────────────┬────────────────────┤
│                       │        Male        │       Female       │
│                       ├────────┬───────────┼────────┬───────────┤
│                       │Gender %│Age Group %│Gender %│Age Group %│
├───────────────────────┼────────┼───────────┼────────┼───────────┤
│Age group 15 or younger│      0%│          .│      0%│          .│
│          16 to 25     │     19%│        54%│     13%│        46%│
│          26 to 35     │     15%│        49%│     13%│        51%│
│          36 to 45     │     16%│        47%│     14%│        53%│
│          46 to 55     │     17%│        45%│     17%│        55%│
│          56 to 65     │     16%│        41%│     19%│        59%│
│          66 or older  │     17%│        36%│     24%│        64%│
└───────────────────────┴────────┴───────────┴────────┴───────────┘

In addition to the standard formats, CTABLES allows the user to specify the following special formats:

FormatDescriptionPositive ExampleNegative Example
NEGPARENw.dEncloses negative numbers in parentheses.42.96(42.96)
NEQUALw.dAdds a N= prefix.N=42.96N=-42.96
PARENw.dEncloses all numbers in parentheses.(42.96)(-42.96)
PCTPARENw.dEncloses all numbers in parentheses with a % suffix.(42.96%)(-42.96%)

Parentheses provide a shorthand to apply summary specifications to multiple variables. For example, both of these commands:

CTABLES /TABLE=ageGroup[COLPCT] + membersOver16[COLPCT] BY gender.
CTABLES /TABLE=(ageGroup + membersOver16)[COLPCT] BY gender.

produce the same output shown below:

                                 Custom Tables
┌─────────────────────────────────────────────────────────────┬───────────────┐
│                                                             │  S3a. GENDER: │
│                                                             ├───────┬───────┤
│                                                             │  Male │ Female│
│                                                             ├───────┼───────┤
│                                                             │ Column│ Column│
│                                                             │   %   │   %   │
├─────────────────────────────────────────────────────────────┼───────┼───────┤
│Age group                                         15 or      │    .0%│    .0%│
│                                                  younger    │       │       │
│                                                  16 to 25   │  19.0%│  13.1%│
│                                                  26 to 35   │  15.2%│  12.7%│
│                                                  36 to 45   │  15.6%│  14.2%│
│                                                  46 to 55   │  16.8%│  16.8%│
│                                                  56 to 65   │  16.5%│  18.9%│
│                                                  66 or older│  17.0%│  24.4%│
├─────────────────────────────────────────────────────────────┼───────┼───────┤
│S1. Including yourself, how many members of this  None       │    .0%│    .0%│
│household are age 16 or older?                    1          │  21.4%│  35.0%│
│                                                  2          │  61.9%│  52.3%│
│                                                  3          │  11.0%│   8.2%│
│                                                  4          │   4.2%│   3.2%│
│                                                  5          │   1.1%│    .9%│
│                                                  6 or more  │    .4%│    .4%│
└─────────────────────────────────────────────────────────────┴───────┴───────┘

The following sections list the available summary functions. After each function's name is given its default label and format. If no format is listed, then the default format is the print format for the variable being summarized.

Summary Functions for Individual Cells

This section lists the summary functions that consider only an individual cell in CTABLES. Only one such summary function, COUNT, may be applied to both categorical and scale variables:

  • COUNT ("Count", F40.0)
    The sum of weights in a cell.

    If CATEGORIES for one or more of the variables in a table include missing values (see Per-Variable Category Options), then some or all of the categories for a cell might be missing values. COUNT counts data included in a cell regardless of whether its categories are missing.

The following summary functions apply only to scale variables or totals and subtotals for categorical variables. Be cautious about interpreting the summary value in the latter case, because it is not necessarily meaningful; however, the mean of a Likert scale, etc. may have a straightforward interpreation.

  • MAXIMUM ("Maximum")
    The largest value.

  • MEAN ("Mean")
    The mean.

  • MEDIAN ("Median")
    The median value.

  • MINIMUM ("Minimum")
    The smallest value.

  • MISSING ("Missing")
    Sum of weights of user- and system-missing values.

  • MODE ("Mode")
    The highest-frequency value. Ties are broken by taking the smallest mode.

  • PTILE n ("Percentile n")
    The Nth percentile, where 0 ≤ N ≤ 100.

  • RANGE ("Range")
    The maximum minus the minimum.

  • SEMEAN ("Std Error of Mean")
    The standard error of the mean.

  • STDDEV ("Std Deviation")
    The standard deviation.

  • SUM ("Sum")
    The sum.

  • TOTALN ("Total N", F40.0)
    The sum of weights in a cell.

    For scale data, COUNT and TOTALN are the same.

    For categorical data, TOTALN counts missing values in excluded categories, that is, user-missing values not in an explicit category list on CATEGORIES (see Per-Variable Category Options), or user-missing values excluded because MISSING=EXCLUDE is in effect on CATEGORIES, or system-missing values. COUNT does not count these.

    See Missing Values for Summary Variables, for details of how CTABLES summarizes missing values.

  • VALIDN ("Valid N", F40.0)
    The sum of valid count weights in included categories.

    For categorical variables, VALIDN does not count missing values regardless of whether they are in included categories via CATEGORIES. VALIDN does not count valid values that are in excluded categories. See Missing Values for Summary Variables for details.

  • VARIANCE ("Variance")
    The variance.

Summary Functions for Groups of Cells

These summary functions summarize over multiple cells within an area of the output chosen by the user and specified as part of the function name. The following basic AREAs are supported, in decreasing order of size:

  • TABLE
    A "section". Stacked variables divide sections of the output from each other. sections may span multiple layers.

  • LAYER
    A section within a single layer.

  • SUBTABLE
    A "subtable", whose contents are the cells that pair an innermost row variable and an innermost column variable within a single layer.

The following shows how the output for the table expression hasBeenPassengerOfDesignatedDriver > hasBeenPassengerOfDrunkDriver BY isLicensedDriver > hasHostedEventWithAlcohol + hasBeenDesignatedDriver BY gender1 is divided up into TABLE, LAYER, and SUBTABLE areas. Each unique value for Table ID is one section, and similarly for Layer ID and Subtable ID. Thus, this output has two TABLE areas (one for isLicensedDriver and one for hasBeenDesignatedDriver), four LAYER areas (for those two variables, per layer), and 12 SUBTABLE areas.

                        Custom Tables
Male
┌─────────────────────────────────┬─────────────────┬──────┐
│                                 │     licensed    │desDrv│
│                                 ├────────┬────────┼───┬──┤
│                                 │   Yes  │   No   │   │  │
│                                 ├────────┼────────┤   │  │
│                                 │ hostAlc│ hostAlc│   │  │
│                                 ├────┬───┼────┬───┤   │  │
│                                 │ Yes│ No│ Yes│ No│Yes│No│
├─────────────────────────────────┼────┼───┼────┼───┼───┼──┤
│desPas Yes druPas Yes Table ID   │   1│  1│   1│  1│  2│ 2│
│                      Layer ID   │   1│  1│   1│  1│  2│ 2│
│                      Subtable ID│   1│  1│   2│  2│  3│ 3│
│                 ────────────────┼────┼───┼────┼───┼───┼──┤
│                  No  Table ID   │   1│  1│   1│  1│  2│ 2│
│                      Layer ID   │   1│  1│   1│  1│  2│ 2│
│                      Subtable ID│   1│  1│   2│  2│  3│ 3│
│      ───────────────────────────┼────┼───┼────┼───┼───┼──┤
│       No  druPas Yes Table ID   │   1│  1│   1│  1│  2│ 2│
│                      Layer ID   │   1│  1│   1│  1│  2│ 2│
│                      Subtable ID│   4│  4│   5│  5│  6│ 6│
│                 ────────────────┼────┼───┼────┼───┼───┼──┤
│                  No  Table ID   │   1│  1│   1│  1│  2│ 2│
│                      Layer ID   │   1│  1│   1│  1│  2│ 2│
│                      Subtable ID│   4│  4│   5│  5│  6│ 6│
└─────────────────────────────────┴────┴───┴────┴───┴───┴──┘

CTABLES also supports the following AREAs that further divide a subtable or a layer within a section:

  • LAYERROW
    LAYERCOL
    A row or column, respectively, in one layer of a section.

  • ROW
    COL
    A row or column, respectively, in a subtable.

The following summary functions for groups of cells are available for each AREA described above, for both categorical and scale variables:

  • areaPCT or areaPCT.COUNT ("Area %", PCT40.1)
    A percentage of total counts within AREA.

  • areaPCT.VALIDN ("Area Valid N %", PCT40.1)
    A percentage of total counts for valid values within AREA.

  • areaPCT.TOTALN ("Area Total N %", PCT40.1)
    A percentage of total counts for all values within AREA.

Scale variables and totals and subtotals for categorical variables may use the following additional group cell summary function:

  • areaPCT.SUM ("Area Sum %", PCT40.1)
    Percentage of the sum of the values within AREA.

Summary Functions for Adjusted Weights

If the WEIGHT subcommand specified an effective weight variable, then the following summary functions use its value instead of the dictionary weight variable. Otherwise, they are equivalent to the summary function without the E-prefix:

  • ECOUNT ("Adjusted Count", F40.0)

  • ETOTALN ("Adjusted Total N", F40.0)

  • EVALIDN ("Adjusted Valid N", F40.0)

Unweighted Summary Functions

The following summary functions with a U-prefix are equivalent to the same ones without the prefix, except that they use unweighted counts:

  • UCOUNT ("Unweighted Count", F40.0)

  • UareaPCT or UareaPCT.COUNT ("Unweighted Area %", PCT40.1)

  • UareaPCT.VALIDN ("Unweighted Area Valid N %", PCT40.1)

  • UareaPCT.TOTALN ("Unweighted Area Total N %", PCT40.1)

  • UMEAN ("Unweighted Mean")

  • UMEDIAN ("Unweighted Median")

  • UMISSING ("Unweighted Missing")

  • UMODE ("Unweighted Mode")

  • UareaPCT.SUM ("Unweighted Area Sum %", PCT40.1)

  • UPTILE n ("Unweighted Percentile n")

  • USEMEAN ("Unweighted Std Error of Mean")

  • USTDDEV ("Unweighted Std Deviation")

  • USUM ("Unweighted Sum")

  • UTOTALN ("Unweighted Total N", F40.0)

  • UVALIDN ("Unweighted Valid N", F40.0)

  • UVARIANCE ("Unweighted Variance", F40.0)

Statistics Positions and Labels

/SLABELS
    [POSITION={COLUMN | ROW | LAYER}]
    [VISIBLE={YES | NO}]

The SLABELS subcommand controls the position and visibility of summary statistics for the TABLE subcommand that it follows.

POSITION sets the axis on which summary statistics appear. With POSITION=COLUMN, which is the default, each summary statistic appears in a column. For example:

CTABLES /TABLE=age [MEAN, MEDIAN] BY gender.

                    Custom Tables
+──────────────────────────+───────────────────────+
│                          │      S3a. GENDER:     │
│                          +───────────+───────────+
│                          │    Male   │   Female  │
│                          +────+──────+────+──────+
│                          │Mean│Median│Mean│Median│
+──────────────────────────+────+──────+────+──────+
│D1. AGE: What is your age?│  46│    45│  50│    52│
+──────────────────────────+────+──────+────+──────+

With POSITION=ROW, each summary statistic appears in a row, as shown below:

CTABLES /TABLE=age [MEAN, MEDIAN] BY gender /SLABELS POSITION=ROW.

                  Custom Tables
+─────────────────────────────────+─────────────+
│                                 │ S3a. GENDER:│
│                                 +─────+───────+
│                                 │ Male│ Female│
+─────────────────────────────────+─────+───────+
│D1. AGE: What is your age? Mean  │   46│     50│
│                           Median│   45│     52│
+─────────────────────────────────+─────+───────+

POSITION=LAYER is also available to place each summary statistic in a separate layer.

Labels for summary statistics are shown by default. Use VISIBLE=NO to suppress them. Because unlabeled data can cause confusion, it should only be considered if the meaning of the data is evident, as in a simple case like this:

CTABLES /TABLE=ageGroup [TABLEPCT] /SLABELS VISIBLE=NO.

         Custom Tables
+───────────────────────+─────+
│Age group 15 or younger│  .0%│
│          16 to 25     │15.7%│
│          26 to 35     │13.8%│
│          36 to 45     │14.8%│
│          46 to 55     │16.8%│
│          56 to 65     │17.8%│
│          66 or older  │21.1%│
+───────────────────────+─────+

Category Label Positions

/CLABELS {AUTO │ {ROWLABELS│COLLABELS}={OPPOSITE│LAYER}}

The CLABELS subcommand controls the position of category labels for the TABLE subcommand that it follows. By default, or if AUTO is specified, category labels for a given variable nest inside the variable's label on the same axis. For example, the command below results in age categories nesting within the age group variable on the rows axis and gender categories within the gender variable on the columns axis:

CTABLES /TABLE ageGroup BY gender.

             Custom Tables
+───────────────────────+────────────+
│                       │S3a. GENDER:│
│                       +─────+──────+
│                       │ Male│Female│
│                       +─────+──────+
│                       │Count│ Count│
+───────────────────────+─────+──────+
│Age group 15 or younger│    0│     0│
│          16 to 25     │  594│   505│
│          26 to 35     │  476│   491│
│          36 to 45     │  489│   548│
│          46 to 55     │  526│   649│
│          56 to 65     │  516│   731│
│          66 or older  │  531│   943│
+───────────────────────+─────+──────+

ROWLABELS=OPPOSITE or COLLABELS=OPPOSITE move row or column variable category labels, respectively, to the opposite axis. The setting affects only the innermost variable or variables, which must be categorical, on the given axis. For example:

CTABLES /TABLE ageGroup BY gender /CLABELS ROWLABELS=OPPOSITE.
CTABLES /TABLE ageGroup BY gender /CLABELS COLLABELS=OPPOSITE.

                                Custom Tables
+─────+──────────────────────────────────────────────────────────────────────
│     │                                      S3a. GENDER:
│     +───────────────────────────────────────────+──────────────────────────
│     │                    Male                   │                   Female
│     +───────+─────+─────+─────+─────+─────+─────+───────+─────+─────+─────+
│     │ 15 or │16 to│26 to│36 to│46 to│56 to│66 or│ 15 or │16 to│26 to│36 to│
│     │younger│  25 │  35 │  45 │  55 │  65 │older│younger│  25 │  35 │  45 │
│     +───────+─────+─────+─────+─────+─────+─────+───────+─────+─────+─────+
│     │ Count │Count│Count│Count│Count│Count│Count│ Count │Count│Count│Count│
+─────+───────+─────+─────+─────+─────+─────+─────+───────+─────+─────+─────+
│Age  │      0│  594│  476│  489│  526│  516│  531│      0│  505│  491│  548│
│group│       │     │     │     │     │     │     │       │     │     │     │
+─────+───────+─────+─────+─────+─────+─────+─────+───────+─────+─────+─────+

+─────+─────────────────+
│     │                 │
│     +─────────────────+
│     │                 │
│     +─────+─────+─────+
│     │46 to│56 to│66 or│
│     │  55 │  65 │older│
│     +─────+─────+─────+
│     │Count│Count│Count│
+─────+─────+─────+─────+
│Age  │  649│  731│  943│
│group│     │     │     │
+─────+─────+─────+─────+

                Custom Tables
+──────────────────────────────+────────────+
│                              │S3a. GENDER:│
│                              +────────────+
│                              │    Count   │
+──────────────────────────────+────────────+
│Age group 15 or younger Male  │           0│
│                        Female│           0│
│         ─────────────────────+────────────+
│          16 to 25      Male  │         594│
│                        Female│         505│
│         ─────────────────────+────────────+
│          26 to 35      Male  │         476│
│                        Female│         491│
│         ─────────────────────+────────────+
│          36 to 45      Male  │         489│
│                        Female│         548│
│         ─────────────────────+────────────+
│          46 to 55      Male  │         526│
│                        Female│         649│
│         ─────────────────────+────────────+
│          56 to 65      Male  │         516│
│                        Female│         731│
│         ─────────────────────+────────────+
│          66 or older   Male  │         531│
│                        Female│         943│
+──────────────────────────────+────────────+

ROWLABELS=LAYER or COLLABELS=LAYER move the innermost row or column variable category labels, respectively, to the layer axis.

Only one axis's labels may be moved, whether to the opposite axis or to the layer axis.

Effect on Summary Statistics

CLABELS primarily affects the appearance of tables, not the data displayed in them. However, CTABLES can affect the values displayed for statistics that summarize areas of a table, since it can change the definitions of these areas.

For example, consider the following syntax and output:

CTABLES /TABLE ageGroup BY gender [ROWPCT, COLPCT].

                     Custom Tables
+───────────────────────+─────────────────────────────+
│                       │         S3a. GENDER:        │
│                       +──────────────+──────────────+
│                       │     Male     │    Female    │
│                       +─────+────────+─────+────────+
│                       │Row %│Column %│Row %│Column %│
+───────────────────────+─────+────────+─────+────────+
│Age group 15 or younger│    .│     .0%│    .│     .0%│
│          16 to 25     │54.0%│   19.0%│46.0%│   13.1%│
│          26 to 35     │49.2%│   15.2%│50.8%│   12.7%│
│          36 to 45     │47.2%│   15.6%│52.8%│   14.2%│
│          46 to 55     │44.8%│   16.8%│55.2%│   16.8%│
│          56 to 65     │41.4%│   16.5%│58.6%│   18.9%│
│          66 or older  │36.0%│   17.0%│64.0%│   24.4%│
+───────────────────────+─────+────────+─────+────────+

Using COLLABELS=OPPOSITE changes the definitions of rows and columns, so that column percentages display what were previously row percentages and the new row percentages become meaningless (because there is only one cell per row):

CTABLES
    /TABLE ageGroup BY gender [ROWPCT, COLPCT]
    /CLABELS COLLABELS=OPPOSITE.

                  Custom Tables
+──────────────────────────────+───────────────+
│                              │  S3a. GENDER: │
│                              +──────+────────+
│                              │ Row %│Column %│
+──────────────────────────────+──────+────────+
│Age group 15 or younger Male  │     .│       .│
│                        Female│     .│       .│
│         ─────────────────────+──────+────────+
│          16 to 25      Male  │100.0%│   54.0%│
│                        Female│100.0%│   46.0%│
│         ─────────────────────+──────+────────+
│          26 to 35      Male  │100.0%│   49.2%│
│                        Female│100.0%│   50.8%│
│         ─────────────────────+──────+────────+
│          36 to 45      Male  │100.0%│   47.2%│
│                        Female│100.0%│   52.8%│
│         ─────────────────────+──────+────────+
│          46 to 55      Male  │100.0%│   44.8%│
│                        Female│100.0%│   55.2%│
│         ─────────────────────+──────+────────+
│          56 to 65      Male  │100.0%│   41.4%│
│                        Female│100.0%│   58.6%│
│         ─────────────────────+──────+────────+
│          66 or older   Male  │100.0%│   36.0%│
│                        Female│100.0%│   64.0%│
+──────────────────────────────+──────+────────+

Moving Categories for Stacked Variables

If CLABELS moves category labels from an axis with stacked variables, the variables that are moved must have the same category specifications (see Per-Variable Category Options) and the same value labels.

The following shows both moving stacked category variables and adapting to the changing definitions of rows and columns:

CTABLES /TABLE (likelihoodOfBeingStoppedByPolice
                + likelihoodOfHavingAnAccident) [COLPCT].
CTABLES /TABLE (likelihoodOfBeingStoppedByPolice
                + likelihoodOfHavingAnAccident) [ROWPCT]
  /CLABELS ROW=OPPOSITE.

                                 Custom Tables
+─────────────────────────────────────────────────────────────────────+───────+
│                                                                     │ Column│
│                                                                     │   %   │
+─────────────────────────────────────────────────────────────────────+───────+
│105b. How likely is it that drivers who have had too     Almost      │  10.2%│
│much to drink to drive safely will A. Get stopped by the certain     │       │
│police?                                                  Very likely │  21.8%│
│                                                         Somewhat    │  40.2%│
│                                                         likely      │       │
│                                                         Somewhat    │  19.0%│
│                                                         unlikely    │       │
│                                                         Very        │   8.9%│
│                                                         unlikely    │       │
+─────────────────────────────────────────────────────────────────────+───────+
│105b. How likely is it that drivers who have had too     Almost      │  15.9%│
│much to drink to drive safely will B. Have an accident?  certain     │       │
│                                                         Very likely │  40.8%│
│                                                         Somewhat    │  35.0%│
│                                                         likely      │       │
│                                                         Somewhat    │   6.2%│
│                                                         unlikely    │       │
│                                                         Very        │   2.0%│
│                                                         unlikely    │       │
+─────────────────────────────────────────────────────────────────────+───────+

                                 Custom Tables
+─────────────────────────────+────────+───────+─────────+──────────+─────────+
│                             │ Almost │  Very │ Somewhat│ Somewhat │   Very  │
│                             │ certain│ likely│  likely │ unlikely │ unlikely│
│                             +────────+───────+─────────+──────────+─────────+
│                             │  Row % │ Row % │  Row %  │   Row %  │  Row %  │
+─────────────────────────────+────────+───────+─────────+──────────+─────────+
│105b. How likely is it that  │   10.2%│  21.8%│    40.2%│     19.0%│     8.9%│
│drivers who have had too much│        │       │         │          │         │
│to drink to drive safely will│        │       │         │          │         │
│A. Get stopped by the police?│        │       │         │          │         │
│105b. How likely is it that  │   15.9%│  40.8%│    35.0%│      6.2%│     2.0%│
│drivers who have had too much│        │       │         │          │         │
│to drink to drive safely will│        │       │         │          │         │
│B. Have an accident?         │        │       │         │          │         │
+─────────────────────────────+────────+───────+─────────+──────────+─────────+

Per-Variable Category Options

/CATEGORIES VARIABLES=variables
    {[value, value...]
   | [ORDER={A | D}]
     [KEY={VALUE | LABEL | summary(variable)}]
     [MISSING={EXCLUDE | INCLUDE}]}
    [TOTAL={NO | YES} [LABEL=string] [POSITION={AFTER | BEFORE}]]
    [EMPTY={INCLUDE | EXCLUDE}]

The CATEGORIES subcommand specifies, for one or more categorical variables, the categories to include and exclude, the sort order for included categories, and treatment of missing values. It also controls the totals and subtotals to display. It may be specified any number of times, each time for a different set of variables. CATEGORIES applies to the table produced by the TABLE subcommand that it follows.

CATEGORIES does not apply to scalar variables.

VARIABLES is required and must list the variables for the subcommand to affect.

The syntax may specify the categories to include and their sort order either explicitly or implicitly. The following sections give the details of each form of syntax, followed by information on totals and subtotals and the EMPTY setting.

Explicit Categories

To use CTABLES to explicitly specify categories to include, list the categories within square brackets in the desired sort order. Use spaces or commas to separate values. Categories not covered by the list are excluded from analysis.

Each element of the list takes one of the following forms:

  • number
    'string'
    A numeric or string category value, for variables that have the corresponding type.

  • 'date'
    'time'
    A date or time category value, for variables that have a date or time print format.

  • min THRU max
    LO THRU max
    min THRU HI
    A range of category values, where min and max each takes one of the forms above, in increasing order.

  • MISSING
    All user-missing values. (To match individual user-missing values, specify their category values.)

  • OTHERNM
    Any non-missing value not covered by any other element of the list (regardless of where OTHERNM is placed in the list).

  • &postcompute
    A computed category name.

  • SUBTOTAL
    HSUBTOTAL
    A subtotal.

If multiple elements of the list cover a given category, the last one in the list takes precedence.

The following example syntax and output show how an explicit category can limit the displayed categories:

CTABLES /TABLE freqOfDriving.
CTABLES /TABLE freqOfDriving /CATEGORIES VARIABLES=freqOfDriving [1, 2, 3].

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│ 1. How often do you usually drive a car or other  Every day           │ 4667│
│motor vehicle?                                     Several days a week │ 1274│
│                                                   Once a week or less │  361│
│                                                   Only certain times a│  130│
│                                                   year                │     │
│                                                   Never               │  540│
+───────────────────────────────────────────────────────────────────────+─────+

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│ 1. How often do you usually drive a car or other     Every day        │ 4667│
│motor vehicle?                                        Several days a   │ 1274│
│                                                      week             │     │
│                                                      Once a week or   │  361│
│                                                      less             │     │
+───────────────────────────────────────────────────────────────────────+─────+

Implicit Categories

In the absence of an explicit list of categories, CATEGORIES allows KEY, ORDER, and MISSING to specify how to select and sort categories.

The KEY setting specifies the sort key. By default, or with KEY=VALUE, categories are sorted by default. Categories may also be sorted by value label, with KEY=LABEL, or by the value of a summary function, e.g. KEY=COUNT.

By default, or with ORDER=A, categories are sorted in ascending order. Specify ORDER=D to sort in descending order.

User-missing values are excluded by default, or with MISSING=EXCLUDE. Specify MISSING=INCLUDE to include user-missing values. The system-missing value is always excluded.

The following example syntax and output show how MISSING=INCLUDE causes missing values to be included in a category list.

CTABLES /TABLE freqOfDriving.
CTABLES /TABLE freqOfDriving
        /CATEGORIES VARIABLES=freqOfDriving MISSING=INCLUDE.

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│ 1. How often do you usually drive a car or other  Every day           │ 4667│
│motor vehicle?                                     Several days a week │ 1274│
│                                                   Once a week or less │  361│
│                                                   Only certain times a│  130│
│                                                   year                │     │
│                                                   Never               │  540│
+───────────────────────────────────────────────────────────────────────+─────+

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│ 1. How often do you usually drive a car or other  Every day           │ 4667│
│motor vehicle?                                     Several days a week │ 1274│
│                                                   Once a week or less │  361│
│                                                   Only certain times a│  130│
│                                                   year                │     │
│                                                   Never               │  540│
│                                                   Don't know          │    8│
│                                                   Refused             │   19│
+───────────────────────────────────────────────────────────────────────+─────+

Totals and Subtotals

CATEGORIES also controls display of totals and subtotals. By default, or with TOTAL=NO, totals are not displayed. Use TOTAL=YES to display a total. By default, the total is labeled "Total"; use LABEL="label" to override it.

Subtotals are also not displayed by default. To add one or more subtotals, use an explicit category list and insert SUBTOTAL or HSUBTOTAL in the position or positions where the subtotal should appear. The subtotal becomes an extra row or column or layer. HSUBTOTAL additionally hides the categories that make up the subtotal. Either way, the default label is "Subtotal", use SUBTOTAL="label" or HSUBTOTAL="label" to specify a custom label.

The following example syntax and output show how to use TOTAL=YES and SUBTOTAL:

CTABLES
    /TABLE freqOfDriving
    /CATEGORIES VARIABLES=freqOfDriving [OTHERNM, SUBTOTAL='Valid Total',
                                         MISSING, SUBTOTAL='Missing Total']
                                        TOTAL=YES LABEL='Overall Total'.

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│ 1. How often do you usually drive a car or other  Every day           │ 4667│
│motor vehicle?                                     Several days a week │ 1274│
│                                                   Once a week or less │  361│
│                                                   Only certain times a│  130│
│                                                   year                │     │
│                                                   Never               │  540│
│                                                   Valid Total         │ 6972│
│                                                   Don't know          │    8│
│                                                   Refused             │   19│
│                                                   Missing Total       │   27│
│                                                   Overall Total       │ 6999│
+───────────────────────────────────────────────────────────────────────+─────+

By default, or with POSITION=AFTER, totals are displayed in the output after the last category and subtotals apply to categories that precede them. With POSITION=BEFORE, totals come before the first category and subtotals apply to categories that follow them.

Only categorical variables may have totals and subtotals. Scalar variables may be "totaled" indirectly by enabling totals and subtotals on a categorical variable within which the scalar variable is summarized. For example, the following syntax produces a mean, count, and valid count across all data by adding a total on the categorical region variable, as shown:

CTABLES /TABLE=region > monthDaysMin1drink [MEAN, VALIDN]
    /CATEGORIES VARIABLES=region TOTAL=YES LABEL='All regions'.

                                 Custom Tables
+───────────────────────────────────────────────────────────+────+─────+──────+
│                                                           │    │     │ Valid│
│                                                           │Mean│Count│   N  │
+───────────────────────────────────────────────────────────+────+─────+──────+
│20. On how many of the thirty days in this  Region NE      │ 5.6│ 1409│   945│
│typical month did you have one or more             MW      │ 5.0│ 1654│  1026│
│alcoholic beverages to drink?                      S       │ 6.0│ 2390│  1285│
│                                                   W       │ 6.5│ 1546│   953│
│                                                   All     │ 5.8│ 6999│  4209│
│                                                   regions │    │     │      │
+───────────────────────────────────────────────────────────+────+─────+──────+

By default, PSPP uses the same summary functions for totals and subtotals as other categories. To summarize totals and subtotals differently, specify the summary functions for totals and subtotals after the ordinary summary functions inside a nested set of [] following TOTALS. For example, the following syntax displays COUNT for individual categories and totals and VALIDN for totals, as shown:

CTABLES
    /TABLE isLicensedDriver [COUNT, TOTALS[COUNT, VALIDN]]
    /CATEGORIES VARIABLES=isLicensedDriver TOTAL=YES MISSING=INCLUDE.

                                 Custom Tables
+────────────────────────────────────────────────────────────────+─────+──────+
│                                                                │     │ Valid│
│                                                                │Count│   N  │
+────────────────────────────────────────────────────────────────+─────+──────+
│D7a. Are you a licensed driver; that is, do you have a Yes      │ 6379│      │
│valid driver's license?                                No       │  572│      │
│                                                       Don't    │    4│      │
│                                                       know     │     │      │
│                                                       Refused  │   44│      │
│                                                       Total    │ 6999│  6951│
+────────────────────────────────────────────────────────────────+─────+──────+

Categories Without Values

Some categories might not be included in the data set being analyzed. For example, our example data set has no cases in the "15 or younger" age group. By default, or with EMPTY=INCLUDE, PSPP includes these empty categories in output tables. To exclude them, specify EMPTY=EXCLUDE.

For implicit categories, empty categories potentially include all the values with value labels for a given variable; for explicit categories, they include all the values listed individually and all values with value labels that are covered by ranges or MISSING or OTHERNM.

The following example syntax and output show the effect of EMPTY=EXCLUDE for the membersOver16 variable, in which 0 is labeled "None" but no cases exist with that value:

CTABLES /TABLE=membersOver16.
CTABLES /TABLE=membersOver16 /CATEGORIES VARIABLES=membersOver16 EMPTY=EXCLUDE.

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│S1. Including yourself, how many members of this household are None    │    0│
│age 16 or older?                                               1       │ 1586│
│                                                               2       │ 3031│
│                                                               3       │  505│
│                                                               4       │  194│
│                                                               5       │   55│
│                                                               6 or    │   21│
│                                                               more    │     │
+───────────────────────────────────────────────────────────────────────+─────+

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│S1. Including yourself, how many members of this household are 1       │ 1586│
│age 16 or older?                                               2       │ 3031│
│                                                               3       │  505│
│                                                               4       │  194│
│                                                               5       │   55│
│                                                               6 or    │   21│
│                                                               more    │     │
+───────────────────────────────────────────────────────────────────────+─────+

Titles

/TITLES
    [TITLE=string...]
    [CAPTION=string...]
    [CORNER=string...]

The TITLES subcommand sets the title, caption, and corner text for the table output for the previous TABLE subcommand. Any number of strings may be specified for each kind of text, with each string appearing on a separate line in the output. The title appears above the table, the caption below the table, and the corner text appears in the table's upper left corner. By default, the title is "Custom Tables" and the caption and corner text are empty. With some table output styles, the corner text is not displayed.

The strings provided in this subcommand may contain the following macro-like keywords that PSPP substitutes at the time that it runs the command:

  • )DATE
    The current date, e.g. MM/DD/YY. The format is locale-dependent.

  • )TIME
    The current time, e.g. HH:MM:SS. The format is locale-dependent.

  • )TABLE
    The expression specified on the TABLE command. Summary and measurement level specifications are omitted, and variable labels are used in place of variable names.

Table Formatting

/FORMAT
    [MINCOLWIDTH={DEFAULT | width}]
    [MAXCOLWIDTH={DEFAULT | width}]
    [UNITS={POINTS | INCHES | CM}]
    [EMPTY={ZERO | BLANK | string}]
    [MISSING=string]

The FORMAT subcommand, which must precede the first TABLE subcommand, controls formatting for all the output tables. FORMAT and all of its settings are optional.

Use MINCOLWIDTH and MAXCOLWIDTH to control the minimum or maximum width of columns in output tables. By default, with DEFAULT, column width varies based on content. Otherwise, specify a number for either or both of these settings. If both are specified, MAXCOLWIDTH must be greater than or equal to MINCOLWIDTH. The default unit, or with UNITS=POINTS, is points (1/72 inch), or specify UNITS=INCHES to use inches or UNITS=CM for centimeters. PSPP does not currently honor any of these settings.

By default, or with EMPTY=ZERO, zero values are displayed in their usual format. Use EMPTY=BLANK to use an empty cell instead, or EMPTY="string" to use the specified string.

By default, missing values are displayed as ., the same as in other tables. Specify MISSING="string" to instead use a custom string.

Display of Variable Labels

/VLABELS
    VARIABLES=variables
    DISPLAY={DEFAULT | NAME | LABEL | BOTH | NONE}

The VLABELS subcommand, which must precede the first TABLE subcommand, controls display of variable labels in all the output tables. VLABELS is optional. It may appear multiple times to adjust settings for different variables.

VARIABLES and DISPLAY are required. The value of DISPLAY controls how variable labels are displayed for the variables listed on VARIABLES. The supported values are:

  • DEFAULT
    Use the setting from SET TVARS).

  • NAME
    Show only a variable name.

  • LABEL
    Show only a variable label.

  • BOTH
    Show variable name and label.

  • NONE
    Show nothing.

Missing Value Treatment

The TABLE subcommand on CTABLES specifies two different kinds of variables: variables that divide tables into cells (which are always categorical) and variables being summarized (which may be categorical or scale). PSPP treats missing values differently in each kind of variable, as described in the sections below.

Missing Values for Cell-Defining Variables

For variables that divide tables into cells, per-variable category options, as described in Per-Variable Category Options, determine which data is analyzed. If any of the categories for such a variable would exclude a case, then that case is not included.

As an example, consider the following entirely artificial dataset, in which x and y are categorical variables with missing value 9, and z is scale:

   Data List
+─+─+─────────+
│x│y│    z    │
+─+─+─────────+
│1│1│        1│
│1│2│       10│
│1│9│      100│
│2│1│     1000│
│2│2│    10000│
│2│9│   100000│
│9│1│  1000000│
│9│2│ 10000000│
│9│9│100000000│
+─+─+─────────+

Using x and y to define cells, and summarizing z, by default PSPP omits all the cases that have x or y (or both) missing:

CTABLES /TABLE x > y > z [SUM].

  Custom Tables
+─────────+─────+
│         │ Sum │
+─────────+─────+
│x 1 y 1 z│    1│
│     ────+─────+
│      2 z│   10│
│ ────────+─────+
│  2 y 1 z│ 1000│
│     ────+─────+
│      2 z│10000│
+─────────+─────+

If, however, we add CATEGORIES specifications to include missing values for y or for x and y, the output table includes them, like so:

CTABLES /TABLE x > y > z [SUM] /CATEGORIES VARIABLES=y MISSING=INCLUDE.
CTABLES /TABLE x > y > z [SUM] /CATEGORIES VARIABLES=x y MISSING=INCLUDE.

   Custom Tables
+─────────+──────+
│         │  Sum │
+─────────+──────+
│x 1 y 1 z│     1│
│     ────+──────+
│      2 z│    10│
│     ────+──────+
│      9 z│   100│
│ ────────+──────+
│  2 y 1 z│  1000│
│     ────+──────+
│      2 z│ 10000│
│     ────+──────+
│      9 z│100000│
+─────────+──────+

    Custom Tables
+─────────+─────────+
│         │   Sum   │
+─────────+─────────+
│x 1 y 1 z│        1│
│     ────+─────────+
│      2 z│       10│
│     ────+─────────+
│      9 z│      100│
│ ────────+─────────+
│  2 y 1 z│     1000│
│     ────+─────────+
│      2 z│    10000│
│     ────+─────────+
│      9 z│   100000│
│ ────────+─────────+
│  9 y 1 z│  1000000│
│     ────+─────────+
│      2 z│ 10000000│
│     ────+─────────+
│      9 z│100000000│
+─────────+─────────+

Missing Values for Summary Variables

For summary variables, values that are valid and in included categories are analyzed, and values that are missing or in excluded categories are not analyzed, with the following exceptions:

  • The VALIDN summary functions (VALIDN, EVALIDN, UVALIDN, areaPCT.VALIDN, and UareaPCT.VALIDN) only count valid values in included categories (not missing values in included categories).

  • The TOTALN summary functions (TOTALN, ETOTALN, UTOTALN, areaPCT.TOTALN), and UareaPCT.TOTALN count all values (valid and missing) in included categories and missing (but not valid) values in excluded categories.

For categorical variables, system-missing values are never in included categories. For scale variables, there is no notion of included and excluded categories, so all values are effectively included.

The following table provides another view of the above rules:

VALIDNotherTOTALN
Categorical variables:
   Valid values in included categoriesyesyesyes
   Missing values in included categories--yesyes
   Missing values in excluded categories----yes
   Valid values in excluded categories------
Scale variables:
   Valid valuesyesyesyes
   User- or system-missing values--yesyes

Scale Missing Values

/SMISSING {VARIABLE | LISTWISE}

The SMISSING subcommand, which must precede the first TABLE subcommand, controls treatment of missing values for scalar variables in producing all the output tables. SMISSING is optional.

With SMISSING=VARIABLE, which is the default, missing values are excluded on a variable-by-variable basis. With SMISSING=LISTWISE, when stacked scalar variables are nested together with a categorical variable, a missing value for any of the scalar variables causes the case to be excluded for all of them.

As an example, consider the following dataset, in which x is a categorical variable and y and z are scale:

   Data List
+─+─────+─────+
│x│  y  │  z  │
+─+─────+─────+
│1│    .│40.00│
│1│10.00│50.00│
│1│20.00│60.00│
│1│30.00│    .│
+─+─────+─────+

With the default missing-value treatment, x's mean is 20, based on the values 10, 20, and 30, and y's mean is 50, based on 40, 50, and 60:

CTABLES /TABLE (y + z) > x.

Custom Tables
+─────+─────+
│     │ Mean│
+─────+─────+
│y x 1│20.00│
+─────+─────+
│z x 1│50.00│
+─────+─────+

By adding SMISSING=LISTWISE, only cases where y and z are both non-missing are considered, so x's mean becomes 15, as the average of 10 and 20, and y's mean becomes 55, the average of 50 and 60:

CTABLES /SMISSING LISTWISE /TABLE (y + z) > x.

Custom Tables
+─────+─────+
│     │ Mean│
+─────+─────+
│y x 1│15.00│
+─────+─────+
│z x 1│55.00│
+─────+─────+

Even with SMISSING=LISTWISE, if y and z are separately nested with x, instead of using a single > operator, missing values revert to being considered on a variable-by-variable basis:

CTABLES /SMISSING LISTWISE /TABLE (y > x) + (z > x).

Custom Tables
+─────+─────+
│     │ Mean│
+─────+─────+
│y x 1│20.00│
+─────+─────+
│z x 1│50.00│
+─────+─────+

Computed Categories

/PCOMPUTE &postcompute=EXPR(expression)
/PPROPERTIES &postcompute...
    [LABEL=string]
    [FORMAT=[summary format]...]
    [HIDESOURCECATS={NO | YES}

"Computed categories", also called "postcomputes", are categories created using arithmetic on categories obtained from the data. The PCOMPUTE subcommand creates a postcompute, which may then be used on CATEGORIES within an explicit category list. Optionally, PPROPERTIES refines how a postcompute is displayed. The following sections provide the details.

PCOMPUTE

/PCOMPUTE &postcompute=EXPR(expression)

The PCOMPUTE subcommand, which must precede the first TABLE command, defines computed categories. It is optional and may be used any number of times to define multiple postcomputes.

Each PCOMPUTE defines one postcompute. Its syntax consists of a name to identify the postcompute as a PSPP identifier prefixed by &, followed by = and a postcompute expression enclosed in EXPR(...). A postcompute expression consists of:

  • [category]
    This form evaluates to the summary statistic for category, e.g. [1] evaluates to the value of the summary statistic associated with category 1. The category may be a number, a quoted string, or a quoted time or date value. All of the categories for a given postcompute must have the same form. The category must appear in all the CATEGORIES list in which the postcompute is used.

  • [min THRU max]
    [LO THRU max]
    [min THRU HI]
    MISSING
    OTHERNM
    These forms evaluate to the summary statistics for a category specified with the same syntax, as described in a previous section (see Explicit Category List). The category must appear in all the CATEGORIES list in which the postcompute is used.

  • SUBTOTAL
    The summary statistic for the subtotal category. This form is allowed only if the CATEGORIES lists that include this postcompute have exactly one subtotal.

  • SUBTOTAL[index]
    The summary statistic for subtotal category index, where 1 is the first subtotal, 2 is the second, and so on. This form may be used for CATEGORIES lists with any number of subtotals.

  • TOTAL
    The summary statistic for the total. The CATEGORIES lsits that include this postcompute must have a total enabled.

  • a + b
    a - b
    a * b
    a / b
    a ** b These forms perform arithmetic on the values of postcompute expressions a and b. The usual operator precedence rules apply.

  • number
    Numeric constants may be used in postcompute expressions.

  • (a)
    Parentheses override operator precedence.

A postcompute is not associated with any particular variable. Instead, it may be referenced within CATEGORIES for any suitable variable (e.g. only a string variable is suitable for a postcompute expression that refers to a string category, only a variable with subtotals for an expression that refers to subtotals, ...).

Normally a named postcompute is defined only once, but if a later PCOMPUTE redefines a postcompute with the same name as an earlier one, the later one take precedence.

The following syntax and output shows how PCOMPUTE can compute a total over subtotals, summing the "Frequent Drivers" and "Infrequent Drivers" subtotals to form an "All Drivers" postcompute. It also shows how to calculate and display a percentage, in this case the percentage of valid responses that report never driving. It uses PPROPERTIES to display the latter in PCT format.

CTABLES
    /PCOMPUTE &all_drivers=EXPR([1 THRU 2] + [3 THRU 4])
    /PPROPERTIES &all_drivers LABEL='All Drivers'
    /PCOMPUTE &pct_never=EXPR([5] / ([1 THRU 2] + [3 THRU 4] + [5]) * 100)
    /PPROPERTIES &pct_never LABEL='% Not Drivers' FORMAT=COUNT PCT40.1
    /TABLE=freqOfDriving BY gender
    /CATEGORIES VARIABLES=freqOfDriving
                             [1 THRU 2, SUBTOTAL='Frequent Drivers',
                              3 THRU 4, SUBTOTAL='Infrequent Drivers',
                              &all_drivers, 5, &pct_never,
                              MISSING, SUBTOTAL='Not Drivers or Missing'].

                                 Custom Tables
+────────────────────────────────────────────────────────────────+────────────+
│                                                                │S3a. GENDER:│
│                                                                +─────+──────+
│                                                                │ Male│Female│
│                                                                +─────+──────+
│                                                                │Count│ Count│
+────────────────────────────────────────────────────────────────+─────+──────+
│ 1. How often do you usually drive a car or Every day           │ 2305│  2362│
│other motor vehicle?                        Several days a week │  440│   834│
│                                            Frequent Drivers    │ 2745│  3196│
│                                            Once a week or less │  125│   236│
│                                            Only certain times a│   58│    72│
│                                            year                │     │      │
│                                            Infrequent Drivers  │  183│   308│
│                                            All Drivers         │ 2928│  3504│
│                                            Never               │  192│   348│
│                                            % Not Drivers       │ 6.2%│  9.0%│
│                                            Don't know          │    3│     5│
│                                            Refused             │    9│    10│
│                                            Not Drivers or      │  204│   363│
│                                            Missing             │     │      │
+────────────────────────────────────────────────────────────────+─────+──────+

PPROPERTIES

/PPROPERTIES &postcompute...
    [LABEL=string]
    [FORMAT=[summary format]...]
    [HIDESOURCECATS={NO | YES}

The PPROPERTIES subcommand, which must appear before TABLE, sets properties for one or more postcomputes defined on prior PCOMPUTE subcommands. The subcommand syntax begins with the list of postcomputes, each prefixed with & as specified on PCOMPUTE.

All of the settings on PPROPERTIES are optional. Use LABEL to set the label shown for the postcomputes in table output. The default label for a postcompute is the expression used to define it.

A postcompute always uses same summary functions as the variable whose categories contain it, but FORMAT allows control over the format used to display their values. It takes a list of summary function names and format specifiers.

By default, or with HIDESOURCECATS=NO, categories referred to by computed categories are displayed like other categories. Use HIDESOURCECATS=YES to hide them.

The previous section provides an example for PPROPERTIES.

Effective Weight

/WEIGHT VARIABLE=variable

The WEIGHT subcommand is optional and must appear before TABLE. If it appears, it must name a numeric variable, known as the "effective weight" or "adjustment weight". The effective weight variable stands in for the dictionary's weight variable, if any, in most calculations in CTABLES. The only exceptions are the COUNT, TOTALN, and VALIDN summary functions, which use the dictionary weight instead.

Weights obtained from the PSPP dictionary are rounded to the nearest integer at the case level. Effective weights are not rounded. Regardless of the weighting source, PSPP does not analyze cases with zero, missing, or negative effective weights.

Hiding Small Counts

/HIDESMALLCOUNTS COUNT=count

The HIDESMALLCOUNTS subcommand is optional. If it specified, then COUNT, ECOUNT, and UCOUNT values in output tables less than the value of count are shown as <count instead of their true values. The value of count must be an integer and must be at least 2.

The following syntax and example shows how to use HIDESMALLCOUNTS:

CTABLES /HIDESMALLCOUNTS COUNT=10 /TABLE placeOfLastDrinkBeforeDrive.

                                 Custom Tables
+───────────────────────────────────────────────────────────────────────+─────+
│                                                                       │Count│
+───────────────────────────────────────────────────────────────────────+─────+
│37. Please think about the most recent occasion that   Other (list)    │<10  │
│you drove within two hours of drinking alcoholic       Your home       │  182│
│beverages. Where did you drink on that occasion?       Friend's home   │  264│
│                                                       Bar/Tavern/Club │  279│
│                                                       Restaurant      │  495│
│                                                       Work            │   21│
│                                                       Bowling alley   │<10  │
│                                                       Hotel/Motel     │<10  │
│                                                       Country Club/   │   17│
│                                                       Golf course     │     │
│                                                       Drank in the    │<10  │
│                                                       car/On the road │     │
│                                                       Sporting event  │   15│
│                                                       Movie theater   │<10  │
│                                                       Shopping/Store/ │<10  │
│                                                       Grocery store   │     │
│                                                       Wedding         │   15│
│                                                       Party at someone│   81│
│                                                       else's home     │     │
│                                                       Park/picnic     │   14│
│                                                       Party at your   │<10  │
│                                                       house           │     │
+───────────────────────────────────────────────────────────────────────+─────+

  1. This is not necessarily a meaningful table. To make it easier to read, short variable labels are used. ↩

FACTOR

FACTOR  {
         VARIABLES=VAR_LIST,
         MATRIX IN ({CORR,COV}={*,FILE_SPEC})
        }

        [ /METHOD = {CORRELATION, COVARIANCE} ]

        [ /ANALYSIS=VAR_LIST ]

        [ /EXTRACTION={PC, PAF}]

        [ /ROTATION={VARIMAX, EQUAMAX, QUARTIMAX, PROMAX[(K)], NOROTATE}]

        [ /PRINT=[INITIAL] [EXTRACTION] [ROTATION] [UNIVARIATE] [CORRELATION] [COVARIANCE] [DET] [KMO] [AIC] [SIG] [ALL] [DEFAULT] ]

        [ /PLOT=[EIGEN] ]

        [ /FORMAT=[SORT] [BLANK(N)] [DEFAULT] ]

        [ /CRITERIA=[FACTORS(N)] [MINEIGEN(L)] [ITERATE(M)] [ECONVERGE (DELTA)] [DEFAULT] ]

        [ /MISSING=[{LISTWISE, PAIRWISE}] [{INCLUDE, EXCLUDE}] ]

The FACTOR command performs Factor Analysis or Principal Axis Factoring on a dataset. It may be used to find common factors in the data or for data reduction purposes.

The VARIABLES subcommand is required (unless the MATRIX IN subcommand is used). It lists the variables which are to partake in the analysis. (The ANALYSIS subcommand may optionally further limit the variables that participate; it is useful primarily in conjunction with MATRIX IN.)

If MATRIX IN instead of VARIABLES is specified, then the analysis is performed on a pre-prepared correlation or covariance matrix file instead of on individual data cases. Typically the matrix file will have been generated by MATRIX DATA or provided by a third party. If specified, MATRIX IN must be followed by COV or CORR, then by = and FILE_SPEC all in parentheses. FILE_SPEC may either be an asterisk, which indicates the currently loaded dataset, or it may be a file name to be loaded. See MATRIX DATA, for the expected format of the file.

The /EXTRACTION subcommand is used to specify the way in which factors (components) are extracted from the data. If PC is specified, then Principal Components Analysis is used. If PAF is specified, then Principal Axis Factoring is used. By default Principal Components Analysis is used.

The /ROTATION subcommand is used to specify the method by which the extracted solution is rotated. Three orthogonal rotation methods are available: VARIMAX (which is the default), EQUAMAX, and QUARTIMAX. There is one oblique rotation method, viz: PROMAX. Optionally you may enter the power of the promax rotation K, which must be enclosed in parentheses. The default value of K is 5. If you don't want any rotation to be performed, the word NOROTATE prevents the command from performing any rotation on the data.

The /METHOD subcommand should be used to determine whether the covariance matrix or the correlation matrix of the data is to be analysed. By default, the correlation matrix is analysed.

The /PRINT subcommand may be used to select which features of the analysis are reported:

  • UNIVARIATE A table of mean values, standard deviations and total weights are printed.
  • INITIAL Initial communalities and eigenvalues are printed.
  • EXTRACTION Extracted communalities and eigenvalues are printed.
  • ROTATION Rotated communalities and eigenvalues are printed.
  • CORRELATION The correlation matrix is printed.
  • COVARIANCE The covariance matrix is printed.
  • DET The determinant of the correlation or covariance matrix is printed.
  • AIC The anti-image covariance and anti-image correlation matrices are printed.
  • KMO The Kaiser-Meyer-Olkin measure of sampling adequacy and the Bartlett test of sphericity is printed.
  • SIG The significance of the elements of correlation matrix is printed.
  • ALL All of the above are printed.
  • DEFAULT Identical to INITIAL and EXTRACTION.

If /PLOT=EIGEN is given, then a "Scree" plot of the eigenvalues is printed. This can be useful for visualizing the factors and deciding which factors (components) should be retained.

The /FORMAT subcommand determined how data are to be displayed in loading matrices. If SORT is specified, then the variables are sorted in descending order of significance. If BLANK(N) is specified, then coefficients whose absolute value is less than N are not printed. If the keyword DEFAULT is specified, or if no /FORMAT subcommand is specified, then no sorting is performed, and all coefficients are printed.

You can use the /CRITERIA subcommand to specify how the number of extracted factors (components) are chosen. If FACTORS(N) is specified, where N is an integer, then N factors are extracted. Otherwise, the MINEIGEN setting is used. MINEIGEN(L) requests that all factors whose eigenvalues are greater than or equal to L are extracted. The default value of L is 1. The ECONVERGE setting has effect only when using iterative algorithms for factor extraction (such as Principal Axis Factoring). ECONVERGE(DELTA) specifies that iteration should cease when the maximum absolute value of the communality estimate between one iteration and the previous is less than DELTA. The default value of DELTA is 0.001.

The ITERATE(M) may appear any number of times and is used for two different purposes. It is used to set the maximum number of iterations (M) for convergence and also to set the maximum number of iterations for rotation. Whether it affects convergence or rotation depends upon which subcommand follows the ITERATE subcommand. If EXTRACTION follows, it affects convergence. If ROTATION follows, it affects rotation. If neither ROTATION nor EXTRACTION follow a ITERATE subcommand, then the entire subcommand is ignored. The default value of M is 25.

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set, then user-missing values are included in the calculations, but system-missing values are not. If EXCLUDE is set, which is the default, user-missing values are excluded as well as system-missing values. This is the default. If LISTWISE is set, then the entire case is excluded from analysis whenever any variable specified in the VARIABLES subcommand contains a missing value.

If PAIRWISE is set, then a case is considered missing only if either of the values for the particular coefficient are missing. The default is LISTWISE.

GLM

GLM DEPENDENT_VARS BY FIXED_FACTORS
     [/METHOD = SSTYPE(TYPE)]
     [/DESIGN = INTERACTION_0 [INTERACTION_1 [... INTERACTION_N]]]
     [/INTERCEPT = {INCLUDE|EXCLUDE}]
     [/MISSING = {INCLUDE|EXCLUDE}]

The GLM procedure can be used for fixed effects factorial Anova.

The DEPENDENT_VARS are the variables to be analysed. You may analyse several variables in the same command in which case they should all appear before the BY keyword.

The FIXED_FACTORS list must be one or more categorical variables. Normally it does not make sense to enter a scalar variable in the FIXED_FACTORS and doing so may cause PSPP to do a lot of unnecessary processing.

The METHOD subcommand is used to change the method for producing the sums of squares. Available values of TYPE are 1, 2 and 3. The default is type 3.

You may specify a custom design using the DESIGN subcommand. The design comprises a list of interactions where each interaction is a list of variables separated by a *. For example the command

GLM subject BY sex age_group race
    /DESIGN = age_group sex group age_group*sex age_group*race

specifies the model

subject = age_group + sex + race + age_group×sex + age_group×race

If no DESIGN subcommand is specified, then the default is all possible combinations of the fixed factors. That is to say

GLM subject BY sex age_group race

implies the model

subject = age_group + sex + race + age_group×sex + age_group×race + sex×race + age_group×sex×race

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set then, for the purposes of GLM analysis, only system-missing values are considered to be missing; user-missing values are not regarded as missing. If EXCLUDE is set, which is the default, then user-missing values are considered to be missing as well as system-missing values. A case for which any dependent variable or any factor variable has a missing value is excluded from the analysis.

LOGISTIC REGRESSION

LOGISTIC REGRESSION [VARIABLES =] DEPENDENT_VAR WITH PREDICTORS
     [/CATEGORICAL = CATEGORICAL_PREDICTORS]
     [{/NOCONST | /ORIGIN | /NOORIGIN }]
     [/PRINT = [SUMMARY] [DEFAULT] [CI(CONFIDENCE)] [ALL]]
     [/CRITERIA = [BCON(MIN_DELTA)] [ITERATE(MAX_INTERATIONS)]
                  [LCON(MIN_LIKELIHOOD_DELTA)] [EPS(MIN_EPSILON)]
                  [CUT(CUT_POINT)]]
     [/MISSING = {INCLUDE|EXCLUDE}]

Bivariate Logistic Regression is used when you want to explain a dichotomous dependent variable in terms of one or more predictor variables.

The minimum command is

LOGISTIC REGRESSION y WITH x1 x2 ... xN.

Here, y is the dependent variable, which must be dichotomous and x1 through xN are the predictor variables whose coefficients the procedure estimates.

By default, a constant term is included in the model. Hence, the full model is $${\bf y} = b_0 + b_1 {\bf x_1} + b_2 {\bf x_2} + \dots + b_n {\bf x_n}.$$

Predictor variables which are categorical in nature should be listed on the /CATEGORICAL subcommand. Simple variables as well as interactions between variables may be listed here.

If you want a model without the constant term b_0, use the keyword /ORIGIN. /NOCONST is a synonym for /ORIGIN.

An iterative Newton-Raphson procedure is used to fit the model. The /CRITERIA subcommand is used to specify the stopping criteria of the procedure, and other parameters. The value of CUT_POINT is used in the classification table. It is the threshold above which predicted values are considered to be 1. Values of CUT_POINT must lie in the range [0,1]. During iterations, if any one of the stopping criteria are satisfied, the procedure is considered complete. The stopping criteria are:

  • The number of iterations exceeds MAX_ITERATIONS. The default value of MAX_ITERATIONS is 20.
  • The change in the all coefficient estimates are less than MIN_DELTA. The default value of MIN_DELTA is 0.001.
  • The magnitude of change in the likelihood estimate is less than MIN_LIKELIHOOD_DELTA. The default value of MIN_LIKELIHOOD_DELTA is zero. This means that this criterion is disabled.
  • The differential of the estimated probability for all cases is less than MIN_EPSILON. In other words, the probabilities are close to zero or one. The default value of MIN_EPSILON is 0.00000001.

The PRINT subcommand controls the display of optional statistics. Currently there is one such option, CI, which indicates that the confidence interval of the odds ratio should be displayed as well as its value. CI should be followed by an integer in parentheses, to indicate the confidence level of the desired confidence interval.

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set, then user-missing values are included in the calculations, but system-missing values are not. If EXCLUDE is set, which is the default, user-missing values are excluded as well as system-missing values. This is the default.

MEANS

MEANS [TABLES =]
      {VAR_LIST}
        [ BY {VAR_LIST} [BY {VAR_LIST} [BY {VAR_LIST} ... ]]]

      [ /{VAR_LIST}
         [ BY {VAR_LIST} [BY {VAR_LIST} [BY {VAR_LIST} ... ]]] ]

      [/CELLS = [MEAN] [COUNT] [STDDEV] [SEMEAN] [SUM] [MIN] [MAX] [RANGE]
        [VARIANCE] [KURT] [SEKURT]
        [SKEW] [SESKEW] [FIRST] [LAST]
        [HARMONIC] [GEOMETRIC]
        [DEFAULT]
        [ALL]
        [NONE] ]

      [/MISSING = [INCLUDE] [DEPENDENT]]

You can use the MEANS command to calculate the arithmetic mean and similar statistics, either for the dataset as a whole or for categories of data.

The simplest form of the command is

MEANS V.

which calculates the mean, count and standard deviation for V. If you specify a grouping variable, for example

MEANS V BY G.

then the means, counts and standard deviations for V after having been grouped by G are calculated. Instead of the mean, count and standard deviation, you could specify the statistics in which you are interested:

MEANS X Y BY G
      /CELLS = HARMONIC SUM MIN.

This example calculates the harmonic mean, the sum and the minimum values of X and Y grouped by G.

The CELLS subcommand specifies which statistics to calculate. The available statistics are:

  • MEAN: The arithmetic mean.
  • COUNT: The count of the values.
  • STDDEV: The standard deviation.
  • SEMEAN: The standard error of the mean.
  • SUM: The sum of the values.
  • MIN: The minimum value.
  • MAX: The maximum value.
  • RANGE: The difference between the maximum and minimum values.
  • VARIANCE: The variance.
  • FIRST: The first value in the category.
  • LAST: The last value in the category.
  • SKEW: The skewness.
  • SESKEW: The standard error of the skewness.
  • KURT: The kurtosis
  • SEKURT: The standard error of the kurtosis.
  • HARMONIC: The harmonic mean.
  • GEOMETRIC: The geometric mean.

In addition, three special keywords are recognized:

  • DEFAULT: This is the same as MEAN COUNT STDDEV.
  • ALL: All of the above statistics are calculated.
  • NONE: No statistics are calculated (only a summary is shown).

More than one "table" can be specified in a single command. Each table is separated by a /. For example

     MEANS TABLES =
           c d e BY x
           /a b BY x y
           /f BY y BY z.

has three tables (the TABLE = is optional). The first table has three dependent variables c, d, and e and a single categorical variable x. The second table has two dependent variables a and b, and two categorical variables x and y. The third table has a single dependent variable f and a categorical variable formed by the combination of y and Z.

By default values are omitted from the analysis only if missing values (either system missing or user missing) for any of the variables directly involved in their calculation are encountered. This behaviour can be modified with the /MISSING subcommand. Three options are possible: TABLE, INCLUDE and DEPENDENT.

/MISSING = INCLUDE says that user missing values, either in the dependent variables or in the categorical variables should be taken at their face value, and not excluded.

/MISSING = DEPENDENT says that user missing values, in the dependent variables should be taken at their face value, however cases which have user missing values for the categorical variables should be omitted from the calculation.

Example

The dataset in repairs.sav contains the mean time between failures (mtbf) for a sample of artifacts produced by different factories and trialed under different operating conditions. Since there are four combinations of categorical variables, by simply looking at the list of data, it would be hard to how the scores vary for each category. The syntax below shows one way of tabulating the mtbf in a way which is easier to understand.

get file='repairs.sav'.

means tables = mtbf
      by factory by environment.

The results are shown below. The figures shown indicate the mean, standard deviation and number of samples in each category. These figures however do not indicate whether the results are statistically significant. For that, you would need to use the procedures ONEWAY, GLM or T-TEST depending on the hypothesis being tested.

                    Case Processing Summary
┌────────────────────────────┬───────────────────────────────┐
│                            │             Cases             │
│                            ├──────────┬─────────┬──────────┤
│                            │ Included │ Excluded│   Total  │
│                            ├──┬───────┼─┬───────┼──┬───────┤
│                            │ N│Percent│N│Percent│ N│Percent│
├────────────────────────────┼──┼───────┼─┼───────┼──┼───────┤
│mtbf * factory * environment│30│ 100.0%│0│    .0%│30│ 100.0%│
└────────────────────────────┴──┴───────┴─┴───────┴──┴───────┘

                                Report
┌────────────────────────────────────────────┬─────┬──┬──────────────┐
│Manufacturing facility Operating Environment│ Mean│ N│Std. Deviation│
├────────────────────────────────────────────┼─────┼──┼──────────────┤
│0                      Temperate            │ 7.26│ 9│          2.57│
│                       Tropical             │ 7.47│ 7│          2.68│
│                       Total                │ 7.35│16│          2.53│
├────────────────────────────────────────────┼─────┼──┼──────────────┤
│1                      Temperate            │13.38│ 6│          7.77│
│                       Tropical             │ 8.20│ 8│          8.39│
│                       Total                │10.42│14│          8.26│
├────────────────────────────────────────────┼─────┼──┼──────────────┤
│Total                  Temperate            │ 9.71│15│          5.91│
│                       Tropical             │ 7.86│15│          6.20│
│                       Total                │ 8.78│30│          6.03│
└────────────────────────────────────────────┴─────┴──┴──────────────┘

PSPP does not limit the number of variables for which you can calculate statistics, nor number of categorical variables per layer, nor the number of layers. However, running MEANS on a large number of variables, or with categorical variables containing a large number of distinct values, may result in an extremely large output, which will not be easy to interpret. So you should consider carefully which variables to select for participation in the analysis.

NPAR TESTS

NPAR TESTS
     nonparametric test subcommands
     .
     .
     .

     [ /STATISTICS={DESCRIPTIVES} ]

     [ /MISSING={ANALYSIS, LISTWISE} {INCLUDE, EXCLUDE} ]

     [ /METHOD=EXACT [ TIMER [(N)] ] ]

NPAR TESTS performs nonparametric tests. Nonparametric tests make very few assumptions about the distribution of the data. One or more tests may be specified by using the corresponding subcommand. If the /STATISTICS subcommand is also specified, then summary statistics are produces for each variable that is the subject of any test.

Certain tests may take a long time to execute, if an exact figure is required. Therefore, by default asymptotic approximations are used unless the subcommand /METHOD=EXACT is specified. Exact tests give more accurate results, but may take an unacceptably long time to perform. If the TIMER keyword is used, it sets a maximum time, after which the test is abandoned, and a warning message printed. The time, in minutes, should be specified in parentheses after the TIMER keyword. If the TIMER keyword is given without this figure, then a default value of 5 minutes is used.

Binomial test

     [ /BINOMIAL[(P)]=VAR_LIST[(VALUE1[, VALUE2)] ] ]

The /BINOMIAL subcommand compares the observed distribution of a dichotomous variable with that of a binomial distribution. The variable P specifies the test proportion of the binomial distribution. The default value of 0.5 is assumed if P is omitted.

If a single value appears after the variable list, then that value is used as the threshold to partition the observed values. Values less than or equal to the threshold value form the first category. Values greater than the threshold form the second category.

If two values appear after the variable list, then they are used as the values which a variable must take to be in the respective category. Cases for which a variable takes a value equal to neither of the specified values, take no part in the test for that variable.

If no values appear, then the variable must assume dichotomous values. If more than two distinct, non-missing values for a variable under test are encountered then an error occurs.

If the test proportion is equal to 0.5, then a two tailed test is reported. For any other test proportion, a one tailed test is reported. For one tailed tests, if the test proportion is less than or equal to the observed proportion, then the significance of observing the observed proportion or more is reported. If the test proportion is more than the observed proportion, then the significance of observing the observed proportion or less is reported. That is to say, the test is always performed in the observed direction.

PSPP uses a very precise approximation to the gamma function to compute the binomial significance. Thus, exact results are reported even for very large sample sizes.

Chi-square Test

     [ /CHISQUARE=VAR_LIST[(LO,HI)] [/EXPECTED={EQUAL|F1, F2 ... FN}] ]

The /CHISQUARE subcommand produces a chi-square statistic for the differences between the expected and observed frequencies of the categories of a variable. Optionally, a range of values may appear after the variable list. If a range is given, then non-integer values are truncated, and values outside the specified range are excluded from the analysis.

The /EXPECTED subcommand specifies the expected values of each category. There must be exactly one non-zero expected value, for each observed category, or the EQUAL keyword must be specified. You may use the notation N*F to specify N consecutive expected categories all taking a frequency of F. The frequencies given are proportions, not absolute frequencies. The sum of the frequencies need not be 1. If no /EXPECTED subcommand is given, then equal frequencies are expected.

Chi-square Example

A researcher wishes to investigate whether there are an equal number of persons of each sex in a population. The sample chosen for invesigation is that from the physiology.sav dataset. The null hypothesis for the test is that the population comprises an equal number of males and females. The analysis is performed as shown below:

get file='physiology.sav'.

npar test
     /chisquare=sex.

There is only one test variable: sex. The other variables in the dataset are ignored.

In the output, shown below, the summary box shows that in the sample, there are more males than females. However the significance of chi-square result is greater than 0.05—the most commonly accepted p-value—and therefore there is not enough evidence to reject the null hypothesis and one must conclude that the evidence does not indicate that there is an imbalance of the sexes in the population.

             Sex of subject
┌──────┬──────────┬──────────┬────────┐
│Value │Observed N│Expected N│Residual│
├──────┼──────────┼──────────┼────────┤
│Male  │        22│     20.00│    2.00│
│Female│        18│     20.00│   ─2.00│
│Total │        40│          │        │
└──────┴──────────┴──────────┴────────┘

         Test Statistics
┌──────────────┬──────────┬──┬───────────┐
│              │Chi─square│df│Asymp. Sig.│
├──────────────┼──────────┼──┼───────────┤
│Sex of subject│       .40│ 1│       .527│
└──────────────┴──────────┴──┴───────────┘

Cochran Q Test

     [ /COCHRAN = VAR_LIST ]

The Cochran Q test is used to test for differences between three or more groups. The data for VAR_LIST in all cases must assume exactly two distinct values (other than missing values).

The value of Q is displayed along with its asymptotic significance based on a chi-square distribution.

Friedman Test

     [ /FRIEDMAN = VAR_LIST ]

The Friedman test is used to test for differences between repeated measures when there is no indication that the distributions are normally distributed.

A list of variables which contain the measured data must be given. The procedure prints the sum of ranks for each variable, the test statistic and its significance.

Kendall's W Test

     [ /KENDALL = VAR_LIST ]

The Kendall test investigates whether an arbitrary number of related samples come from the same population. It is identical to the Friedman test except that the additional statistic W, Kendall's Coefficient of Concordance is printed. It has the range [0,1]—a value of zero indicates no agreement between the samples whereas a value of unity indicates complete agreement.

Kolmogorov-Smirnov Test

     [ /KOLMOGOROV-SMIRNOV ({NORMAL [MU, SIGMA], UNIFORM [MIN, MAX], POISSON [LAMBDA], EXPONENTIAL [SCALE] }) = VAR_LIST ]

The one sample Kolmogorov-Smirnov subcommand is used to test whether or not a dataset is drawn from a particular distribution. Four distributions are supported: normal, uniform, Poisson and exponential.

Ideally you should provide the parameters of the distribution against which you wish to test the data. For example, with the normal distribution the mean (MU) and standard deviation (SIGMA) should be given; with the uniform distribution, the minimum (MIN) and maximum (MAX) value should be provided. However, if the parameters are omitted they are imputed from the data. Imputing the parameters reduces the power of the test so should be avoided if possible.

In the following example, two variables score and age are tested to see if they follow a normal distribution with a mean of 3.5 and a standard deviation of 2.0.

  NPAR TESTS
        /KOLMOGOROV-SMIRNOV (NORMAL 3.5 2.0) = score age.

If the variables need to be tested against different distributions, then a separate subcommand must be used. For example the following syntax tests score against a normal distribution with mean of 3.5 and standard deviation of 2.0 whilst age is tested against a normal distribution of mean 40 and standard deviation 1.5.

  NPAR TESTS
        /KOLMOGOROV-SMIRNOV (NORMAL 3.5 2.0) = score
        /KOLMOGOROV-SMIRNOV (NORMAL 40 1.5) =  age.

The abbreviated subcommand K-S may be used in place of KOLMOGOROV-SMIRNOV.

Kruskal-Wallis Test

     [ /KRUSKAL-WALLIS = VAR_LIST BY VAR (LOWER, UPPER) ]

The Kruskal-Wallis test is used to compare data from an arbitrary number of populations. It does not assume normality. The data to be compared are specified by VAR_LIST. The categorical variable determining the groups to which the data belongs is given by VAR. The limits LOWER and UPPER specify the valid range of VAR. If UPPER is smaller than LOWER, the PSPP will assume their values to be reversed. Any cases for which VAR falls outside [LOWER, UPPER] are ignored.

The mean rank of each group as well as the chi-squared value and significance of the test are printed. The abbreviated subcommand K-W may be used in place of KRUSKAL-WALLIS.

Mann-Whitney U Test

     [ /MANN-WHITNEY = VAR_LIST BY var (GROUP1, GROUP2) ]

The Mann-Whitney subcommand is used to test whether two groups of data come from different populations. The variables to be tested should be specified in VAR_LIST and the grouping variable, that determines to which group the test variables belong, in VAR. VAR may be either a string or an alpha variable. GROUP1 and GROUP2 specify the two values of VAR which determine the groups of the test data. Cases for which the VAR value is neither GROUP1 or GROUP2 are ignored.

The value of the Mann-Whitney U statistic, the Wilcoxon W, and the significance are printed. You may abbreviated the subcommand MANN-WHITNEY to M-W.

McNemar Test

     [ /MCNEMAR VAR_LIST [ WITH VAR_LIST [ (PAIRED) ]]]

Use McNemar's test to analyse the significance of the difference between pairs of correlated proportions.

If the WITH keyword is omitted, then tests for all combinations of the listed variables are performed. If the WITH keyword is given, and the (PAIRED) keyword is also given, then the number of variables preceding WITH must be the same as the number following it. In this case, tests for each respective pair of variables are performed. If the WITH keyword is given, but the (PAIRED) keyword is omitted, then tests for each combination of variable preceding WITH against variable following WITH are performed.

The data in each variable must be dichotomous. If there are more than two distinct variables an error will occur and the test will not be run.

Median Test

     [ /MEDIAN [(VALUE)] = VAR_LIST BY VARIABLE (VALUE1, VALUE2) ]

The median test is used to test whether independent samples come from populations with a common median. The median of the populations against which the samples are to be tested may be given in parentheses immediately after the /MEDIAN subcommand. If it is not given, the median is imputed from the union of all the samples.

The variables of the samples to be tested should immediately follow the = sign. The keyword BY must come next, and then the grouping variable. Two values in parentheses should follow. If the first value is greater than the second, then a 2-sample test is performed using these two values to determine the groups. If however, the first variable is less than the second, then a k sample test is conducted and the group values used are all values encountered which lie in the range [VALUE1,VALUE2].

Runs Test

     [ /RUNS ({MEAN, MEDIAN, MODE, VALUE})  = VAR_LIST ]

The /RUNS subcommand tests whether a data sequence is randomly ordered.

It works by examining the number of times a variable's value crosses a given threshold. The desired threshold must be specified within parentheses. It may either be specified as a number or as one of MEAN, MEDIAN or MODE. Following the threshold specification comes the list of variables whose values are to be tested.

The subcommand shows the number of runs, the asymptotic significance based on the length of the data.

Sign Test

     [ /SIGN VAR_LIST [ WITH VAR_LIST [ (PAIRED) ]]]

The /SIGN subcommand tests for differences between medians of the variables listed. The test does not make any assumptions about the distribution of the data.

If the WITH keyword is omitted, then tests for all combinations of the listed variables are performed. If the WITH keyword is given, and the (PAIRED) keyword is also given, then the number of variables preceding WITH must be the same as the number following it. In this case, tests for each respective pair of variables are performed. If the WITH keyword is given, but the (PAIRED) keyword is omitted, then tests for each combination of variable preceding WITH against variable following WITH are performed.

Wilcoxon Matched Pairs Signed Ranks Test

     [ /WILCOXON VAR_LIST [ WITH VAR_LIST [ (PAIRED) ]]]

The /WILCOXON subcommand tests for differences between medians of the variables listed. The test does not make any assumptions about the variances of the samples. It does however assume that the distribution is symmetrical.

If the WITH keyword is omitted, then tests for all combinations of the listed variables are performed. If the WITH keyword is given, and the (PAIRED) keyword is also given, then the number of variables preceding WITH must be the same as the number following it. In this case, tests for each respective pair of variables are performed. If the WITH keyword is given, but the (PAIRED) keyword is omitted, then tests for each combination of variable preceding WITH against variable following WITH are performed.

T-TEST

T-TEST
        /MISSING={ANALYSIS,LISTWISE} {EXCLUDE,INCLUDE}
        /CRITERIA=CI(CONFIDENCE)


(One Sample mode.)
        TESTVAL=TEST_VALUE
        /VARIABLES=VAR_LIST


(Independent Samples mode.)
        GROUPS=var(VALUE1 [, VALUE2])
        /VARIABLES=VAR_LIST


(Paired Samples mode.)
        PAIRS=VAR_LIST [WITH VAR_LIST [(PAIRED)] ]

The T-TEST procedure outputs tables used in testing hypotheses about means. It operates in one of three modes:

Each of these modes are described in more detail below. There are two optional subcommands which are common to all modes.

The /CRITERIA subcommand tells PSPP the confidence interval used in the tests. The default value is 0.95.

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set, then user-missing values are included in the calculations, but system-missing values are not. If EXCLUDE is set, which is the default, user-missing values are excluded as well as system-missing values. This is the default.

If LISTWISE is set, then the entire case is excluded from analysis whenever any variable specified in the /VARIABLES, /PAIRS or /GROUPS subcommands contains a missing value. If ANALYSIS is set, then missing values are excluded only in the analysis for which they would be needed. This is the default.

One Sample Mode

The TESTVAL subcommand invokes the One Sample mode. This mode is used to test a population mean against a hypothesized mean. The value given to the TESTVAL subcommand is the value against which you wish to test. In this mode, you must also use the /VARIABLES subcommand to tell PSPP which variables you wish to test.

Example

A researcher wishes to know whether the weight of persons in a population is different from the national average. The samples are drawn from the population under investigation and recorded in the file physiology.sav. From the Department of Health, she knows that the national average weight of healthy adults is 76.8kg. Accordingly the TESTVAL is set to 76.8. The null hypothesis therefore is that the mean average weight of the population from which the sample was drawn is 76.8kg.

As previously noted, one sample in the dataset contains a weight value which is clearly incorrect. So this is excluded from the analysis using the SELECT command.

GET FILE='physiology.sav'.

SELECT IF (weight > 0).

T-TEST TESTVAL = 76.8
   /VARIABLES = weight.

The output below shows that the mean of our sample differs from the test value by -1.40kg. However the significance is very high (0.610). So one cannot reject the null hypothesis, and must conclude there is not enough evidence to suggest that the mean weight of the persons in our population is different from 76.8kg.

                 One─Sample Statistics
┌───────────────────┬──┬─────┬──────────────┬─────────┐
│                   │ N│ Mean│Std. Deviation│S.E. Mean│
├───────────────────┼──┼─────┼──────────────┼─────────┤
│Weight in kilograms│39│75.40│         17.08│     2.73│
└───────────────────┴──┴─────┴──────────────┴─────────┘

                                One─Sample Test
┌──────────────┬──────────────────────────────────────────────────────────────┐
│              │                       Test Value = 76.8                      │
│              ├────┬──┬────────────┬────────────┬────────────────────────────┤
│              │    │  │            │            │ 95% Confidence Interval of │
│              │    │  │            │            │       the Difference       │
│              │    │  │  Sig. (2─  │    Mean    ├──────────────┬─────────────┤
│              │  t │df│   tailed)  │ Difference │     Lower    │    Upper    │
├──────────────┼────┼──┼────────────┼────────────┼──────────────┼─────────────┤
│Weight in     │─.51│38│        .610│       ─1.40│         ─6.94│         4.13│
│kilograms     │    │  │            │            │              │             │
└──────────────┴────┴──┴────────────┴────────────┴──────────────┴─────────────┘

Independent Samples Mode

The GROUPS subcommand invokes Independent Samples mode or 'Groups' mode. This mode is used to test whether two groups of values have the same population mean. In this mode, you must also use the /VARIABLES subcommand to tell PSPP the dependent variables you wish to test.

The variable given in the GROUPS subcommand is the independent variable which determines to which group the samples belong. The values in parentheses are the specific values of the independent variable for each group. If the parentheses are omitted and no values are given, the default values of 1.0 and 2.0 are assumed.

If the independent variable is numeric, it is acceptable to specify only one value inside the parentheses. If you do this, cases where the independent variable is greater than or equal to this value belong to the first group, and cases less than this value belong to the second group. When using this form of the GROUPS subcommand, missing values in the independent variable are excluded on a listwise basis, regardless of whether /MISSING=LISTWISE was specified.

Example

A researcher wishes to know whether within a population, adult males are taller than adult females. The samples are drawn from the population under investigation and recorded in the file physiology.sav.

As previously noted, one sample in the dataset contains a height value which is clearly incorrect. So this is excluded from the analysis using the SELECT command.

get file='physiology.sav'.

select if (height >= 200).

t-test /variables = height
       /groups = sex(0,1).

The null hypothesis is that both males and females are on average of equal height.

From the output, shown below, one can clearly see that the sample mean height is greater for males than for females. However in order to see if this is a significant result, one must consult the T-Test table.

The T-Test table contains two rows; one for use if the variance of the samples in each group may be safely assumed to be equal, and the second row if the variances in each group may not be safely assumed to be equal.

In this case however, both rows show a 2-tailed significance less than 0.001 and one must therefore reject the null hypothesis and conclude that within the population the mean height of males and of females are unequal.

                         Group Statistics
┌────────────────────────────┬──┬───────┬──────────────┬─────────┐
│                      Group │ N│  Mean │Std. Deviation│S.E. Mean│
├────────────────────────────┼──┼───────┼──────────────┼─────────┤
│Height in millimeters Male  │22│1796.49│         49.71│    10.60│
│                      Female│17│1610.77│         25.43│     6.17│
└────────────────────────────┴──┴───────┴──────────────┴─────────┘

                          Independent Samples Test
┌─────────────────────┬──────────┬──────────────────────────────────────────
│                     │ Levene's │
│                     │ Test for │
│                     │ Equality │
│                     │    of    │
│                     │ Variances│              T─Test for Equality of Means
│                     ├────┬─────┼─────┬─────┬───────┬──────────┬──────────┐
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │       │          │          │
│                     │    │     │     │     │  Sig. │          │          │
│                     │    │     │     │     │  (2─  │   Mean   │Std. Error│
│                     │  F │ Sig.│  t  │  df │tailed)│Difference│Difference│
├─────────────────────┼────┼─────┼─────┼─────┼───────┼──────────┼──────────┤
│Height in   Equal    │ .97│ .331│14.02│37.00│   .000│    185.72│     13.24│
│millimeters variances│    │     │     │     │       │          │          │
│            assumed  │    │     │     │     │       │          │          │
│            Equal    │    │     │15.15│32.71│   .000│    185.72│     12.26│
│            variances│    │     │     │     │       │          │          │
│            not      │    │     │     │     │       │          │          │
│            assumed  │    │     │     │     │       │          │          │
└─────────────────────┴────┴─────┴─────┴─────┴───────┴──────────┴──────────┘

┌─────────────────────┬─────────────┐
│                     │             │
│                     │             │
│                     │             │
│                     │             │
│                     │             │
│                     ├─────────────┤
│                     │     95%     │
│                     │  Confidence │
│                     │ Interval of │
│                     │     the     │
│                     │  Difference │
│                     ├──────┬──────┤
│                     │ Lower│ Upper│
├─────────────────────┼──────┼──────┤
│Height in   Equal    │158.88│212.55│
│millimeters variances│      │      │
│            assumed  │      │      │
│            Equal    │160.76│210.67│
│            variances│      │      │
│            not      │      │      │
│            assumed  │      │      │
└─────────────────────┴──────┴──────┘

Paired Samples Mode

The PAIRS subcommand introduces Paired Samples mode. Use this mode when repeated measures have been taken from the same samples. If the WITH keyword is omitted, then tables for all combinations of variables given in the PAIRS subcommand are generated. If the WITH keyword is given, and the (PAIRED) keyword is also given, then the number of variables preceding WITH must be the same as the number following it. In this case, tables for each respective pair of variables are generated. In the event that the WITH keyword is given, but the (PAIRED) keyword is omitted, then tables for each combination of variable preceding WITH against variable following WITH are generated.

ONEWAY

ONEWAY
        [/VARIABLES = ] VAR_LIST BY VAR
        /MISSING={ANALYSIS,LISTWISE} {EXCLUDE,INCLUDE}
        /CONTRAST= VALUE1 [, VALUE2] ... [,VALUEN]
        /STATISTICS={DESCRIPTIVES,HOMOGENEITY}
        /POSTHOC={BONFERRONI, GH, LSD, SCHEFFE, SIDAK, TUKEY, ALPHA ([VALUE])}

The ONEWAY procedure performs a one-way analysis of variance of variables factored by a single independent variable. It is used to compare the means of a population divided into more than two groups.

The dependent variables to be analysed should be given in the VARIABLES subcommand. The list of variables must be followed by the BY keyword and the name of the independent (or factor) variable.

You can use the STATISTICS subcommand to tell PSPP to display ancillary information. The options accepted are:

  • DESCRIPTIVES: Displays descriptive statistics about the groups factored by the independent variable.
  • HOMOGENEITY: Displays the Levene test of Homogeneity of Variance for the variables and their groups.

The CONTRAST subcommand is used when you anticipate certain differences between the groups. The subcommand must be followed by a list of numerals which are the coefficients of the groups to be tested. The number of coefficients must correspond to the number of distinct groups (or values of the independent variable). If the total sum of the coefficients are not zero, then PSPP will display a warning, but will proceed with the analysis. The CONTRAST subcommand may be given up to 10 times in order to specify different contrast tests. The MISSING subcommand defines how missing values are handled. If LISTWISE is specified then cases which have missing values for the independent variable or any dependent variable are ignored. If ANALYSIS is specified, then cases are ignored if the independent variable is missing or if the dependent variable currently being analysed is missing. The default is ANALYSIS. A setting of EXCLUDE means that variables whose values are user-missing are to be excluded from the analysis. A setting of INCLUDE means they are to be included. The default is EXCLUDE.

Using the POSTHOC subcommand you can perform multiple pairwise comparisons on the data. The following comparison methods are available:

  • LSD: Least Significant Difference.
  • TUKEY: Tukey Honestly Significant Difference.
  • BONFERRONI: Bonferroni test.
  • SCHEFFE: Scheffé's test.
  • SIDAK: Sidak test.
  • GH: The Games-Howell test.

Use the optional syntax ALPHA(VALUE) to indicate that ONEWAY should perform the posthoc tests at a confidence level of VALUE. If ALPHA(VALUE) is not specified, then the confidence level used is 0.05.

QUICK CLUSTER

QUICK CLUSTER VAR_LIST
      [/CRITERIA=CLUSTERS(K) [MXITER(MAX_ITER)] CONVERGE(EPSILON) [NOINITIAL]]
      [/MISSING={EXCLUDE,INCLUDE} {LISTWISE, PAIRWISE}]
      [/PRINT={INITIAL} {CLUSTER}]
      [/SAVE[=[CLUSTER[(MEMBERSHIP_VAR)]] [DISTANCE[(DISTANCE_VAR)]]]

The QUICK CLUSTER command performs k-means clustering on the dataset. This is useful when you wish to allocate cases into clusters of similar values and you already know the number of clusters.

The minimum specification is QUICK CLUSTER followed by the names of the variables which contain the cluster data. Normally you will also want to specify /CRITERIA=CLUSTERS(K) where K is the number of clusters. If this is not specified, then K defaults to 2.

If you use /CRITERIA=NOINITIAL then a naive algorithm to select the initial clusters is used. This will provide for faster execution but less well separated initial clusters and hence possibly an inferior final result.

QUICK CLUSTER uses an iterative algorithm to select the clusters centers. The subcommand /CRITERIA=MXITER(MAX_ITER) sets the maximum number of iterations. During classification, PSPP will continue iterating until until MAX_ITER iterations have been done or the convergence criterion (see below) is fulfilled. The default value of MAX_ITER is 2.

If however, you specify /CRITERIA=NOUPDATE then after selecting the initial centers, no further update to the cluster centers is done. In this case, MAX_ITER, if specified, is ignored.

The subcommand /CRITERIA=CONVERGE(EPSILON) is used to set the convergence criterion. The value of convergence criterion is EPSILON times the minimum distance between the initial cluster centers. Iteration stops when the mean cluster distance between one iteration and the next is less than the convergence criterion. The default value of EPSILON is zero.

The MISSING subcommand determines the handling of missing variables. If INCLUDE is set, then user-missing values are considered at their face value and not as missing values. If EXCLUDE is set, which is the default, user-missing values are excluded as well as system-missing values.

If LISTWISE is set, then the entire case is excluded from the analysis whenever any of the clustering variables contains a missing value. If PAIRWISE is set, then a case is considered missing only if all the clustering variables contain missing values. Otherwise it is clustered on the basis of the non-missing values. The default is LISTWISE.

The PRINT subcommand requests additional output to be printed. If INITIAL is set, then the initial cluster memberships will be printed. If CLUSTER is set, the cluster memberships of the individual cases are displayed (potentially generating lengthy output).

You can specify the subcommand SAVE to ask that each case's cluster membership and the euclidean distance between the case and its cluster center be saved to a new variable in the active dataset. To save the cluster membership use the CLUSTER keyword and to save the distance use the DISTANCE keyword. Each keyword may optionally be followed by a variable name in parentheses to specify the new variable which is to contain the saved parameter. If no variable name is specified, then PSPP will create one.

RANK

RANK
        [VARIABLES=] VAR_LIST [{A,D}] [BY VAR_LIST]
        /TIES={MEAN,LOW,HIGH,CONDENSE}
        /FRACTION={BLOM,TUKEY,VW,RANKIT}
        /PRINT[={YES,NO}
        /MISSING={EXCLUDE,INCLUDE}

        /RANK [INTO VAR_LIST]
        /NTILES(k) [INTO VAR_LIST]
        /NORMAL [INTO VAR_LIST]
        /PERCENT [INTO VAR_LIST]
        /RFRACTION [INTO VAR_LIST]
        /PROPORTION [INTO VAR_LIST]
        /N [INTO VAR_LIST]
        /SAVAGE [INTO VAR_LIST]

The RANK command ranks variables and stores the results into new variables.

The VARIABLES subcommand, which is mandatory, specifies one or more variables whose values are to be ranked. After each variable, A or D may appear, indicating that the variable is to be ranked in ascending or descending order. Ascending is the default. If a BY keyword appears, it should be followed by a list of variables which are to serve as group variables. In this case, the cases are gathered into groups, and ranks calculated for each group.

The TIES subcommand specifies how tied values are to be treated. The default is to take the mean value of all the tied cases.

The FRACTION subcommand specifies how proportional ranks are to be calculated. This only has any effect if NORMAL or PROPORTIONAL rank functions are requested.

The PRINT subcommand may be used to specify that a summary of the rank variables created should appear in the output.

The function subcommands are RANK, NTILES, NORMAL, PERCENT, RFRACTION, PROPORTION, and SAVAGE. Any number of function subcommands may appear. If none are given, then the default is RANK. The NTILES subcommand must take an integer specifying the number of partitions into which values should be ranked. Each subcommand may be followed by the INTO keyword and a list of variables which are the variables to be created and receive the rank scores. There may be as many variables specified as there are variables named on the VARIABLES subcommand. If fewer are specified, then the variable names are automatically created.

The MISSING subcommand determines how user missing values are to be treated. A setting of EXCLUDE means that variables whose values are user-missing are to be excluded from the rank scores. A setting of INCLUDE means they are to be included. The default is EXCLUDE.

REGRESSION

The REGRESSION procedure fits linear models to data via least-squares estimation. The procedure is appropriate for data which satisfy those assumptions typical in linear regression:

  • The data set contains \(n\) observations of a dependent variable, say \(y_1,...,y_n\), and \(n\) observations of one or more explanatory variables. Let \(x_{11}, x_{12}, ..., x_{1n}\) denote the \(n\) observations of the first explanatory variable; \(x_{21},...,x_{2n}\) denote the \(n\) observations of the second explanatory variable; \(x_{k1},...,x_{kn}\) denote the \(n\) observations of the kth explanatory variable.

  • The dependent variable \(y\) has the following relationship to the explanatory variables: \(y_i = b_0 + b_1 x_{1i} + ... + b_k x_{ki} + z_i\) where \(b_0, b_1, ..., b_k\) are unknown coefficients, and \(z_1,...,z_n\) are independent, normally distributed "noise" terms with mean zero and common variance. The noise, or "error" terms are unobserved. This relationship is called the "linear model".

    The REGRESSION procedure estimates the coefficients \(b_0,...,b_k\) and produces output relevant to inferences for the linear model.

Syntax

REGRESSION
        /VARIABLES=VAR_LIST
        /DEPENDENT=VAR_LIST
        /STATISTICS={ALL, DEFAULTS, R, COEFF, ANOVA, BCOV, CI[CONF, TOL]}
        { /ORIGIN | /NOORIGIN }
        /SAVE={PRED, RESID}

The REGRESSION procedure reads the active dataset and outputs statistics relevant to the linear model specified by the user.

The VARIABLES subcommand, which is required, specifies the list of variables to be analyzed. Keyword VARIABLES is required. The DEPENDENT subcommand specifies the dependent variable of the linear model. The DEPENDENT subcommand is required. All variables listed in the VARIABLES subcommand, but not listed in the DEPENDENT subcommand, are treated as explanatory variables in the linear model.

All other subcommands are optional:

The STATISTICS subcommand specifies which statistics are to be displayed. The following keywords are accepted:

  • ALL
    All of the statistics below.
  • R
    The ratio of the sums of squares due to the model to the total sums of squares for the dependent variable.
  • COEFF
    A table containing the estimated model coefficients and their standard errors.
  • CI (CONF)
    This item is only relevant if COEFF has also been selected. It specifies that the confidence interval for the coefficients should be printed. The optional value CONF, which must be in parentheses, is the desired confidence level expressed as a percentage.
  • ANOVA
    Analysis of variance table for the model.
  • BCOV
    The covariance matrix for the estimated model coefficients.
  • TOL
    The variance inflation factor and its reciprocal. This has no effect unless COEFF is also given.
  • DEFAULT
    The same as if R, COEFF, and ANOVA had been selected. This is what you get if the /STATISTICS command is not specified, or if it is specified without any parameters.

The ORIGIN and NOORIGIN subcommands are mutually exclusive. ORIGIN indicates that the regression should be performed through the origin. You should use this option if, and only if you have reason to believe that the regression does indeed pass through the origin -- that is to say, the value b_0 above, is zero. The default is NOORIGIN.

The SAVE subcommand causes PSPP to save the residuals or predicted values from the fitted model to the active dataset. PSPP will store the residuals in a variable called RES1 if no such variable exists, RES2 if RES1 already exists, RES3 if RES1 and RES2 already exist, etc. It will choose the name of the variable for the predicted values similarly, but with PRED as a prefix. When SAVE is used, PSPP ignores TEMPORARY, treating temporary transformations as permanent.

Example

The following PSPP syntax will generate the default output and save the predicted values and residuals to the active dataset.

title 'Demonstrate REGRESSION procedure'.
data list / v0 1-2 (A) v1 v2 3-22 (10).
begin data.
b  7.735648 -23.97588
b  6.142625 -19.63854
a  7.651430 -25.26557
c  6.125125 -16.57090
a  8.245789 -25.80001
c  6.031540 -17.56743
a  9.832291 -28.35977
c  5.343832 -16.79548
a  8.838262 -29.25689
b  6.200189 -18.58219
end data.
list.
regression /variables=v0 v1 v2 /statistics defaults /dependent=v2
           /save pred resid /method=enter.

RELIABILITY

RELIABILITY
        /VARIABLES=VAR_LIST
        /SCALE (NAME) = {VAR_LIST, ALL}
        /MODEL={ALPHA, SPLIT[(N)]}
        /SUMMARY={TOTAL,ALL}
        /MISSING={EXCLUDE,INCLUDE}

The RELIABILITY command performs reliability analysis on the data.

The VARIABLES subcommand is required. It determines the set of variables upon which analysis is to be performed.

The SCALE subcommand determines the variables for which reliability is to be calculated. If SCALE is omitted, then analysis for all variables named in the VARIABLES subcommand are used. Optionally, the NAME parameter may be specified to set a string name for the scale.

The MODEL subcommand determines the type of analysis. If ALPHA is specified, then Cronbach's Alpha is calculated for the scale. If the model is SPLIT, then the variables are divided into 2 subsets. An optional parameter N may be given, to specify how many variables to be in the first subset. If N is omitted, then it defaults to one half of the variables in the scale, or one half minus one if there are an odd number of variables. The default model is ALPHA.

By default, any cases with user missing, or system missing values for any variables given in the VARIABLES subcommand are omitted from the analysis. The MISSING subcommand determines whether user missing values are included or excluded in the analysis.

The SUMMARY subcommand determines the type of summary analysis to be performed. Currently there is only one type: SUMMARY=TOTAL, which displays per-item analysis tested against the totals.

Example

Before analysing the results of a survey—particularly for a multiple choice survey—it is desirable to know whether the respondents have considered their answers or simply provided random answers.

In the following example the survey results from the file hotel.sav are used. All five survey questions are included in the reliability analysis. However, before running the analysis, the data must be preprocessed. An examination of the survey questions reveals that two questions, viz: v3 and v5 are negatively worded, whereas the others are positively worded. All questions must be based upon the same scale for the analysis to be meaningful. One could use the RECODE command, however a simpler way is to use COMPUTE and this is what is done in the syntax below.

get file="hotel.sav".

* Recode V3 and V5 inverting the sense of the values.
compute v3 = 6 - v3.
compute v5 = 6 - v5.

reliability
   /variables= all
   /model=alpha.

In this case, all variables in the data set are used, so we can use the special keyword ALL.

The output, below, shows that Cronbach's Alpha is 0.11 which is a value normally considered too low to indicate consistency within the data. This is possibly due to the small number of survey questions. The survey should be redesigned before serious use of the results are applied.

Scale: ANY

Case Processing Summary
┌────────┬──┬───────┐
│Cases   │ N│Percent│
├────────┼──┼───────┤
│Valid   │17│ 100.0%│
│Excluded│ 0│    .0%│
│Total   │17│ 100.0%│
└────────┴──┴───────┘

    Reliability Statistics
┌────────────────┬──────────┐
│Cronbach's Alpha│N of Items│
├────────────────┼──────────┤
│             .11│         5│
└────────────────┴──────────┘

ROC

ROC
        VAR_LIST BY STATE_VAR (STATE_VALUE)
        /PLOT = { CURVE [(REFERENCE)], NONE }
        /PRINT = [ SE ] [ COORDINATES ]
        /CRITERIA = [ CUTOFF({INCLUDE,EXCLUDE}) ]
          [ TESTPOS ({LARGE,SMALL}) ]
          [ CI (CONFIDENCE) ]
          [ DISTRIBUTION ({FREE, NEGEXPO }) ]
        /MISSING={EXCLUDE,INCLUDE}

The ROC command is used to plot the receiver operating characteristic curve of a dataset, and to estimate the area under the curve. This is useful for analysing the efficacy of a variable as a predictor of a state of nature.

The mandatory VAR_LIST is the list of predictor variables. The variable STATE_VAR is the variable whose values represent the actual states, and STATE_VALUE is the value of this variable which represents the positive state.

The optional subcommand PLOT is used to determine if and how the ROC curve is drawn. The keyword CURVE means that the ROC curve should be drawn, and the optional keyword REFERENCE, which should be enclosed in parentheses, says that the diagonal reference line should be drawn. If the keyword NONE is given, then no ROC curve is drawn. By default, the curve is drawn with no reference line.

The optional subcommand PRINT determines which additional tables should be printed. Two additional tables are available. The SE keyword says that standard error of the area under the curve should be printed as well as the area itself. In addition, a p-value for the null hypothesis that the area under the curve equals 0.5 is printed. The COORDINATES keyword says that a table of coordinates of the ROC curve should be printed.

The CRITERIA subcommand has four optional parameters:

  • The TESTPOS parameter may be LARGE or SMALL. LARGE is the default, and says that larger values in the predictor variables are to be considered positive. SMALL indicates that smaller values should be considered positive.

  • The CI parameter specifies the confidence interval that should be printed. It has no effect if the SE keyword in the PRINT subcommand has not been given.

  • The DISTRIBUTION parameter determines the method to be used when estimating the area under the curve. There are two possibilities, viz: FREE and NEGEXPO. The FREE method uses a non-parametric estimate, and the NEGEXPO method a bi-negative exponential distribution estimate. The NEGEXPO method should only be used when the number of positive actual states is equal to the number of negative actual states. The default is FREE.

  • The CUTOFF parameter is for compatibility and is ignored.

The MISSING subcommand determines whether user missing values are to be included or excluded in the analysis. The default behaviour is to exclude them. Cases are excluded on a listwise basis; if any of the variables in VAR_LIST or if the variable STATE_VAR is missing, then the entire case is excluded.

Matrices

Some PSPP procedures work with matrices by producing numeric matrices that report results of data analysis, or by consuming matrices as a basis for further analysis. This chapter documents the format of data files that store these matrices and commands for working with them, as well as PSPP's general-purpose facility for matrix operations.

Matrix Files

A matrix file is an SPSS system file that conforms to the dictionary and case structure described in this section. Procedures that read matrices from files expect them to be in the matrix file format. Procedures that write matrices also use this format.

Text files that contain matrices can be converted to matrix file format. The MATRIX DATA command can read a text file as a matrix file.

A matrix file's dictionary must have the following variables in the specified order:

  1. Zero or more numeric split variables. These are included by procedures when SPLIT FILE is active. MATRIX DATA assigns split variables format F4.0.

  2. ROWTYPE_, a string variable with width 8. This variable indicates the kind of matrix or vector that a given case represents. The supported row types are listed below.

  3. Zero or more numeric factor variables. These are included by procedures that divide data into cells. For within-cell data, factor variables are filled with non-missing values; for pooled data, they are missing. MATRIX DATA assigns factor variables format F4.0.

  4. VARNAME_, a string variable. Matrix data includes one row per continuous variable (see below), naming each continuous variable in order. This column is blank for vector data. MATRIX DATA makes VARNAME_ wide enough for the name of any of the continuous variables, but at least 8 bytes.

  5. One or more numeric continuous variables. These are the variables whose data was analyzed to produce the matrices. MATRIX DATA assigns continuous variables format F10.4.

Case weights are ignored in matrix files.

Row Types

Matrix files support a fixed set of types of matrix and vector data. The ROWTYPE_ variable in each case of a matrix file indicates its row type.

The supported matrix row types are listed below. Each type is listed with the keyword that identifies it in ROWTYPE_. All supported types of matrices are square, meaning that each matrix must include one row per continuous variable, with the VARNAME_ variable indicating each continuous variable in turn in the same order as the dictionary.

  • CORR
    Correlation coefficients.

  • COV
    Covariance coefficients.

  • MAT
    General-purpose matrix.

  • N_MATRIX
    Counts.

  • PROX
    Proximities matrix.

The supported vector row types are listed below, along with their associated keyword. Vector row types only require a single row, whose VARNAME_ is blank:

  • COUNT
    Unweighted counts.

  • DFE
    Degrees of freedom.

  • MEAN
    Means.

  • MSE
    Mean squared errors.

  • N
    Counts.

  • STDDEV
    Standard deviations.

Only the row types listed above may appear in matrix files. The MATRIX DATA command, however, accepts the additional row types listed below, which it changes into matrix file row types as part of its conversion process:

  • N_VECTOR
    Synonym for N.

  • SD
    Synonym for STDDEV.

  • N_SCALAR
    Accepts a single number from the MATRIX DATA input and writes it as an N row with the number replicated across all the continuous variables.

MATRIX DATA

MATRIX DATA
        VARIABLES=VARIABLES
        [FILE={'FILE_NAME' | INLINE}
        [/FORMAT=[{LIST | FREE}]
                 [{UPPER | LOWER | FULL}]
                 [{DIAGONAL | NODIAGONAL}]]
        [/SPLIT=SPLIT_VARS]
        [/FACTORS=FACTOR_VARS]
        [/N=N]

The following subcommands are only needed when ROWTYPE_ is not
specified on the VARIABLES subcommand:
        [/CONTENTS={CORR,COUNT,COV,DFE,MAT,MEAN,MSE,
                    N_MATRIX,N|N_VECTOR,N_SCALAR,PROX,SD|STDDEV}]
        [/CELLS=N_CELLS]

The MATRIX DATA command convert matrices and vectors from text format into the matrix file format for use by procedures that read matrices. It reads a text file or inline data and outputs to the active file, replacing any data already in the active dataset. The matrix file may then be used by other commands directly from the active file, or it may be written to a .sav file using the SAVE command.

The text data read by MATRIX DATA can be delimited by spaces or commas. A plus or minus sign, except immediately following a d or e, also begins a new value. Optionally, values may be enclosed in single or double quotes.

MATRIX DATA can read the types of matrix and vector data supported in matrix files (see Row Types).

The FILE subcommand specifies the source of the command's input. To read input from a text file, specify its name in quotes. To supply input inline, omit FILE or specify INLINE. Inline data must directly follow MATRIX DATA, inside BEGIN DATA.

VARIABLES is the only required subcommand. It names the variables present in each input record in the order that they appear. (MATRIX DATA reorders the variables in the matrix file it produces, if needed to fit the matrix file format.) The variable list must include split variables and factor variables, if they are present in the data, in addition to the continuous variables that form matrix rows and columns. It may also include a special variable named ROWTYPE_.

Matrix data may include split variables or factor variables or both. List split variables, if any, on the SPLIT subcommand and factor variables, if any, on the FACTORS subcommand. Split and factor variables must be numeric. Split and factor variables must also be listed on VARIABLES, with one exception: if VARIABLES does not include ROWTYPE_, then SPLIT may name a single variable that is not in VARIABLES (see Example 8).

The FORMAT subcommand accepts settings to describe the format of the input data:

  • LIST (default)
    FREE

    LIST requires each row to begin at the start of a new input line. FREE allows rows to begin in the middle of a line. Either setting allows a single row to continue across multiple input lines.

  • LOWER (default)
    UPPER
    FULL

    With LOWER, only the lower triangle is read from the input data and the upper triangle is mirrored across the main diagonal. UPPER behaves similarly for the upper triangle. FULL reads the entire matrix.

  • DIAGONAL (default)
    NODIAGONAL

    With DIAGONAL, the main diagonal is read from the input data. With NODIAGONAL, which is incompatible with FULL, the main diagonal is not read from the input data but instead set to 1 for correlation matrices and system-missing for others.

The N subcommand is a way to specify the size of the population. It is equivalent to specifying an N vector with the specified value for each split file.

MATRIX DATA supports two different ways to indicate the kinds of matrices and vectors present in the data, depending on whether a variable with the special name ROWTYPE_ is present in VARIABLES. The following subsections explain MATRIX DATA syntax and behavior in each case.

With ROWTYPE_

If VARIABLES includes ROWTYPE_, each case's ROWTYPE_ indicates the type of data contained in the row. See Row Types for a list of supported row types.

Example 1: Defaults with ROWTYPE_

This example shows a simple use of MATRIX DATA with ROWTYPE_ plus 8 variables named var01 through var08.

Because ROWTYPE_ is the first variable in VARIABLES, it appears first on each line. The first three lines in the example data have ROWTYPE_ values of MEAN, SD, and N. These indicate that these lines contain vectors of means, standard deviations, and counts, respectively, for var01 through var08 in order.

The remaining 8 lines have a ROWTYPE_ of CORR which indicates that the values are correlation coefficients. Each of the lines corresponds to a row in the correlation matrix: the first line is for var01, the next line for var02, and so on. The input only contains values for the lower triangle, including the diagonal, since FORMAT=LOWER DIAGONAL is the default.

With ROWTYPE_, the CONTENTS subcommand is optional and the CELLS subcommand may not be used.

MATRIX DATA
    VARIABLES=ROWTYPE_ var01 TO var08.
BEGIN DATA.
MEAN  24.3   5.4  69.7  20.1  13.4   2.7  27.9   3.7
SD     5.7   1.5  23.5   5.8   2.8   4.5   5.4   1.5
N       92    92    92    92    92    92    92    92
CORR  1.00
CORR   .18  1.00
CORR  -.22  -.17  1.00
CORR   .36   .31  -.14  1.00
CORR   .27   .16  -.12   .22  1.00
CORR   .33   .15  -.17   .24   .21  1.00
CORR   .50   .29  -.20   .32   .12   .38  1.00
CORR   .17   .29  -.05   .20   .27   .20   .04  1.00
END DATA.

Example 2: FORMAT=UPPER NODIAGONAL

This syntax produces the same matrix file as example 1, but it uses FORMAT=UPPER NODIAGONAL to specify the upper triangle and omit the diagonal. Because the matrix's ROWTYPE_ is CORR, PSPP automatically fills in the diagonal with 1.

MATRIX DATA
    VARIABLES=ROWTYPE_ var01 TO var08
    /FORMAT=UPPER NODIAGONAL.
BEGIN DATA.
MEAN  24.3   5.4  69.7  20.1  13.4   2.7  27.9   3.7
SD     5.7   1.5  23.5   5.8   2.8   4.5   5.4   1.5
N       92    92    92    92    92    92    92    92
CORR         .17   .50  -.33   .27   .36  -.22   .18
CORR               .29   .29  -.20   .32   .12   .38
CORR                     .05   .20  -.15   .16   .21
CORR                           .20   .32  -.17   .12
CORR                                 .27   .12  -.24
CORR                                      -.20  -.38
CORR                                             .04
END DATA.

Example 3: N subcommand

This syntax uses the N subcommand in place of an N vector. It produces the same matrix file as examples 1 and 2.

MATRIX DATA
    VARIABLES=ROWTYPE_ var01 TO var08
    /FORMAT=UPPER NODIAGONAL
    /N 92.
BEGIN DATA.
MEAN  24.3   5.4  69.7  20.1  13.4   2.7  27.9   3.7
SD     5.7   1.5  23.5   5.8   2.8   4.5   5.4   1.5
CORR         .17   .50  -.33   .27   .36  -.22   .18
CORR               .29   .29  -.20   .32   .12   .38
CORR                     .05   .20  -.15   .16   .21
CORR                           .20   .32  -.17   .12
CORR                                 .27   .12  -.24
CORR                                      -.20  -.38
CORR                                             .04
END DATA.

Example 4: Split variables

This syntax defines two matrices, using the variable s1 to distinguish between them. Notice how the order of variables in the input matches their order on VARIABLES. This example also uses FORMAT=FULL.

MATRIX DATA
    VARIABLES=s1 ROWTYPE_  var01 TO var04
    /SPLIT=s1
    /FORMAT=FULL.
BEGIN DATA.
0 MEAN 34 35 36 37
0 SD   22 11 55 66
0 N    99 98 99 92
0 CORR  1 .9 .8 .7
0 CORR .9  1 .6 .5
0 CORR .8 .6  1 .4
0 CORR .7 .5 .4  1
1 MEAN 44 45 34 39
1 SD   23 15 51 46
1 N    98 34 87 23
1 CORR  1 .2 .3 .4
1 CORR .2  1 .5 .6
1 CORR .3 .5  1 .7
1 CORR .4 .6 .7  1
END DATA.

Example 5: Factor variables

This syntax defines a matrix file that includes a factor variable f1. The data includes mean, standard deviation, and count vectors for two values of the factor variable, plus a correlation matrix for pooled data.

MATRIX DATA
    VARIABLES=ROWTYPE_ f1 var01 TO var04
    /FACTOR=f1.
BEGIN DATA.
MEAN 0 34 35 36 37
SD   0 22 11 55 66
N    0 99 98 99 92
MEAN 1 44 45 34 39
SD   1 23 15 51 46
N    1 98 34 87 23
CORR .  1
CORR . .9  1
CORR . .8 .6  1
CORR . .7 .5 .4  1
END DATA.

Without ROWTYPE_

If VARIABLES does not contain ROWTYPE_, the CONTENTS subcommand defines the row types that appear in the file and their order. If CONTENTS is omitted, CONTENTS=CORR is assumed.

Factor variables without ROWTYPE_ introduce special requirements, illustrated below in Examples 8 and 9.

Example 6: Defaults without ROWTYPE_

This example shows a simple use of MATRIX DATA with 8 variables named var01 through var08, without ROWTYPE_. This yields the same matrix file as Example 1.

MATRIX DATA
    VARIABLES=var01 TO var08
   /CONTENTS=MEAN SD N CORR.
BEGIN DATA.
24.3   5.4  69.7  20.1  13.4   2.7  27.9   3.7
 5.7   1.5  23.5   5.8   2.8   4.5   5.4   1.5
  92    92    92    92    92    92    92    92
1.00
 .18  1.00
-.22  -.17  1.00
 .36   .31  -.14  1.00
 .27   .16  -.12   .22  1.00
 .33   .15  -.17   .24   .21  1.00
 .50   .29  -.20   .32   .12   .38  1.00
 .17   .29  -.05   .20   .27   .20   .04  1.00
END DATA.

Example 7: Split variables with explicit values

This syntax defines two matrices, using the variable s1 to distinguish between them. Each line of data begins with s1. This yields the same matrix file as Example 4.

MATRIX DATA
    VARIABLES=s1 var01 TO var04
    /SPLIT=s1
    /FORMAT=FULL
    /CONTENTS=MEAN SD N CORR.
BEGIN DATA.
0 34 35 36 37
0 22 11 55 66
0 99 98 99 92
0  1 .9 .8 .7
0 .9  1 .6 .5
0 .8 .6  1 .4
0 .7 .5 .4  1
1 44 45 34 39
1 23 15 51 46
1 98 34 87 23
1  1 .2 .3 .4
1 .2  1 .5 .6
1 .3 .5  1 .7
1 .4 .6 .7  1
END DATA.

Example 8: Split variable with sequential values

Like this previous example, this syntax defines two matrices with split variable s1. In this case, though, s1 is not listed in VARIABLES, which means that its value does not appear in the data. Instead, MATRIX DATA reads matrix data until the input is exhausted, supplying 1 for the first split, 2 for the second, and so on.

MATRIX DATA
    VARIABLES=var01 TO var04
    /SPLIT=s1
    /FORMAT=FULL
    /CONTENTS=MEAN SD N CORR.
BEGIN DATA.
34 35 36 37
22 11 55 66
99 98 99 92
 1 .9 .8 .7
.9  1 .6 .5
.8 .6  1 .4
.7 .5 .4  1
44 45 34 39
23 15 51 46
98 34 87 23
 1 .2 .3 .4
.2  1 .5 .6
.3 .5  1 .7
.4 .6 .7  1
END DATA.

Factor variables without ROWTYPE_

Without ROWTYPE_, factor variables introduce two new wrinkles to MATRIX DATA syntax. First, the CELLS subcommand must declare the number of combinations of factor variables present in the data. If there is, for example, one factor variable for which the data contains three values, one would write CELLS=3; if there are two (or more) factor variables for which the data contains five combinations, one would use CELLS=5; and so on.

Second, the CONTENTS subcommand must distinguish within-cell data from pooled data by enclosing within-cell row types in parentheses. When different within-cell row types for a single factor appear in subsequent lines, enclose the row types in a single set of parentheses; when different factors' values for a given within-cell row type appear in subsequent lines, enclose each row type in individual parentheses.

Without ROWTYPE_, input lines for pooled data do not include factor values, not even as missing values, but input lines for within-cell data do.

The following examples aim to clarify this syntax.

Example 9: Factor variables, grouping within-cell records by factor

This syntax defines the same matrix file as Example 5, without using ROWTYPE_. It declares CELLS=2 because the data contains two values (0 and 1) for factor variable f1. Within-cell vector row types MEAN, SD, and N are in a single set of parentheses on CONTENTS because they are grouped together in subsequent lines for a single factor value. The data lines with the pooled correlation matrix do not have any factor values.

MATRIX DATA
    VARIABLES=f1 var01 TO var04
    /FACTOR=f1
    /CELLS=2
    /CONTENTS=(MEAN SD N) CORR.
BEGIN DATA.
0 34 35 36 37
0 22 11 55 66
0 99 98 99 92
1 44 45 34 39
1 23 15 51 46
1 98 34 87 23
   1
  .9  1
  .8 .6  1
  .7 .5 .4  1
END DATA.

Example 10: Factor variables, grouping within-cell records by row type

This syntax defines the same matrix file as the previous example. The only difference is that the within-cell vector rows are grouped differently: two rows of means (one for each factor), followed by two rows of standard deviations, followed by two rows of counts.

MATRIX DATA
    VARIABLES=f1 var01 TO var04
    /FACTOR=f1
    /CELLS=2
    /CONTENTS=(MEAN) (SD) (N) CORR.
BEGIN DATA.
0 34 35 36 37
1 44 45 34 39
0 22 11 55 66
1 23 15 51 46
0 99 98 99 92
1 98 34 87 23
   1
  .9  1
  .8 .6  1
  .7 .5 .4  1
END DATA.

MCONVERT

MCONVERT
    [[MATRIX=]
     [IN({‘*’|'FILE'})]
     [OUT({‘*’|'FILE'})]]
    [/{REPLACE,APPEND}].

The MCONVERT command converts matrix data from a correlation matrix and a vector of standard deviations into a covariance matrix, or vice versa.

By default, MCONVERT both reads and writes the active file. Use the MATRIX subcommand to specify other files. To read a matrix file, specify its name inside parentheses following IN. To write a matrix file, specify its name inside parentheses following OUT. Use * to explicitly specify the active file for input or output.

When MCONVERT reads the input, by default it substitutes a correlation matrix and a vector of standard deviations each time it encounters a covariance matrix, and vice versa. Specify /APPEND to instead have MCONVERT add the other form of data without removing the existing data. Use /REPLACE to explicitly request removing the existing data.

The MCONVERT command requires its input to be a matrix file. Use MATRIX DATA to convert text input into matrix file format.

MATRIX…END MATRIX

Summary

MATRIX.
…matrix commands…
END MATRIX.

The following basic matrix commands are supported:

COMPUTE variable[(index[,index])]=expression.
CALL procedure(argument, …).
PRINT [expression]
      [/FORMAT=format]
      [/TITLE=title]
      [/SPACE={NEWPAGE | n}]
      [{/RLABELS=string… | /RNAMES=expression}]
      [{/CLABELS=string… | /CNAMES=expression}].

The following matrix commands offer support for flow control:

DO IF expression.
  …matrix commands…
[ELSE IF expression.
  …matrix commands…]…
[ELSE
  …matrix commands…]
END IF.

LOOP [var=first TO last [BY step]] [IF expression].
  …matrix commands…
END LOOP [IF expression].

BREAK.

The following matrix commands support matrix input and output:

READ variable[(index[,index])]
     [/FILE=file]
     /FIELD=first TO last [BY width]
     [/FORMAT=format]
     [/SIZE=expression]
     [/MODE={RECTANGULAR | SYMMETRIC}]
     [/REREAD].
WRITE expression
      [/OUTFILE=file]
      /FIELD=first TO last [BY width]
      [/MODE={RECTANGULAR | TRIANGULAR}]
      [/HOLD]
      [/FORMAT=format].
GET variable[(index[,index])]
    [/FILE={file | *}]
    [/VARIABLES=variable…]
    [/NAMES=expression]
    [/MISSING={ACCEPT | OMIT | number}]
    [/SYSMIS={OMIT | number}].
SAVE expression
     [/OUTFILE={file | *}]
     [/VARIABLES=variable…]
     [/NAMES=expression]
     [/STRINGS=variable…].
MGET [/FILE=file]
     [/TYPE={COV | CORR | MEAN | STDDEV | N | COUNT}].
MSAVE expression
      /TYPE={COV | CORR | MEAN | STDDEV | N | COUNT}
      [/OUTFILE=file]
      [/VARIABLES=variable…]
      [/SNAMES=variable…]
      [/SPLIT=expression]
      [/FNAMES=variable…]
      [/FACTOR=expression].

The following matrix commands provide additional support:

DISPLAY [{DICTIONARY | STATUS}].
RELEASE variable….

MATRIX and END MATRIX enclose a special PSPP sub-language, called the matrix language. The matrix language does not require an active dataset to be defined and only a few of the matrix language commands work with any datasets that are defined. Each instance of MATRIX…END MATRIX is a separate program whose state is independent of any instance, so that variables declared within a matrix program are forgotten at its end.

The matrix language works with matrices, where a "matrix" is a rectangular array of real numbers. An N×M matrix has N rows and M columns. Some special cases are important: a N×1 matrix is a "column vector", a 1×N is a "row vector", and a 1×1 matrix is a "scalar".

The matrix language also has limited support for matrices that contain 8-byte strings instead of numbers. Strings longer than 8 bytes are truncated, and shorter strings are padded with spaces. String matrices are mainly useful for labeling rows and columns when printing numerical matrices with the MATRIX PRINT command. Arithmetic operations on string matrices will not produce useful results. The user should not mix strings and numbers within a matrix.

The matrix language does not work with cases. A variable in the matrix language represents a single matrix.

The matrix language does not support missing values.

MATRIX is a procedure, so it cannot be enclosed inside DO IF, LOOP, etc.

Macros defined before a matrix program may be used within a matrix program, and macros may expand to include entire matrix programs. The DEFINE command to define new macros may not appear within a matrix program.

The following sections describe the details of the matrix language: first, the syntax of matrix expressions, then each of the supported commands. The COMMENT command is also supported.

Matrix Expressions

Many matrix commands use expressions. A matrix expression may use the following operators, listed in descending order of operator precedence. Within a single level, operators associate from left to right.

The operators are described in more detail below. Matrix Functions documents matrix functions.

Expressions appear in the matrix language in some contexts where there would be ambiguity whether / is an operator or a separator between subcommands. In these contexts, only the operators with higher precedence than / are allowed outside parentheses. Later sections call these "restricted expressions".

Matrix Construction Operator {}

Use the {} operator to construct matrices. Within the curly braces, commas separate elements within a row and semicolons separate rows. The following examples show a 2×3 matrix, a 1×4 row vector, a 3×1 column vector, and a scalar.

{1, 2, 3; 4, 5, 6}            ⇒    [1 2 3]
                                   [4 5 6]  
{3.14, 6.28, 9.24, 12.57}     ⇒    [3.14 6.28 9.42 12.57]  
{1.41; 1.73; 2}               ⇒    [1.41]
                                   [1.73]
                                   [2.00]  
{5}                           ⇒    5

Curly braces are not limited to holding numeric literals. They can contain calculations, and they can paste together matrices and vectors in any way as long as the result is rectangular. For example, if m is matrix {1, 2; 3, 4}, r is row vector {5, 6}, and c is column vector {7, 8}, then curly braces can be used as follows:

{m, c; r, 10}                 ⇒    [1 2 7]
                                   [3 4 8]
                                   [5 6 10]  
{c, 2 * c, T(r)}              ⇒    [7 14 5]
                                   [8 16 6]

The final example above uses the transposition function T.

Integer Sequence Operator :

The syntax FIRST:LAST:STEP yields a row vector of consecutive integers from FIRST to LAST counting by STEP. The final :STEP is optional and defaults to 1 when omitted.

FIRST, LAST, and STEP must each be a scalar and should be an integer (any fractional part is discarded). Because : has a high precedence, operands other than numeric literals must usually be parenthesized.

When STEP is positive (or omitted) and END < START, or if STEP is negative and END > START, then the result is an empty matrix. If STEP is 0, then PSPP reports an error.

Here are some examples:

1:6                           ⇒    {1, 2, 3, 4, 5, 6}
1:6:2                         ⇒    {1, 3, 5}
-1:-5:-1                      ⇒    {-1, -2, -3, -4, -5}
-1:-5                         ⇒    {}
2:1:0                         ⇒    (error)

Index Operator ()

The result of the submatrix or indexing operator, written M(RINDEX, CINDEX), contains the rows of M whose indexes are given in vector RINDEX and the columns whose indexes are given in vector CINDEX.

In the simplest case, if RINDEX and CINDEX are both scalars, the result is also a scalar:

{10, 20; 30, 40}(1, 1)        ⇒    10
{10, 20; 30, 40}(1, 2)        ⇒    20
{10, 20; 30, 40}(2, 1)        ⇒    30
{10, 20; 30, 40}(2, 2)        ⇒    40

If the index arguments have multiple elements, then the result includes multiple rows or columns:

{10, 20; 30, 40}(1:2, 1)      ⇒    {10; 30}
{10, 20; 30, 40}(2, 1:2)      ⇒    {30, 40}
{10, 20; 30, 40}(1:2, 1:2)    ⇒    {10, 20; 30, 40}

The special argument : may stand in for all the rows or columns in the matrix being indexed, like this:

{10, 20; 30, 40}(:, 1)        ⇒    {10; 30}
{10, 20; 30, 40}(2, :)        ⇒    {30, 40}
{10, 20; 30, 40}(:, :)        ⇒    {10, 20; 30, 40}

The index arguments do not have to be in order, and they may contain repeated values, like this:

{10, 20; 30, 40}({2, 1}, 1)   ⇒    {30; 10}
{10, 20; 30, 40}(2, {2; 2;    ⇒    {40, 40, 30}
1})
{10, 20; 30, 40}(2:1:-1, :)   ⇒    {30, 40; 10, 20}

When the matrix being indexed is a row or column vector, only a single index argument is needed, like this:

{11, 12, 13, 14, 15}(2:4)     ⇒    {12, 13, 14}
{11; 12; 13; 14; 15}(2:4)     ⇒    {12; 13; 14}

When an index is not an integer, PSPP discards the fractional part. It is an error for an index to be less than 1 or greater than the number of rows or columns:

{11, 12, 13, 14}({2.5,        ⇒    {12, 14}
4.6})
{11; 12; 13; 14}(0)           ⇒    (error)

Unary Operators

The unary operators take a single operand of any dimensions and operate on each of its elements independently. The unary operators are:

  • -: Inverts the sign of each element.
  • +: No change.
  • NOT: Logical inversion: each positive value becomes 0 and each zero or negative value becomes 1.

Examples:

-{1, -2; 3, -4}               ⇒    {-1, 2; -3, 4}
+{1, -2; 3, -4}               ⇒    {1, -2; 3, -4}
NOT {1, 0; -1, 1}             ⇒    {0, 1; 1, 0}

Elementwise Binary Operators

The elementwise binary operators require their operands to be matrices with the same dimensions. Alternatively, if one operand is a scalar, then its value is treated as if it were duplicated to the dimensions of the other operand. The result is a matrix of the same size as the operands, in which each element is the result of the applying the operator to the corresponding elements of the operands.

The elementwise binary operators are listed below.

  • The arithmetic operators, for familiar arithmetic operations:

    • +: Addition.

    • -: Subtraction.

    • *: Multiplication, if one operand is a scalar. (Otherwise this is matrix multiplication, described below.)

    • / or &/: Division.

    • &*: Multiplication.

    • &**: Exponentiation.

  • The relational operators, whose results are 1 when a comparison is true and 0 when it is false:

    • < or LT: Less than.

    • <= or LE: Less than or equal.

    • = or EQ: Equal.

    • > or GT: Greater than.

    • >= or GE: Greater than or equal.

    • <> or ~= or NE: Not equal.

  • The logical operators, which treat positive operands as true and nonpositive operands as false. They yield 0 for false and 1 for true:

    • AND: True if both operands are true.

    • OR: True if at least one operand is true.

    • XOR: True if exactly one operand is true.

Examples:

1 + 2                         ⇒    3
1 + {3; 4}                    ⇒    {4; 5}
{66, 77; 88, 99} + 5          ⇒    {71, 82; 93, 104}
{4, 8; 3, 7} + {1, 0; 5, 2}   ⇒    {5, 8; 8, 9}
{1, 2; 3, 4} < {4, 3; 2, 1}   ⇒    {1, 1; 0, 0}
{1, 3; 2, 4} >= 3             ⇒    {0, 1; 0, 1}
{0, 0; 1, 1} AND {0, 1; 0,    ⇒    {0, 0; 0, 1}
1}

Matrix Multiplication Operator *

If A is an M×N matrix and B is an N×P matrix, then A*B is the M×P matrix multiplication product C. PSPP reports an error if the number of columns in A differs from the number of rows in B.

The * operator performs elementwise multiplication (see above) if one of its operands is a scalar.

No built-in operator yields the inverse of matrix multiplication. Instead, multiply by the result of INV or GINV.

Some examples:

{1, 2, 3} * {4; 5; 6}         ⇒    32
{4; 5; 6} * {1, 2, 3}         ⇒    {4,  8, 12;
                                    5, 10, 15;
                                    6, 12, 18}

Matrix Exponentiation Operator **

The result of A**B is defined as follows when A is a square matrix and B is an integer scalar:

  • For B > 0, A**B is A*…*A, where there are B As. (PSPP implements this efficiently for large B, using exponentiation by squaring.)

  • For B < 0, A**B is INV(A**(-B)).

  • For B = 0, A**B is the identity matrix.

PSPP reports an error if A is not square or B is not an integer.

Examples:

{2, 5; 1, 4}**3               ⇒    {48, 165; 33, 114}
{2, 5; 1, 4}**0               ⇒    {1, 0; 0, 1}
10*{4, 7; 2, 6}**-1           ⇒    {6, -7; -2, 4}

Matrix Functions

The matrix language support numerous functions in multiple categories. The following subsections document each of the currently supported functions. The first letter of each parameter's name indicate the required argument type:

  • S: A scalar.

  • N: A nonnegative integer scalar. (Non-integers are accepted and silently rounded down to the nearest integer.)

  • V: A row or column vector.

  • M: A matrix.

Elementwise Functions

These functions act on each element of their argument independently, like the elementwise operators.

  • ABS(M)
    Takes the absolute value of each element of M.

    ABS({-1, 2; -3, 0}) ⇒ {1, 2; 3, 0}

  • ARSIN(M)
    ARTAN(M)
    Computes the inverse sine or tangent, respectively, of each element in M. The results are in radians, between \(-\pi/2\) and \(+\pi/2\), inclusive.

    The value of \(\pi\) can be computed as 4*ARTAN(1).

    ARSIN({-1, 0, 1}) ⇒ {-1.57, 0, 1.57} (approximately)

ARTAN({-5, -1, 1, 5}) ⇒ {-1.37, -.79, .79, 1.37} (approximately)

  • COS(M)
    SIN(M)
    Computes the cosine or sine, respectively, of each element in M, which must be in radians.

    COS({0.785, 1.57; 3.14, 1.57 + 3.14}) ⇒ {.71, 0; -1, 0}
(approximately)

  • EXP(M)
    Computes \(e^x\) for each element \(x\) in M.

    EXP({2, 3; 4, 5}) ⇒ {7.39, 20.09; 54.6, 148.4} (approximately)

  • LG10(M)
    LN(M)
    Takes the logarithm with base 10 or base \(e\), respectively, of each element in M.

    LG10({1, 10, 100, 1000}) ⇒ {0, 1, 2, 3}
LG10(0) ⇒ (error)

LN({EXP(1), 1, 2, 3, 4}) ⇒ {1, 0, .69, 1.1, 1.39} (approximately)
LN(0) ⇒ (error)

  • MOD(M, S)
    Takes each element in M modulo nonzero scalar value S, that is, the remainder of division by S. The sign of the result is the same as the sign of the dividend.

    MOD({5, 4, 3, 2, 1, 0}, 3) ⇒ {2, 1, 0, 2, 1, 0}
MOD({5, 4, 3, 2, 1, 0}, -3) ⇒ {2, 1, 0, 2, 1, 0}
MOD({-5, -4, -3, -2, -1, 0}, 3) ⇒ {-2, -1, 0, -2, -1, 0}
MOD({-5, -4, -3, -2, -1, 0}, -3) ⇒ {-2, -1, 0, -2, -1, 0}
MOD({5, 4, 3, 2, 1, 0}, 1.5) ⇒ {.5, 1.0, .0, .5, 1.0, .0}
MOD({5, 4, 3, 2, 1, 0}, 0) ⇒ (error)

  • RND(M)
    TRUNC(M)
    Rounds each element of M to an integer. RND rounds to the nearest integer, with halves rounded to even integers, and TRUNC rounds toward zero.

    RND({-1.6, -1.5, -1.4}) ⇒ {-2, -2, -1}
RND({-.6, -.5, -.4}) ⇒ {-1, 0, 0}
RND({.4, .5, .6} ⇒ {0, 0, 1}
RND({1.4, 1.5, 1.6}) ⇒ {1, 2, 2}

TRUNC({-1.6, -1.5, -1.4}) ⇒ {-1, -1, -1}
TRUNC({-.6, -.5, -.4}) ⇒ {0, 0, 0}
TRUNC({.4, .5, .6} ⇒ {0, 0, 0}
TRUNC({1.4, 1.5, 1.6}) ⇒ {1, 1, 1}

  • SQRT(M)
    Takes the square root of each element of M, which must not be negative.

    SQRT({0, 1, 2, 4, 9, 81}) ⇒ {0, 1, 1.41, 2, 3, 9} (approximately)
SQRT(-1) ⇒ (error)

Logical Functions

  • ALL(M)
    Returns a scalar with value 1 if all of the elements in M are nonzero, or 0 if at least one element is zero.

    ALL({1, 2, 3} < {2, 3, 4}) ⇒ 1
ALL({2, 2, 3} < {2, 3, 4}) ⇒ 0
ALL({2, 3, 3} < {2, 3, 4}) ⇒ 0
ALL({2, 3, 4} < {2, 3, 4}) ⇒ 0

  • ANY(M)
    Returns a scalar with value 1 if any of the elements in M is nonzero, or 0 if all of them are zero.

    ANY({1, 2, 3} < {2, 3, 4}) ⇒ 1
ANY({2, 2, 3} < {2, 3, 4}) ⇒ 1
ANY({2, 3, 3} < {2, 3, 4}) ⇒ 1
ANY({2, 3, 4} < {2, 3, 4}) ⇒ 0

Matrix Construction Functions

  • BLOCK(M1, …, MN)
    Returns a block diagonal matrix with as many rows as the sum of its arguments' row counts and as many columns as the sum of their columns. Each argument matrix is placed along the main diagonal of the result, and all other elements are zero.

    BLOCK({1, 2; 3, 4}, 5, {7; 8; 9}, {10, 11}) ⇒
   1   2   0   0   0   0
   3   4   0   0   0   0
   0   0   5   0   0   0
   0   0   0   7   0   0
   0   0   0   8   0   0
   0   0   0   9   0   0
   0   0   0   0  10  11

  • IDENT(N)
    IDENT(NR, NC)
    Returns an identity matrix, whose main diagonal elements are one and whose other elements are zero. The returned matrix has N rows and columns or NR rows and NC columns, respectively.

    IDENT(1) ⇒ 1
IDENT(2) ⇒
  1  0
  0  1
IDENT(3, 5) ⇒
  1  0  0  0  0
  0  1  0  0  0
  0  0  1  0  0
IDENT(5, 3) ⇒
  1  0  0
  0  1  0
  0  0  1
  0  0  0
  0  0  0

  • MAGIC(N)
    Returns an N×N matrix that contains each of the integers 1…N once, in which each column, each row, and each diagonal sums to \(n(n^2+1)/2\). There are many magic squares with given dimensions, but this function always returns the same one for a given value of N.

    MAGIC(3) ⇒ {8, 1, 6; 3, 5, 7; 4, 9, 2}
MAGIC(4) ⇒ {1, 5, 12, 16; 15, 11, 6, 2; 14, 8, 9, 3; 4, 10, 7, 13}

  • MAKE(NR, NC, S)
    Returns an NR×NC matrix whose elements are all S.

    MAKE(1, 2, 3) ⇒ {3, 3}
MAKE(2, 1, 4) ⇒ {4; 4}
MAKE(2, 3, 5) ⇒ {5, 5, 5; 5, 5, 5}

  • MDIAG(V)
    Given N-element vector V, returns a N×N matrix whose main diagonal is copied from V. The other elements in the returned vector are zero.

    Use CALL SETDIAG to replace the main diagonal of a matrix in-place.

    MDIAG({1, 2, 3, 4}) ⇒
  1  0  0  0
  0  2  0  0
  0  0  3  0
  0  0  0  4

  • RESHAPE(M, NR, NC)
    Returns an NR×NC matrix whose elements come from M, which must have the same number of elements as the new matrix, copying elements from M to the new matrix row by row.

    RESHAPE(1:12, 1, 12) ⇒
   1   2   3   4   5   6   7   8   9  10  11  12
RESHAPE(1:12, 2, 6) ⇒
   1   2   3   4   5   6
   7   8   9  10  11  12
RESHAPE(1:12, 3, 4) ⇒
   1   2   3   4
   5   6   7   8
   9  10  11  12
RESHAPE(1:12, 4, 3) ⇒
   1   2   3
   4   5   6
   7   8   9
  10  11  12

  • T(M)
    TRANSPOS(M)
    Returns M with rows exchanged for columns.

    T({1, 2, 3}) ⇒ {1; 2; 3}
T({1; 2; 3}) ⇒ {1, 2, 3}

  • UNIFORM(NR, NC)
    Returns a NR×NC matrix in which each element is randomly chosen from a uniform distribution of real numbers between 0 and 1. Random number generation honors the current seed setting.

    The following example shows one possible output, but of course every result will be different (given different seeds):

    UNIFORM(4, 5)*10 ⇒
  7.71  2.99   .21  4.95  6.34
  4.43  7.49  8.32  4.99  5.83
  2.25   .25  1.98  7.09  7.61
  2.66  1.69  2.64   .88  1.50

Minimum, Maximum, and Sum Functions

  • CMIN(M)
    CMAX(M)
    CSUM(M)
    CSSQ(M)
    Returns a row vector with the same number of columns as M, in which each element is the minimum, maximum, sum, or sum of squares, respectively, of the elements in the same column of M.

    CMIN({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {1, 2, 3}
CMAX({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {7, 8, 9}
CSUM({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {12, 15, 18}
CSSQ({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {66, 93, 126}

  • MMIN(M)
    MMAX(M)
    MSUM(M)
    MSSQ(M)
    Returns the minimum, maximum, sum, or sum of squares, respectively, of the elements of M.

    MMIN({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ 1
MMAX({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ 9
MSUM({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ 45
MSSQ({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ 285

  • RMIN(M)
    RMAX(M)
    RSUM(M)
    RSSQ(M)
    Returns a column vector with the same number of rows as M, in which each element is the minimum, maximum, sum, or sum of squares, respectively, of the elements in the same row of M.

    RMIN({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {1; 4; 7}
RMAX({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {3; 6; 9}
RSUM({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {6; 15; 24}
RSSQ({1, 2, 3; 4, 5, 6; 7, 8, 9} ⇒ {14; 77; 194}

  • SSCP(M)
    Returns \({\bf M}^{\bf T} × \bf M\).

    SSCP({1, 2, 3; 4, 5, 6}) ⇒ {17, 22, 27; 22, 29, 36; 27, 36, 45}

  • TRACE(M)
    Returns the sum of the elements along M's main diagonal, equivalent to MSUM(DIAG(M)).

    TRACE(MDIAG(1:5)) ⇒ 15

Matrix Property Functions

  • NROW(M)
    NCOL(M)
    Returns the number of row or columns, respectively, in M.

    NROW({1, 0; -2, -3; 3, 3}) ⇒ 3
NROW(1:5) ⇒ 1

NCOL({1, 0; -2, -3; 3, 3}) ⇒ 2
NCOL(1:5) ⇒ 5

  • DIAG(M)
    Returns a column vector containing a copy of M's main diagonal. The vector's length is the lesser of NCOL(M) and NROW(M).

    DIAG({1, 0; -2, -3; 3, 3}) ⇒ {1; -3}

Matrix Rank Ordering Functions

The GRADE and RANK functions each take a matrix M and return a matrix R with the same dimensions. Each element in R ranges between 1 and the number of elements N in M, inclusive. When the elements in M all have unique values, both of these functions yield the same results: the smallest element in M corresponds to value 1 in R, the next smallest to 2, and so on, up to the largest to N. When multiple elements in M have the same value, these functions use different rules for handling the ties.

  • GRADE(M)
    Returns a ranking of M, turning duplicate values into sequential ranks. The returned matrix always contains each of the integers 1 through the number of elements in the matrix exactly once.

    GRADE({1, 0, 3; 3, 1, 2; 3, 0, 5}) ⇒ {3, 1, 6; 7, 4, 5; 8, 2, 9}

  • RNKORDER(M)
    Returns a ranking of M, turning duplicate values into the mean of their sequential ranks.

    RNKORDER({1, 0, 3; 3, 1, 2; 3, 0, 5})
 ⇒ {3.5, 1.5, 7; 7, 3.5, 5; 7, 1.5, 9}

One may use GRADE to sort a vector:

COMPUTE v(GRADE(v))=v.   /* Sort v in ascending order.
COMPUTE v(GRADE(-v))=v.  /* Sort v in descending order.

Matrix Algebra Functions

  • CHOL(M)
    Matrix M must be an N×N symmetric positive-definite matrix. Returns an N×N matrix B such that \({\bf B}^{\bf T}×{\bf B}=\bf M\).

    CHOL({4, 12, -16; 12, 37, -43; -16, -43, 98}) ⇒
  2  6 -8
  0  1  5
  0  0  3

  • DESIGN(M)
    Returns a design matrix for M. The design matrix has the same number of rows as M. Each column C in M, from left to right, yields a group of columns in the output. For each unique value V in C, from top to bottom, add a column to the output in which V becomes 1 and other values become 0.

    PSPP issues a warning if a column only contains a single unique value.

    DESIGN({1; 2; 3}) ⇒ {1, 0, 0; 0, 1, 0; 0, 0, 1}
DESIGN({5; 8; 5}) ⇒ {1, 0; 0, 1; 1, 0}
DESIGN({1, 5; 2, 8; 3, 5})
 ⇒ {1, 0, 0, 1, 0; 0, 1, 0, 0, 1; 0, 0, 1, 1, 0}
DESIGN({5; 5; 5}) ⇒ (warning)

  • DET(M)
    Returns the determinant of square matrix M.

    DET({3, 7; 1, -4}) ⇒ -19

  • EVAL(M)
    Returns a column vector containing the eigenvalues of symmetric matrix M, sorted in ascending order.

    Use CALL EIGEN to compute eigenvalues and eigenvectors of a matrix.

    EVAL({2, 0, 0; 0, 3, 4; 0, 4, 9}) ⇒ {11; 2; 1}

  • GINV(M)
    Returns the K×N matrix A that is the "generalized inverse" of N×K matrix M, defined such that \({\bf M}×{\bf A}×{\bf M}={\bf M}\) and \({\bf A}×{\bf M}×{\bf A}={\bf A}\).

    GINV({1, 2}) ⇒ {.2; .4} (approximately)
{1:9} * GINV(1:9) * {1:9} ⇒ {1:9} (approximately)

  • GSCH(M)
    M must be a N×M matrix, M ≥ N, with rank N. Returns an N×N orthonormal basis for M, obtained using the Gram-Schmidt process.

    GSCH({3, 2; 1, 2}) * SQRT(10) ⇒ {3, -1; 1, 3} (approximately)

  • INV(M)
    Returns the N×N matrix A that is the inverse of N×N matrix M, defined such that \({\bf M}×{\bf A} = {\bf A}×{\bf M} = {\bf I}\), where I is the identity matrix. M must not be singular, that is, \(\det({\bf M}) ≠ 0\).

    INV({4, 7; 2, 6}) ⇒ {.6, -.7; -.2, .4} (approximately)

  • KRONEKER(MA, MB)
    Returns the PM×QN matrix P that is the Kroneker product of M×N matrix MA and P×Q matrix MB. One may view P as the concatenation of multiple P×Q blocks, each of which is the scalar product of MB by a different element of MA. For example, when A is a 2×2 matrix, KRONEKER(A, B) is equivalent to {A(1,1)*B, A(1,2)*B; A(2,1)*B, A(2,2)*B}.

    KRONEKER({1, 2; 3, 4}, {0, 5; 6, 7}) ⇒
   0   5   0  10
   6   7  12  14
   0  15   0  20
  18  21  24  28

  • RANK(M)
    Returns the rank of matrix M, a integer scalar whose value is the dimension of the vector space spanned by its columns or, equivalently, by its rows.

    RANK({1, 0, 1; -2, -3, 1; 3, 3, 0}) ⇒ 2
RANK({1, 1, 0, 2; -1, -1, 0, -2}) ⇒ 1
RANK({1, -1; 1, -1; 0, 0; 2, -2}) ⇒ 1
RANK({1, 2, 1; -2, -3, 1; 3, 5, 0}) ⇒ 2
RANK({1, 0, 2; 2, 1, 0; 3, 2, 1}) ⇒ 3

  • SOLVE(MA, MB)
    MA must be an N×N matrix, with \(\det({\bf MA}) ≠ 0\), and MB an P×Q matrix. Returns an P×Q matrix X such that \({\bf MA} × {\bf X} = {\bf MB}\).

    All of the following examples show approximate results:

    SOLVE({2, 3; 4, 9}, {6, 2; 15, 5}) ⇒
   1.50    .50
   1.00    .33
SOLVE({1, 3, -2; 3, 5, 6; 2, 4, 3}, {5; 7; 8}) ⇒
 -15.00
   8.00
   2.00
SOLVE({2, 1, -1; -3, -1, 2; -2, 1, 2}, {8; -11; -3}) ⇒
   2.00
   3.00
  -1.00

  • SVAL(M)

    Given P×Q matrix M, returns a \(\min(N,K)\)-element column vector containing the singular values of M in descending order.

    Use CALL SVD to compute the full singular value decomposition of a matrix.

    SVAL({1, 1; 0, 0}) ⇒ {1.41; .00}
SVAL({1, 0, 1; 0, 1, 1; 0, 0, 0}) ⇒ {1.73; 1.00; .00}
SVAL({2, 4; 1, 3; 0, 0; 0, 0}) ⇒ {5.46; .37}

  • SWEEP(M, NK)
    Given P×Q matrix M and integer scalar \(k\) = NK such that \(1 ≤ k ≤ \min(R,C)\), returns the P×Q sweep matrix A.

    If \({\bf M}_{kk} ≠ 0\), then:

    $$ \begin{align} A_{kk} &= 1/M_{kk},\\ A_{ik} &= -M_{ik}/M_{kk} \text{ for } i ≠ k,\\ A_{kj} &= M_{kj}/M_{kk} \text{ for } j ≠ k,\\ A_{ij} &= M_{ij} - M_{ik}M_{kj}/M_{kk} \text{ for } i ≠ k \text{ and } j ≠ k. \end{align} $$

    If \({\bf M}_{kk}\) = 0, then:

    $$ \begin{align} A_{ik} &= A_{ki} = 0, \\ A_{ij} &= M_{ij}, \text{ for } i ≠ k \text{ and } j ≠ k. \end{align} $$

    Given M = {0, 1, 2; 3, 4, 5; 6, 7, 8}, then (approximately):

    SWEEP(M, 1) ⇒
   .00   .00   .00
   .00  4.00  5.00
   .00  7.00  8.00
SWEEP(M, 2) ⇒
  -.75  -.25   .75
   .75   .25  1.25
   .75 -1.75  -.75
SWEEP(M, 3) ⇒
 -1.50  -.75  -.25
  -.75  -.38  -.63
   .75   .88   .13

Matrix Statistical Distribution Functions

The matrix language can calculate several functions of standard statistical distributions using the same syntax and semantics as in PSPP transformation expressions. See Statistical Distribution Functions for details.

The matrix language extends the PDF, CDF, SIG, IDF, NPDF, and NCDF functions by allowing the first parameters to each of these functions to be a vector or matrix with any dimensions. In addition, CDF.BVNOR and PDF.BVNOR allow either or both of their first two parameters to be vectors or matrices; if both are non-scalar then they must have the same dimensions. In each case, the result is a matrix or vector with the same dimensions as the input populated with elementwise calculations.

EOF Function

This function works with files being used on the READ statement.

  • EOF(FILE)

    Given a file handle or file name FILE, returns an integer scalar 1 if the last line in the file has been read or 0 if more lines are available. Determining this requires attempting to read another line, which means that REREAD on the next READ command following EOF on the same file will be ineffective.

The EOF function gives a matrix program the flexibility to read a file with text data without knowing the length of the file in advance. For example, the following program will read all the lines of data in data.txt, each consisting of three numbers, as rows in matrix data:

MATRIX.
COMPUTE data={}.
LOOP IF NOT EOF('data.txt').
  READ row/FILE='data.txt'/FIELD=1 TO 1000/SIZE={1,3}.
  COMPUTE data={data; row}.
END LOOP.
PRINT data.
END MATRIX.

COMPUTE Command

COMPUTE variable[(index[,index])]=expression.

The COMPUTE command evaluates an expression and assigns the result to a variable or a submatrix of a variable. Assigning to a submatrix uses the same syntax as the index operator.

CALL Command

A matrix function returns a single result. The CALL command implements procedures, which take a similar syntactic form to functions but yield results by modifying their arguments rather than returning a value.

Output arguments to a CALL procedure must be a single variable name.

The following procedures are implemented via CALL to allow them to return multiple results. For these procedures, the output arguments need not name existing variables; if they do, then their previous values are replaced:

  • CALL EIGEN(M, EVEC, EVAL)

    Computes the eigenvalues and eigenvector of symmetric N×N matrix M. Assigns the eigenvectors of M to the columns of N×N matrix EVEC and the eigenvalues in descending order to N-element column vector EVAL.

    Use the EVAL function to compute just the eigenvalues of a symmetric matrix.

    For example, the following matrix language commands:

    CALL EIGEN({1, 0; 0, 1}, evec, eval).
PRINT evec.
PRINT eval.

CALL EIGEN({3, 2, 4; 2, 0, 2; 4, 2, 3}, evec2, eval2).
PRINT evec2.
PRINT eval2.

    yield this output:

    evec
  1  0
  0  1

eval
  1
  1

evec2
  -.6666666667   .0000000000   .7453559925
  -.3333333333  -.8944271910  -.2981423970
  -.6666666667   .4472135955  -.5962847940

eval2
  8.0000000000
 -1.0000000000
 -1.0000000000

  • CALL SVD(M, U, S, V)

    Computes the singular value decomposition of P×Q matrix M, assigning S a P×Q diagonal matrix and to U and V unitary P×Q matrices such that M = U×S×V^T. The main diagonal of Q contains the singular values of M.

    Use the SVAL function to compute just the singular values of a matrix.

    For example, the following matrix program:

    CALL SVD({3, 2, 2; 2, 3, -2}, u, s, v).
PRINT (u * s * T(v))/FORMAT F5.1.

    yields this output:

    (u * s * T(v))
   3.0   2.0   2.0
   2.0   3.0  -2.0

The final procedure is implemented via CALL to allow it to modify a matrix instead of returning a modified version. For this procedure, the output argument must name an existing variable.

  • CALL SETDIAG(M, V)

    Replaces the main diagonal of N×P matrix M by the contents of K-element vector V. If K = 1, so that V is a scalar, replaces all of the diagonal elements of M by V. If K < \min(N,P), only the upper K diagonal elements are replaced; if K > \min(N,P), then the extra elements of V are ignored.

    Use the MDIAG function to construct a new matrix with a specified main diagonal.

    For example, this matrix program:

    COMPUTE x={1, 2, 3; 4, 5, 6; 7, 8, 9}.
CALL SETDIAG(x, 10).
PRINT x.

    outputs the following:

    x
  10   2   3
   4  10   6
   7   8  10

PRINT [expression]
      [/FORMAT=format]
      [/TITLE=title]
      [/SPACE={NEWPAGE | n}]
      [{/RLABELS=string… | /RNAMES=expression}]
      [{/CLABELS=string… | /CNAMES=expression}].

The PRINT command is commonly used to display a matrix. It evaluates the restricted EXPRESSION, if present, and outputs it either as text or a pivot table, depending on the setting of MDISPLAY.

Use the FORMAT subcommand to specify a format, such as F8.2, for displaying the matrix elements. FORMAT is optional for numerical matrices. When it is omitted, PSPP chooses how to format entries automatically using \(m\), the magnitude of the largest-magnitude element in the matrix to be displayed:

  1. If \(m < 10^{11}\) and the matrix's elements are all integers, PSPP chooses the narrowest F format that fits \(m\) plus a sign. For example, if the matrix is {1:10}, then \(m = 10\), which fits in 3 columns with room for a sign, the format is F3.0.

  2. Otherwise, if \(m ≥ 10^9\) or \(m ≤ 10^{-4}\), PSPP scales all of the numbers in the matrix by \(10^x\), where \(x\) is the exponent that would be used to display \(m\) in scientific notation. For example, for \(m = 5.123×10^{20}\), the scale factor is \(10^{20}\). PSPP displays the scaled values in format F13.10 and notes the scale factor in the output.

  3. Otherwise, PSPP displays the matrix values, without scaling, in format F13.10.

The optional TITLE subcommand specifies a title for the output text or table, as a quoted string. When it is omitted, the syntax of the matrix expression is used as the title.

Use the SPACE subcommand to request extra space above the matrix output. With a numerical argument, it adds the specified number of lines of blank space above the matrix. With NEWPAGE as an argument, it prints the matrix at the top of a new page. The SPACE subcommand has no effect when a matrix is output as a pivot table.

The RLABELS and RNAMES subcommands, which are mutually exclusive, can supply a label to accompany each row in the output. With RLABELS, specify the labels as comma-separated strings or other tokens. With RNAMES, specify a single expression that evaluates to a vector of strings. Either way, if there are more labels than rows, the extra labels are ignored, and if there are more rows than labels, the extra rows are unlabeled. For output to a pivot table with RLABELS, the labels can be any length; otherwise, the labels are truncated to 8 bytes.

The CLABELS and CNAMES subcommands work for labeling columns as RLABELS and RNAMES do for labeling rows.

When the EXPRESSION is omitted, PRINT does not output a matrix. Instead, it outputs only the text specified on TITLE, if any, preceded by any space specified on the SPACE subcommand, if any. Any other subcommands are ignored, and the command acts as if MDISPLAY is set to TEXT regardless of its actual setting.

Example

The following syntax demonstrates two different ways to label the rows and columns of a matrix with PRINT:

MATRIX.
COMPUTE m={1, 2, 3; 4, 5, 6; 7, 8, 9}.
PRINT m/RLABELS=a, b, c/CLABELS=x, y, z.

COMPUTE rlabels={"a", "b", "c"}.
COMPUTE clabels={"x", "y", "z"}.
PRINT m/RNAMES=rlabels/CNAMES=clabels.
END MATRIX.

With MDISPLAY=TEXT (the default), this program outputs the following (twice):

m
                x        y        z
a               1        2        3
b               4        5        6
c               7        8        9

With SET MDISPLAY=TABLES. added above MATRIX., the output becomes the following (twice):

    m
┌─┬─┬─┬─┐
│ │x│y│z│
├─┼─┼─┼─┤
│a│1│2│3│
│b│4│5│6│
│c│7│8│9│
└─┴─┴─┴─┘

DO IF Command

DO IF expression.
  …matrix commands…
[ELSE IF expression.
  …matrix commands…]…
[ELSE
  …matrix commands…]
END IF.

A DO IF command evaluates its expression argument. If the DO IF expression evaluates to true, then PSPP executes the associated commands. Otherwise, PSPP evaluates the expression on each ELSE IF clause (if any) in order, and executes the commands associated with the first one that yields a true value. Finally, if the DO IF and all the ELSE IF expressions all evaluate to false, PSPP executes the commands following the ELSE clause (if any).

Each expression on DO IF and ELSE IF must evaluate to a scalar. Positive scalars are considered to be true, and scalars that are zero or negative are considered to be false.

Example

The following matrix language fragment sets b to the term following a in the Juggler sequence:

DO IF MOD(a, 2) = 0.
  COMPUTE b = TRUNC(a &** (1/2)).
ELSE.
  COMPUTE b = TRUNC(a &** (3/2)).
END IF.

LOOP and BREAK Commands

LOOP [var=first TO last [BY step]] [IF expression].
  …matrix commands…
END LOOP [IF expression].

BREAK.

The LOOP command executes a nested group of matrix commands, called the loop's "body", repeatedly. It has three optional clauses that control how many times the loop body executes. Regardless of these clauses, the global MXLOOPS setting, which defaults to 40, also limits the number of iterations of a loop. To iterate more times, raise the maximum with SET MXLOOPS outside of the MATRIX command.

The optional index clause causes VAR to be assigned successive values on each trip through the loop: first FIRST, then FIRST + STEP, then FIRST + 2 × STEP, and so on. The loop ends when VAR > LAST, for positive STEP, or VAR < LAST, for negative STEP. If STEP is not specified, it defaults to 1. All the index clause expressions must evaluate to scalars, and non-integers are rounded toward zero. If STEP evaluates as zero (or rounds to zero), then the loop body never executes.

The optional IF on LOOP is evaluated before each iteration through the loop body. If its expression, which must evaluate to a scalar, is zero or negative, then the loop terminates without executing the loop body.

The optional IF on END LOOP is evaluated after each iteration through the loop body. If its expression, which must evaluate to a scalar, is zero or negative, then the loop terminates.

Example

The following computes and prints \(l(n)\), whose value is the number of steps in the Juggler sequence for \(n\), for \( 2 \le n \le 10\):

COMPUTE l = {}.
LOOP n = 2 TO 10.
  COMPUTE a = n.
  LOOP i = 1 TO 100.
    DO IF MOD(a, 2) = 0.
      COMPUTE a = TRUNC(a &** (1/2)).
    ELSE.
      COMPUTE a = TRUNC(a &** (3/2)).
    END IF.
  END LOOP IF a = 1.
  COMPUTE l = {l; i}.
END LOOP.
PRINT l.

BREAK Command

The BREAK command may be used inside a loop body, ordinarily within a DO IF command. If it is executed, then the loop terminates immediately, jumping to the command just following END LOOP. When multiple LOOP commands nest, BREAK terminates the innermost loop.

Example

The following example is a revision of the one above that shows how BREAK could substitute for the index and IF clauses on LOOP and END LOOP:

COMPUTE l = {}.
LOOP n = 2 TO 10.
  COMPUTE a = n.
  COMPUTE i = 1.
  LOOP.
    DO IF MOD(a, 2) = 0.
      COMPUTE a = TRUNC(a &** (1/2)).
    ELSE.
      COMPUTE a = TRUNC(a &** (3/2)).
    END IF.
    DO IF a = 1.
      BREAK.
    END IF.
    COMPUTE i = i + 1.
  END LOOP.
  COMPUTE l = {l; i}.
END LOOP.
PRINT l.

READ and WRITE Commands

The READ and WRITE commands perform matrix input and output with text files. They share the following syntax for specifying how data is divided among input lines:

/FIELD=first TO last [BY width]
[/FORMAT=format]

Both commands require the FIELD subcommand. It specifies the range of columns, from FIRST to LAST, inclusive, that the data occupies on each line of the file. The leftmost column is column 1. The columns must be literal numbers, not expressions. To use entire lines, even if they might be very long, specify a column range such as 1 TO 100000.

The FORMAT subcommand is optional for numerical matrices. For string matrix input and output, specify an A format. In addition to FORMAT, the optional BY specification on FIELD determine the meaning of each text line:

  • With neither BY nor FORMAT, the numbers in the text file are in F format separated by spaces or commas. For WRITE, PSPP uses as many digits of precision as needed to accurately represent the numbers in the matrix.

  • BY width divides the input area into fixed-width fields with the given width. The input area must be a multiple of width columns wide. Numbers are read or written as Fwidth.0 format.

  • FORMAT="countF" divides the input area into integer count equal-width fields per line. The input area must be a multiple of count columns wide. Another format type may be substituted for F.

  • FORMAT=Fw[.d] divides the input area into fixed-width fields with width w. The input area must be a multiple of w columns wide. Another format type may be substituted for F. The READ command disregards d.

  • FORMAT=F specifies format F without indicating a field width. Another format type may be substituted for F. The WRITE command accepts this form, but it has no effect unless BY is also used to specify a field width.

If BY and FORMAT both specify or imply a field width, then they must indicate the same field width.

READ Command

READ variable[(index[,index])]
     [/FILE=file]
     /FIELD=first TO last [BY width]
     [/FORMAT=format]
     [/SIZE=expression]
     [/MODE={RECTANGULAR | SYMMETRIC}]
     [/REREAD].

The READ command reads from a text file into a matrix variable. Specify the target variable just after the command name, either just a variable name to create or replace an entire variable, or a variable name followed by an indexing expression to replace a submatrix of an existing variable.

The FILE subcommand is required in the first READ command that appears within MATRIX. It specifies the text file to be read, either as a file name in quotes or a file handle previously declared on FILE HANDLE. Later READ commands (in syntax order) use the previous referenced file if FILE is omitted.

The FIELD and FORMAT subcommands specify how input lines are interpreted. FIELD is required, but FORMAT is optional. See READ and WRITE Commands, for details.

The SIZE subcommand is required for reading into an entire variable. Its restricted expression argument should evaluate to a 2-element vector {N, M} or {N; M}, which indicates a N×M matrix destination. A scalar N is also allowed and indicates a N×1 column vector destination. When the destination is a submatrix, SIZE is optional, and if it is present then it must match the size of the submatrix.

By default, or with MODE=RECTANGULAR, the command reads an entry for every row and column. With MODE=SYMMETRIC, the command reads only the entries on and below the matrix's main diagonal, and copies the entries above the main diagonal from the corresponding symmetric entries below it. Only square matrices may use MODE=SYMMETRIC.

Ordinarily, each READ command starts from a new line in the text file. Specify the REREAD subcommand to instead start from the last line read by the previous READ command. This has no effect for the first READ command to read from a particular file. It is also ineffective just after a command that uses the EOF matrix function on a particular file, because EOF has to try to read the next line from the file to determine whether the file contains more input.

Example 1: Basic Use

The following matrix program reads the same matrix {1, 2, 4; 2, 3, 5; 4, 5, 6} into matrix variables v, w, and x:

READ v /FILE='input.txt' /FIELD=1 TO 100 /SIZE={3, 3}.
READ w /FIELD=1 TO 100 /SIZE={3; 3} /MODE=SYMMETRIC.
READ x /FIELD=1 TO 100 BY 1/SIZE={3, 3} /MODE=SYMMETRIC.

given that input.txt contains the following:

1, 2, 4
2, 3, 5
4, 5, 6
1
2 3
4 5 6
1
23
456

The READ command will read as many lines of input as needed for a particular row, so it's also acceptable to break any of the lines above into multiple lines. For example, the first line 1, 2, 4 could be written with a line break following either or both commas.

Example 2: Reading into a Submatrix

The following reads a 5×5 matrix from input2.txt, reversing the order of the rows:

COMPUTE m = MAKE(5, 5, 0).
LOOP r = 5 TO 1 BY -1.
  READ m(r, :) /FILE='input2.txt' /FIELD=1 TO 100.
END LOOP.

Example 3: Using REREAD

Suppose each of the 5 lines in a file input3.txt starts with an integer COUNT followed by COUNT numbers, e.g.:

1 5
3 1 2 3
5 6 -1 2 5 1
2 8 9
3 1 3 2

Then, the following reads this file into a matrix m:

COMPUTE m = MAKE(5, 5, 0).
LOOP i = 1 TO 5.
  READ count /FILE='input3.txt' /FIELD=1 TO 1 /SIZE=1.
  READ m(i, 1:count) /FIELD=3 TO 100 /REREAD.
END LOOP.

WRITE Command

WRITE expression
      [/OUTFILE=file]
      /FIELD=first TO last [BY width]
      [/FORMAT=format]
      [/MODE={RECTANGULAR | TRIANGULAR}]
      [/HOLD].

The WRITE command evaluates expression and writes its value to a text file in a specified format. Write the expression to evaluate just after the command name.

The OUTFILE subcommand is required in the first WRITE command that appears within MATRIX. It specifies the text file to be written, either as a file name in quotes or a file handle previously declared on FILE HANDLE. Later WRITE commands (in syntax order) use the previous referenced file if FILE is omitted.

The FIELD and FORMAT subcommands specify how output lines are formed. FIELD is required, but FORMAT is optional. See READ and WRITE Commands, for details.

By default, or with MODE=RECTANGULAR, the command writes an entry for every row and column. With MODE=TRIANGULAR, the command writes only the entries on and below the matrix's main diagonal. Entries above the diagonal are not written. Only square matrices may be written with MODE=TRIANGULAR.

Ordinarily, each WRITE command writes complete lines to the output file. With HOLD, the final line written by WRITE will be held back for the next WRITE command to augment. This can be useful to write more than one matrix on a single output line.

Example 1: Basic Usage

This matrix program:

WRITE {1, 2; 3, 4} /OUTFILE='matrix.txt' /FIELD=1 TO 80.

writes the following to matrix.txt:

 1 2
 3 4

Example 2: Triangular Matrix

This matrix program:

WRITE MAGIC(5) /OUTFILE='matrix.txt' /FIELD=1 TO 80 BY 5 /MODE=TRIANGULAR.

writes the following to matrix.txt:

    17
    23    5
     4    6   13
    10   12   19   21
    11   18   25    2    9

GET Command

GET variable[(index[,index])]
    [/FILE={file | *}]
    [/VARIABLES=variable…]
    [/NAMES=variable]
    [/MISSING={ACCEPT | OMIT | number}]
    [/SYSMIS={OMIT | number}].

The READ command reads numeric data from an SPSS system file, SPSS/PC+ system file, or SPSS portable file into a matrix variable or submatrix:

  • To read data into a variable, specify just its name following GET. The variable need not already exist; if it does, it is replaced. The variable will have as many columns as there are variables specified on the VARIABLES subcommand and as many rows as there are cases in the input file.

  • To read data into a submatrix, specify the name of an existing variable, followed by an indexing expression, just after GET. The submatrix must have as many columns as variables specified on VARIABLES and as many rows as cases in the input file.

Specify the name or handle of the file to be read on FILE. Use *, or simply omit the FILE subcommand, to read from the active file. Reading from the active file is only permitted if it was already defined outside MATRIX.

List the variables to be read as columns in the matrix on the VARIABLES subcommand. The list can use TO for collections of variables or ALL for all variables. If VARIABLES is omitted, all variables are read. Only numeric variables may be read.

If a variable is named on NAMES, then the names of the variables read as data columns are stored in a string vector within the given name, replacing any existing matrix variable with that name. Variable names are truncated to 8 bytes.

The MISSING and SYSMIS subcommands control the treatment of missing values in the input file. By default, any user- or system-missing data in the variables being read from the input causes an error that prevents GET from executing. To accept missing values, specify one of the following settings on MISSING:

  • ACCEPT: Accept user-missing values with no change.

    By default, system-missing values still yield an error. Use the SYSMIS subcommand to change this treatment:

    • OMIT: Skip any case that contains a system-missing value.

    • number: Recode the system-missing value to number.

  • OMIT: Skip any case that contains any user- or system-missing value.

  • number: Recode all user- and system-missing values to number.

The SYSMIS subcommand has an effect only with MISSING=ACCEPT.

SAVE Command

SAVE expression
     [/OUTFILE={file | *}]
     [/VARIABLES=variable…]
     [/NAMES=expression]
     [/STRINGS=variable…].

The SAVE matrix command evaluates expression and writes the resulting matrix to an SPSS system file. In the system file, each matrix row becomes a case and each column becomes a variable.

Specify the name or handle of the SPSS system file on the OUTFILE subcommand, or * to write the output as the new active file. The OUTFILE subcommand is required on the first SAVE command, in syntax order, within MATRIX. For SAVE commands after the first, the default output file is the same as the previous.

When multiple SAVE commands write to one destination within a single MATRIX, the later commands append to the same output file. All the matrices written to the file must have the same number of columns. The VARIABLES, NAMES, and STRINGS subcommands are honored only for the first SAVE command that writes to a given file.

By default, SAVE names the variables in the output file COL1 through COLn. Use VARIABLES or NAMES to give the variables meaningful names. The VARIABLES subcommand accepts a comma-separated list of variable names. Its alternative, NAMES, instead accepts an expression that must evaluate to a row or column string vector of names. The number of names need not exactly match the number of columns in the matrix to be written: extra names are ignored; extra columns use default names.

By default, SAVE assumes that the matrix to be written is all numeric. To write string columns, specify a comma-separated list of the string columns' variable names on STRINGS.

MGET Command

MGET [/FILE=file]
     [/TYPE={COV | CORR | MEAN | STDDEV | N | COUNT}].

The MGET command reads the data from a matrix file into matrix variables.

All of MGET's subcommands are optional. Specify the name or handle of the matrix file to be read on the FILE subcommand; if it is omitted, then the command reads the active file.

By default, MGET reads all of the data from the matrix file. Specify a space-delimited list of matrix types on TYPE to limit the kinds of data to the one specified:

  • COV: Covariance matrix.
  • CORR: Correlation coefficient matrix.
  • MEAN: Vector of means.
  • STDDEV: Vector of standard deviations.
  • N: Vector of case counts.
  • COUNT: Vector of counts.

MGET reads the entire matrix file and automatically names, creates, and populates matrix variables using its contents. It constructs the name of each variable by concatenating the following:

  • A 2-character prefix that identifies the type of the matrix:

    • CV: Covariance matrix.
    • CR: Correlation coefficient matrix.
    • MN: Vector of means.
    • SD: Vector of standard deviations.
    • NC: Vector of case counts.
    • CN: Vector of counts.

  • If the matrix file has factor variables, Fn, where n is a number identifying a group of factors: F1 for the first group, F2 for the second, and so on. This part is omitted for pooled data (where the factors all have the system-missing value).

  • If the matrix file has split file variables, Sn, where n is a number identifying a split group: S1 for the first group, S2 for the second, and so on.

If MGET chooses the name of an existing variable, it issues a warning and does not change the variable.

MSAVE Command

MSAVE expression
      /TYPE={COV | CORR | MEAN | STDDEV | N | COUNT}
      [/FACTOR=expression]
      [/SPLIT=expression]
      [/OUTFILE=file]
      [/VARIABLES=variable…]
      [/SNAMES=variable…]
      [/FNAMES=variable…].

The MSAVE command evaluates the expression specified just after the command name, and writes the resulting matrix to a matrix file.

The TYPE subcommand is required. It specifies the ROWTYPE_ to write along with this matrix.

The FACTOR and SPLIT subcommands are required on the first MSAVE if and only if the matrix file has factor or split variables, respectively. After that, their values are carried along from one MSAVE command to the next in syntax order as defaults. Each one takes an expression that must evaluate to a vector with the same number of entries as the matrix has factor or split variables, respectively. Each MSAVE only writes data for a single combination of factor and split variables, so many MSAVE commands (or one inside a loop) may be needed to write a complete set.

The remaining MSAVE subcommands define the format of the matrix file. All of the MSAVE commands within a given matrix program write to the same matrix file, so these subcommands are only meaningful on the first MSAVE command within a matrix program. (If they are given again on later MSAVE commands, then they must have the same values as on the first.)

The OUTFILE subcommand specifies the name or handle of the matrix file to be written. Output must go to an external file, not a data set or the active file.

The VARIABLES subcommand specifies a comma-separated list of the names of the continuous variables to be written to the matrix file. The TO keyword can be used to define variables named with consecutive integer suffixes. These names become column names and names that appear in VARNAME_ in the matrix file. ROWTYPE_ and VARNAME_ are not allowed on VARIABLES. If VARIABLES is omitted, then PSPP uses the names COL1, COL2, and so on.

The FNAMES subcommand may be used to supply a comma-separated list of factor variable names. The default names are FAC1, FAC2, and so on.

The SNAMES subcommand can supply a comma-separated list of split variable names. The default names are SPL1, SPL2, and so on.

DISPLAY Command

DISPLAY [{DICTIONARY | STATUS}].

The DISPLAY command makes PSPP display a table with the name and dimensions of each matrix variable. The DICTIONARY and STATUS keywords are accepted but have no effect.

RELEASE Command

RELEASE variable….

The RELEASE command accepts a comma-separated list of matrix variable names. It deletes each variable and releases the memory associated with it.

The END MATRIX command releases all matrix variables.

Utility Commands

This chapter describes commands that don't fit in other categories.

Most of these commands are not affected by commands like IF and LOOP: they take effect only once, unconditionally, at the time that they are encountered in the input.

ADD DOCUMENT

ADD DOCUMENT
    'line one' 'line two' ... 'last line' .

ADD DOCUMENT adds one or more lines of descriptive commentary to the active dataset. Documents added in this way are saved to system files. They can be viewed using SYSFILE INFO or DISPLAY DOCUMENTS. They can be removed from the active dataset with DROP DOCUMENTS.

Each line of documentary text must be enclosed in quotation marks, and may not be more than 80 bytes long. See also DOCUMENT.

CACHE

CACHE.

This command is accepted, for compatibility, but it has no effect.

CD

CD 'new directory' .

CD changes the current directory. The new directory becomes that specified by the command.

COMMENT

Comment commands:
    COMMENT comment text ... .
    *comment text ... .

Comments within a line of syntax:
    FREQUENCIES /VARIABLES=v0 v1 v2.  /* All our categorical variables.

COMMENT is ignored. It is used to provide information to the author and other readers of the PSPP syntax file.

COMMENT can extend over any number of lines. It ends at a dot at the end of a line or a blank line. The comment may contain any characters.

PSPP also supports comments within a line of syntax, introduced with /*. These comments end at the first */ or at the end of the line, whichever comes first. A line that contains just this kind of comment is considered blank and ends the current command.

DOCUMENT

DOCUMENT DOCUMENTARY_TEXT.

DOCUMENT adds one or more lines of descriptive commentary to the active dataset. Documents added in this way are saved to system files. They can be viewed using SYSFILE INFO or DISPLAY DOCUMENTS. They can be removed from the active dataset with DROP DOCUMENTS.

Specify the text of the document following the DOCUMENT keyword. It is interpreted literally—any quotes or other punctuation marks are included in the file. You can extend the documentary text over as many lines as necessary, including blank lines to separate paragraphs. Lines are truncated at 80 bytes. Don't forget to terminate the command with a dot at the end of a line. See also ADD DOCUMENT.

DISPLAY DOCUMENTS

DISPLAY DOCUMENTS.

DISPLAY DOCUMENTS displays the documents in the active dataset. Each document is preceded by a line giving the time and date that it was added. See also DOCUMENT.

DISPLAY FILE LABEL

DISPLAY FILE LABEL.

DISPLAY FILE LABEL displays the file label contained in the active dataset, if any. See also FILE LABEL.

This command is a PSPP extension.

DROP DOCUMENTS

DROP DOCUMENTS.

DROP DOCUMENTS removes all documents from the active dataset. New documents can be added with DOCUMENT.

DROP DOCUMENTS changes only the active dataset. It does not modify any system files stored on disk.

ECHO

ECHO 'arbitrary text' .

Use ECHO to write arbitrary text to the output stream. The text should be enclosed in quotation marks following the normal rules for string tokens. string tokens.

ERASE

ERASE FILE "FILE_NAME".

ERASE FILE deletes a file from the local file system. The file's name must be quoted. This command cannot be used if the SAFER setting is active.

EXECUTE

EXECUTE.

EXECUTE causes the active dataset to be read and all pending transformations to be executed.

FILE LABEL

FILE LABEL file label.

FILE LABEL provides a title for the active dataset. This title is saved into system files and portable files that are created during this PSPP run.

The file label should not be quoted. If quotes are included, they are become part of the file label.

FINISH

FINISH.

FINISH terminates the current PSPP session and returns control to the operating system.

HOST

In the syntax below, the square brackets must be included in the command syntax and do not indicate that that their contents are optional.

HOST COMMAND=['COMMAND'...]
     TIMELIMIT=SECS.

HOST executes one or more commands, each provided as a string in the required COMMAND subcommand, in the shell of the underlying operating system. PSPP runs each command in a separate shell process and waits for it to finish before running the next one. If a command fails (with a nonzero exit status, or because it is killed by a signal), then PSPP does not run any remaining commands.

PSPP provides /dev/null as the shell's standard input. If a process needs to read from stdin, redirect from a file or device, or use a pipe.

PSPP displays the shell's standard output and standard error as PSPP output. Redirect to a file or /dev/null or another device if this is not desired.

By default, PSPP waits as long as necessary for the series of commands to complete. Use the optional TIMELIMIT subcommand to limit the execution time to the specified number of seconds.

PSPP built for mingw does not support all the features of HOST.

PSPP rejects this command if the SAFER setting is active.

Example

The following example runs rsync to copy a file from a remote server to the local file data.txt, writing rsync's own output to rsync-log.txt. PSPP displays the command's error output, if any. If rsync needs to prompt the user (e.g. to obtain a password), the command fails. Only if the rsync succeeds, PSPP then runs the sha512sum command.

HOST COMMAND=['rsync remote:data.txt data.txt > rsync-log.txt'
              'sha512sum -c data.txt.sha512sum].

INCLUDE

INCLUDE [FILE=]'FILE_NAME' [ENCODING='ENCODING'].

INCLUDE causes the PSPP command processor to read an additional command file as if it were included bodily in the current command file. If errors are encountered in the included file, then command processing stops and no more commands are processed. Include files may be nested to any depth, up to the limit of available memory.

The INSERT command is a more flexible alternative to INCLUDE. An INCLUDE command acts the same as INSERT with ERROR=STOP CD=NO SYNTAX=BATCH specified.

The optional ENCODING subcommand has the same meaning as with INSERT.

INSERT

INSERT [FILE=]'FILE_NAME'
   [CD={NO,YES}]
   [ERROR={CONTINUE,STOP}]
   [SYNTAX={BATCH,INTERACTIVE}]
   [ENCODING={LOCALE, 'CHARSET_NAME'}].

INSERT is similar to INCLUDE but more flexible. It causes the command processor to read a file as if it were embedded in the current command file.

If CD=YES is specified, then before including the file, the current directory becomes the directory of the included file. The default setting is CD=NO. This directory remains current until it is changed explicitly (with the CD command, or a subsequent INSERT command with the CD=YES option). It does not revert to its original setting even after the included file is finished processing.

If ERROR=STOP is specified, errors encountered in the inserted file causes processing to immediately cease. Otherwise processing continues at the next command. The default setting is ERROR=CONTINUE.

If SYNTAX=INTERACTIVE is specified then the syntax contained in the included file must conform to interactive syntax conventions. The default conventions](../language/basics/syntax-variants.md). The default setting is SYNTAX=BATCH.

ENCODING optionally specifies the character set used by the included file. Its argument, which is not case-sensitive, must be in one of the following forms:

  • LOCALE
    The encoding used by the system locale, or as overridden by SET LOCALE. On GNU/Linux and other Unix-like systems, environment variables, e.g. LANG or LC_ALL, determine the system locale.

  • 'CHARSET_NAME'
    An IANA character set name. Some examples are ASCII (United States), ISO-8859-1 (western Europe), EUC-JP (Japan), and windows-1252 (Windows). Not all systems support all character sets.

  • Auto,ENCODING
    Automatically detects whether a syntax file is encoded in a Unicode encoding such as UTF-8, UTF-16, or UTF-32. If it is not, then PSPP generally assumes that the file is encoded in ENCODING (an IANA character set name). However, if ENCODING is UTF-8, and the syntax file is not valid UTF-8, PSPP instead assumes that the file is encoded in windows-1252.

    For best results, ENCODING should be an ASCII-compatible encoding (the most common locale encodings are all ASCII-compatible), because encodings that are not ASCII compatible cannot be automatically distinguished from UTF-8.

  • Auto
    Auto,Locale
    Automatic detection, as above, with the default encoding taken from the system locale or the setting on SET LOCALE.

When ENCODING is not specified, the default is taken from the --syntax-encoding command option, if it was specified, and otherwise it is Auto.

OUTPUT

In the syntax below, the characters [ and ] are literals. They must appear in the syntax to be interpreted:

OUTPUT MODIFY
     /SELECT TABLES
     /TABLECELLS SELECT = [ CLASS... ]
     FORMAT = FMT_SPEC.

OUTPUT changes the appearance of the tables in which results are printed. In particular, it can be used to set the format and precision to which results are displayed.

After running this command, the default table appearance parameters will have been modified and each new output table generated uses the new parameters.

Following /TABLECELLS SELECT = a list of cell classes must appear, enclosed in square brackets. This list determines the classes of values should be selected for modification. Each class can be:

  • RESIDUAL: Residual values. Default: F40.2.

  • CORRELATION: Correlations. Default: F40.3.

  • PERCENT: Percentages. Default: PCT40.1.

  • SIGNIFICANCE: Significance of tests (p-values). Default: F40.3.

  • COUNT: Counts or sums of weights. For a weighted data set, the default is the weight variable's print format. For an unweighted data set, the default is F40.0.

For most other numeric values that appear in tables, SET FORMAT) may be used to specify the format.

FMT_SPEC must be a valid output format. Not all possible format](../language/datasets/formats/index.md). Not all possible formats are meaningful for all classes.

PERMISSIONS

PERMISSIONS
        FILE='FILE_NAME'
        /PERMISSIONS = {READONLY,WRITEABLE}.

PERMISSIONS changes the permissions of a file. There is one mandatory subcommand which specifies the permissions to which the file should be changed. If you set a file's permission to READONLY, then the file will become unwritable either by you or anyone else on the system. If you set the permission to WRITEABLE, then the file becomes writeable by you; the permissions afforded to others are unchanged. This command cannot be used if the SAFER setting is active.

PRESERVE…RESTORE

PRESERVE.
...
RESTORE.

PRESERVE saves all of the settings that SET can adjust. A later RESTORE command restores those settings.

PRESERVE can be nested up to five levels deep.

SET

SET

(data input)
        /BLANKS={SYSMIS,'.',number}
        /DECIMAL={DOT,COMMA}
        /FORMAT=FMT_SPEC
        /EPOCH={AUTOMATIC,YEAR}
        /RIB={NATIVE,MSBFIRST,LSBFIRST}

(interaction)
        /MXERRS=MAX_ERRS
        /MXWARNS=MAX_WARNINGS
        /WORKSPACE=WORKSPACE_SIZE

(syntax execution)
        /LOCALE='LOCALE'
        /MXLOOPS=MAX_LOOPS
        /SEED={RANDOM,SEED_VALUE}
        /UNDEFINED={WARN,NOWARN}
        /FUZZBITS=FUZZBITS
        /SCALEMIN=COUNT

(data output)
        /CC{A,B,C,D,E}='STRING'
        /DECIMAL={DOT,COMMA}
        /FORMAT=FMT_SPEC
        /LEADZERO={ON,OFF}
        /MDISPLAY={TEXT,TABLES}
        /SMALL=NUMBER
        /WIB={NATIVE,MSBFIRST,LSBFIRST}

(output routing)
        /ERRORS={ON,OFF,TERMINAL,LISTING,BOTH,NONE}
        /MESSAGES={ON,OFF,TERMINAL,LISTING,BOTH,NONE}
        /PRINTBACK={ON,OFF,TERMINAL,LISTING,BOTH,NONE}
        /RESULTS={ON,OFF,TERMINAL,LISTING,BOTH,NONE}

(output driver options)
        /HEADERS={NO,YES,BLANK}
        /LENGTH={NONE,N_LINES}
        /WIDTH={NARROW,WIDTH,N_CHARACTERS}
        /TNUMBERS={VALUES,LABELS,BOTH}
        /TVARS={NAMES,LABELS,BOTH}
        /TLOOK={NONE,FILE}

(journal)
        /JOURNAL={ON,OFF} ['FILE_NAME']

(system files)
        /SCOMPRESSION={ON,OFF}

(security)
        /SAFER=ON
        /LOCALE='STRING'

(macros)
        /MEXPAND={ON,OFF}
        /MPRINT={ON,OFF}
        /MITERATE=NUMBER
        /MNEST=NUMBER

(not yet implemented)
        /BASETEXTDIRECTION={AUTOMATIC,RIGHTTOLEFT,LEFTTORIGHT}
        /BLOCK='C'
        /BOX={'XXX','XXXXXXXXXXX'}
        /CACHE={ON,OFF}
        /CELLSBREAK=NUMBER
        /COMPRESSION={ON,OFF}
        /CMPTRANS={ON,OFF}
        /HEADER={NO,YES,BLANK}

SET allows the user to adjust several parameters relating to PSPP's execution. Since there are many subcommands to this command, its subcommands are examined in groups.

For subcommands that take boolean values, ON and YES are synonymous, as are OFF and NO, when used as subcommand values.

Data Input

SET
        /BLANKS={SYSMIS,'.',number}
        /DECIMAL={DOT,COMMA}
        /FORMAT=FMT_SPEC
        /EPOCH={AUTOMATIC,YEAR}
        /RIB={NATIVE,MSBFIRST,LSBFIRST}

The data input subcommands affect the way that data is read from data files. The data input subcommands are:

  • BLANKS
    This is the value assigned to an item data item that is empty or contains only white space. An argument of SYSMIS or '.' causes the system-missing value to be assigned to null items. This is the default. Any real value may be assigned.

  • DECIMAL
    This value may be set to DOT or COMMA. Setting it to DOT causes the decimal point character to be . and the grouping character to be ,. Setting it to COMMA causes the decimal point character to be , and the grouping character to be .. If the setting is COMMA, then , is not treated as a field separator in the DATA LIST command. The default value is determined from the system locale.

  • FORMAT
    Changes the default numeric input/output format. The default is initially F8.2.

  • EPOCH
    Specifies the range of years used when a 2-digit year is read from a data file or used in a date construction expression. If a 4-digit year is specified for the epoch, then 2-digit years are interpreted starting from that year, known as the epoch. If AUTOMATIC (the default) is specified, then the epoch begins 69 years before the current date.

  • RIB
    PSPP extension to set the byte ordering (endianness) used for reading data in IB or PIB format. In MSBFIRST ordering, the most-significant byte appears at the left end of a IB or PIB field. In LSBFIRST ordering, the least-significant byte appears at the left end. NATIVE, the default, is equivalent to MSBFIRST or LSBFIRST depending on the native format of the machine running PSPP.

Interaction

SET
        /MXERRS=MAX_ERRS
        /MXWARNS=MAX_WARNINGS
        /WORKSPACE=WORKSPACE_SIZE

Interaction subcommands affect the way that PSPP interacts with an online user. The interaction subcommands are

  • MXERRS
    The maximum number of errors before PSPP halts processing of the current command file. The default is 50.

  • MXWARNS
    The maximum number of warnings + errors before PSPP halts processing the current command file. The special value of zero means that all warning situations should be ignored. No warnings are issued, except a single initial warning advising you that warnings will not be given. The default value is 100.

Syntax Execution

SET
        /LOCALE='LOCALE'
        /MXLOOPS=MAX_LOOPS
        /SEED={RANDOM,SEED_VALUE}
        /UNDEFINED={WARN,NOWARN}
        /FUZZBITS=FUZZBITS
        /SCALEMIN=COUNT

Syntax execution subcommands control the way that PSPP commands execute. The syntax execution subcommands are

  • LOCALE
    Overrides the system locale for the purpose of reading and writing syntax and data files. The argument should be a locale name in the general form LANGUAGE_COUNTRY.ENCODING, where LANGUAGE and COUNTRY are 2-character language and country abbreviations, respectively, and ENCODING is an IANA character set name. Example locales are en_US.UTF-8 (UTF-8 encoded English as spoken in the United States) and ja_JP.EUC-JP (EUC-JP encoded Japanese as spoken in Japan).

  • MXLOOPS
    The maximum number of iterations for an uncontrolled LOOP, and for any loop in the matrix language. The default MXLOOPS is 40.

  • SEED
    The initial pseudo-random number seed. Set it to a real number or to RANDOM, to obtain an initial seed from the current time of day.

  • UNDEFINED
    Currently not used.

  • FUZZBITS
    The maximum number of bits of errors in the least-significant places to accept for rounding up a value that is almost halfway between two possibilities for rounding with the RND. The default FUZZBITS is 6.

  • SCALEMIN
    The minimum number of distinct valid values for PSPP to assume that a variable has a scale measurement level.

  • WORKSPACE
    The maximum amount of memory (in kilobytes) that PSPP uses to store data being processed. If memory in excess of the workspace size is required, then PSPP starts to use temporary files to store the data. Setting a higher value means that procedures run faster, but may cause other applications to run slower. On platforms without virtual memory management, setting a very large workspace may cause PSPP to abort.

Data Output

SET
        /CC{A,B,C,D,E}='STRING'
        /DECIMAL={DOT,COMMA}
        /FORMAT=FMT_SPEC
        /LEADZERO={ON,OFF}
        /MDISPLAY={TEXT,TABLES}
        /SMALL=NUMBER
        /WIB={NATIVE,MSBFIRST,LSBFIRST}

Data output subcommands affect the format of output data. These subcommands are

  • CCA
    CCB
    CCC
    CCD
    CCE
    Set up custom currency formats.

  • DECIMAL
    The default DOT setting causes the decimal point character to be .. A setting of COMMA causes the decimal point character to be ,.

  • FORMAT
    Allows the default numeric input/output format to be specified. The default is F8.2.

  • LEADZERO
    Controls whether numbers with magnitude less than one are displayed with a zero before the decimal point. For example, with SET LEADZERO=OFF, which is the default, one-half is shown as 0.5, and with SET LEADZERO=ON, it is shown as .5. This setting affects only the F, COMMA, and DOT formats.

  • MDISPLAY
    Controls how the PRINT command within MATRIX...END MATRIX outputs matrices. With the default TEXT, PRINT outputs matrices as text. Change this setting to TABLES to instead output matrices as pivot tables.

  • SMALL
    This controls how PSPP formats small numbers in pivot tables, in cases where PSPP does not otherwise have a well-defined format for the numbers. When such a number has a magnitude less than the value set here, PSPP formats the number in scientific notation; otherwise, it formats it in standard notation. The default is 0.0001. Set a value of 0 to disable scientific notation.

  • WIB
    PSPP extension to set the byte ordering (endianness) used for writing data in IB or PIB format. In MSBFIRST ordering, the most-significant byte appears at the left end of a IB or PIB field. In LSBFIRST ordering, the least-significant byte appears at the left end. NATIVE, the default, is equivalent to MSBFIRST or LSBFIRST depending on the native format of the machine running PSPP.

Output Routing

SET
        /ERRORS={ON,OFF,TERMINAL,LISTING,BOTH,NONE}
        /MESSAGES={ON,OFF,TERMINAL,LISTING,BOTH,NONE}
        /PRINTBACK={ON,OFF,TERMINAL,LISTING,BOTH,NONE}
        /RESULTS={ON,OFF,TERMINAL,LISTING,BOTH,NONE}

In the PSPP text-based interface, the output routing subcommands affect where output is sent. The following values are allowed for each of these subcommands:

  • OFF
    NONE
    Discard this kind of output.

  • TERMINAL
    Write this output to the terminal, but not to listing files and other output devices.

  • LISTING
    Write this output to listing files and other output devices, but not to the terminal.

  • ON
    BOTH
    Write this type of output to all output devices.

These output routing subcommands are:

  • ERRORS
    Applies to error and warning messages. The default is BOTH.

  • MESSAGES
    Applies to notes. The default is BOTH.

  • PRINTBACK
    Determines whether the syntax used for input is printed back as part of the output. The default is NONE.

  • RESULTS
    Applies to everything not in one of the above categories, such as the results of statistical procedures. The default is BOTH.

These subcommands have no effect on output in the PSPP GUI environment.

Output Driver

SET
        /HEADERS={NO,YES,BLANK}
        /LENGTH={NONE,N_LINES}
        /WIDTH={NARROW,WIDTH,N_CHARACTERS}
        /TNUMBERS={VALUES,LABELS,BOTH}
        /TVARS={NAMES,LABELS,BOTH}
        /TLOOK={NONE,FILE}

Output driver option subcommands affect output drivers' settings. These subcommands are:

  • HEADERS

  • LENGTH

  • WIDTH

  • TNUMBERS
    The TNUMBERS option sets the way in which values are displayed in output tables. The valid settings are VALUES, LABELS and BOTH. If TNUMBERS is set to VALUES, then all values are displayed with their literal value (which for a numeric value is a number and for a string value an alphanumeric string). If TNUMBERS is set to LABELS, then values are displayed using their assigned value labels, if any. If the value has no label, then the literal value is used for display. If TNUMBERS is set to BOTH, then values are displayed with both their label (if any) and their literal value in parentheses.

  • TVARS
    The TVARS option sets the way in which variables are displayed in output tables. The valid settings are NAMES, LABELS and BOTH. If TVARS is set to NAMES, then all variables are displayed using their names. If TVARS is set to LABELS, then variables are displayed using their variable label, if one has been set. If no label has been set, then the name is used. If TVARS is set to BOTH, then variables are displayed with both their label (if any) and their name in parentheses.

  • TLOOK
    The TLOOK option sets the style used for subsequent table output. Specifying NONE makes PSPP use the default built-in style. Otherwise, specifying FILE makes PSPP search for an .stt or .tlo file in the same way as specifying --table-look=FILE the PSPP command line (*note Main Options::).

Journal

SET
        /JOURNAL={ON,OFF} ['FILE_NAME']

Journal subcommands affect logging of commands executed to external files. These subcommands are

  • JOURNAL
    LOG
    These subcommands, which are synonyms, control the journal. The default is ON, which causes commands entered interactively to be written to the journal file. Commands included from syntax files that are included interactively and error messages printed by PSPP are also written to the journal file, prefixed by >. OFF disables use of the journal.

    The journal is named pspp.jnl by default. A different name may be specified.

System Files

SET
        /SCOMPRESSION={ON,OFF}

System file subcommands affect the default format of system files produced by PSPP. These subcommands are

Security

SET
        /SAFER=ON
        /LOCALE='STRING'

Security subcommands affect the operations that commands are allowed to perform. The security subcommands are

  • SAFER
    Setting this option disables the following operations:

    • The ERASE command.
    • The HOST command.
    • The PERMISSIONS command.
    • Pipes (file names beginning or ending with |).

    Be aware that this setting does not guarantee safety (commands can still overwrite files, for instance) but it is an improvement. When set, this setting cannot be reset during the same session, for obvious security reasons.

  • LOCALE
    This item is used to set the default character encoding. The encoding may be specified either as an IANA encoding name or alias, or as a locale name. If given as a locale name, only the character encoding of the locale is relevant.

    System files written by PSPP use this encoding. System files read by PSPP, for which the encoding is unknown, are interpreted using this encoding.

    The full list of valid encodings and locale names/alias are operating system dependent. The following are all examples of acceptable syntax on common GNU/Linux systems.

    SET LOCALE='iso-8859-1'.

SET LOCALE='ru_RU.cp1251'.

SET LOCALE='japanese'.

    Contrary to intuition, this command does not affect any aspect of the system's locale.

Macros

SET
        /MEXPAND={ON,OFF}
        /MPRINT={ON,OFF}
        /MITERATE=NUMBER
        /MNEST=NUMBER

The following subcommands affect the interpretation of macros. For more information, see Macro Settings.

  • MEXPAND
    Controls whether macros are expanded. The default is ON.

  • MPRINT
    Controls whether the expansion of macros is included in output. This is separate from whether command syntax in general is included in output. The default is OFF.

  • MITERATE
    Limits the number of iterations executed in !DO loops within macros. This does not affect other language constructs such as LOOP…END LOOP. This must be set to a positive integer. The default is 1000.

  • MNEST
    Limits the number of levels of nested macro expansions. This must be set to a positive integer. The default is 50.

Not Yet Implemented

SET
        /BASETEXTDIRECTION={AUTOMATIC,RIGHTTOLEFT,LEFTTORIGHT}
        /BLOCK='C'
        /BOX={'XXX','XXXXXXXXXXX'}
        /CACHE={ON,OFF}
        /CELLSBREAK=NUMBER
        /COMPRESSION={ON,OFF}
        /CMPTRANS={ON,OFF}
        /HEADER={NO,YES,BLANK}

The following subcommands are not yet implemented, but PSPP accepts them and ignores the settings:

  • BASETEXTDIRECTION
  • BLOCK
  • BOX
  • CACHE
  • CELLSBREAK
  • COMPRESSION
  • CMPTRANS
  • HEADER

SHOW

SHOW
        [ALL]
        [BLANKS]
        [CC]
        [CCA]
        [CCB]
        [CCC]
        [CCD]
        [CCE]
        [COPYING]
        [DECIMAL]
        [DIRECTORY]
        [ENVIRONMENT]
        [FORMAT]
        [FUZZBITS]
        [LENGTH]
        [MEXPAND]
        [MPRINT]
        [MITERATE]
        [MNEST]
        [MXERRS]
        [MXLOOPS]
        [MXWARNS]
        [N]
        [SCOMPRESSION]
        [SYSTEM]
        [TEMPDIR]
        [UNDEFINED]
        [VERSION]
        [WARRANTY]
        [WEIGHT]
        [WIDTH]

SHOW displays PSPP's settings and status. Parameters that can be changed using SET, can be examined using SHOW using the subcommand with the same name. SHOW supports the following additional subcommands:

  • ALL
    Show all settings.
  • CC
    Show all custom currency settings (CCA through CCE).
  • DIRECTORY
    Shows the current working directory.
  • ENVIRONMENT
    Shows the operating system details.
  • N
    Reports the number of cases in the active dataset. The reported number is not weighted. If no dataset is defined, then Unknown is reported.
  • SYSTEM
    Shows information about how PSPP was built. This information is useful in bug reports.
  • TEMPDIR
    Shows the path of the directory where temporary files are stored.
  • VERSION
    Shows the version of this installation of PSPP.
  • WARRANTY
    Show details of the lack of warranty for PSPP.
  • COPYING or LICENSE
    Display the terms of PSPP's copyright licence.

Specifying SHOW without any subcommands is equivalent to SHOW ALL.

SUBTITLE

SUBTITLE 'SUBTITLE_STRING'.
  or
SUBTITLE SUBTITLE_STRING.

SUBTITLE provides a subtitle to a particular PSPP run. This subtitle appears at the top of each output page below the title, if headers are enabled on the output device.

Specify a subtitle as a string in quotes. The alternate syntax that did not require quotes is now obsolete. If it is used then the subtitle is converted to all uppercase.

TITLE

TITLE 'TITLE_STRING'.
  or
TITLE TITLE_STRING.

TITLE provides a title to a particular PSPP run. This title appears at the top of each output page, if headers are enabled on the output device.

Specify a title as a string in quotes. The alternate syntax that did not require quotes is now obsolete. If it is used then the title is converted to all uppercase.

System File Format

An SPSS system file holds a set of cases and dictionary information that describes how they may be interpreted. The system file format dates back 40+ years and has evolved greatly over that time to support new features, but in a way to facilitate interchange between even the oldest and newest versions of software. This chapter describes the system file format.

Introduction

System files use four data types: 8-bit characters, 32-bit integers, 64-bit integers, and 64-bit floating points, called here char’, int32’, int64’, and flt64’, respectively. Data is not necessarily aligned on a word or double-word boundary: the long variable name record and very long string record have arbitrary byte length and can therefore cause all data coming after them in the file to be misaligned.

Integer data in system files may be big-endian or little-endian. A reader may detect the endianness of a system file by examining layout_code in the file header record.

Floating-point data in system files may nominally be in IEEE 754, IBM, or VAX formats. A reader may detect the floating-point format in use by examining bias in the file header record. Only files with IEEE 754 floating point data have actually been encountered.

PSPP detects big-endian and little-endian integer formats in system files and translates as necessary. PSPP also detects the floating-point format in use, as well as the endianness of IEEE 754 floating-point numbers, and translates as needed. However, only IEEE 754 numbers with the same endianness as integer data in the same file have actually been observed in system files, and it is likely that other formats are obsolete or were never used.

System files use a few floating point values for special purposes:

  • SYSMIS

    The system-missing value is represented by the largest possible negative number in the floating point format (-DBL_MAX or f64::MIN).

  • HIGHEST

    HIGHEST is used as the high end of a missing value range with an unbounded maximum. It is represented by the largest possible positive number (DBL_MAX or f64::MAX).

  • LOWEST

    LOWEST is used as the low end of a missing value range with an unbounded minimum. It was originally represented by the second-largest negative number (in IEEE 754 format, 0xffeffffffffffffe). System files written by SPSS 21 and later instead use the largest negative number (-DBL_MAX or f64::MIN), the same value as SYSMIS. This does not lead to ambiguity because LOWEST appears in system files only in missing value ranges, which never contain SYSMIS.

System files may use most character encodings based on an 8-bit unit. UTF-16 and UTF-32, based on wider units, appear to be unacceptable. rec_type in the file header record is sufficient to distinguish between ASCII and EBCDIC based encodings. The best way to determine the specific encoding in use is to consult the character encoding record, if present, and failing that character_code in the machine integer info record. The same encoding should be used for the dictionary and the data in the file, although it is possible to artificially synthesize files that use different encodings.

System File Record Structure

System files are divided into records with the following format:

     int32               type;
     char                data[];

This header does not identify the length of the data or any information about what it contains, so the system file reader must understand the format of data based on type. However, records with type 7, called “extension records”, have a stricter format:

     int32               type;
     int32               subtype;
     int32               size;
     int32               count;
     char                data[size * count];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. This value identifies a particular kind of extension record.

  • int32 size;

    The size of each piece of data that follows the header, in bytes. Known extension records use 1, 4, or 8, for char, int32, and flt64 format data, respectively.

  • int32 count;

    The number of pieces of data that follow the header.

  • char data[size * count];

    Data, whose format and interpretation depend on the subtype.

An extension record contains exactly size * count bytes of data, which allows a reader that does not understand an extension record to skip it. Extension records provide only nonessential information, so this allows for files written by newer software to preserve backward compatibility with older or less capable readers.

Records in a system file must appear in the following order:

  • File header record.

  • Variable records.

  • All pairs of value labels records and value label variables records, if present.

  • Document record, if present.

  • Extension (type 7) records, in ascending numerical order of their subtypes.

    System files written by SPSS include at most one of each kind of extension record. This is generally true of system files written by other software as well, with known exceptions noted below in the individual sections about each type of record.

  • Dictionary termination record.

  • Data record.

We advise authors of programs that read system files to tolerate format variations. Various kinds of misformatting and corruption have been observed in system files written by SPSS and other software alike. In particular, because extension records provide nonessential information, it is generally better to ignore an extension record entirely than to refuse to read a system file.

The following sections describe the known kinds of records.

File Header Record

A system file begins with the file header, with the following format:

     char                rec_type[4];
     char                prod_name[60];
     int32               layout_code;
     int32               nominal_case_size;
     int32               compression;
     int32               weight_index;
     int32               ncases;
     flt64               bias;
     char                creation_date[9];
     char                creation_time[8];
     char                file_label[64];
     char                padding[3];

  • char rec_type[4];

    Record type code, either $FL2 for system files with uncompressed data or data compressed with simple bytecode compression, or $FL3 for system files with ZLIB compressed data.

    This is truly a character field that uses the character encoding as other strings. Thus, in a file with an ASCII-based character encoding this field contains 24 46 4c 32 or 24 46 4c 33, and in a file with an EBCDIC-based encoding this field contains 5b c6 d3 f2. (No EBCDIC-based ZLIB-compressed files have been observed.)

  • char prod_name[60];

    Product identification string. This always begins with the characters @(#) SPSS DATA FILE. PSPP uses the remaining characters to give its version and the operating system name; for example, GNU pspp 0.1.4 - sparc-sun-solaris2.5.2. The string is truncated if it would be longer than 60 characters; otherwise it is padded on the right with spaces.

    The product name field allow readers to behave differently based on quirks in the way that particular software writes system files. See Value Labels Records, for the detail of the quirk that the PSPP system file reader tolerates in files written by ReadStat, which has https://github.com/WizardMac/ReadStat in prod_name.

  • int32 layout_code;

    Normally set to 2, although a few system files have been spotted in the wild with a value of 3 here. PSPP use this value to determine the file's integer endianness.

  • int32 nominal_case_size;

    Number of data elements per case. This is the number of variables, except that long string variables add extra data elements (one for every 8 characters after the first 8). However, string variables do not contribute to this value beyond the first 255 bytes. Further, some software always writes -1 or 0 in this field. In general, it is unsafe for systems reading system files to rely upon this value.

  • int32 compression;

    Set to 0 if the data in the file is not compressed, 1 if the data is compressed with simple bytecode compression, 2 if the data is ZLIB compressed. This field has value 2 if and only if rec_type is $FL3.

  • int32 weight_index;

    If one of the variables in the data set is used as a weighting variable, set to the dictionary index of that variable. Otherwise, set to 0.

  • int32 ncases;

    Set to the number of cases in the file if it is known, or -1 otherwise.

    In the general case it is not possible to determine the number of cases that will be output to a system file at the time that the header is written. The way that this is dealt with is by writing the entire system file, including the header, then seeking back to the beginning of the file and writing just the ncases field. For files in which this is not valid, the seek operation fails. In this case, ncases remains -1.

  • flt64 bias;

    Compression bias, usually 100. Only integers between 1 - bias and 251 - bias can be compressed.

    By assuming that its value is 100, PSPP uses bias to determine the file's floating-point format and endianness. If the compression bias is not 100, PSPP cannot auto-detect the floating-point format and assumes that it is IEEE 754 format with the same endianness as the system file's integers, which is correct for all known system files.

  • char creation_date[9];

    Date of creation of the system file, in dd mmm yy format, with the month as standard English abbreviations, using an initial capital letter and following with lowercase.

    Some files in the corpus have the date in dd-mmm-yy format.

  • char creation_time[8];

    Time of creation of the system file, in hh:mm:ss format and using 24-hour time.

  • char file_label[64];

    File label declared by the user, if any. Padded on the right with spaces.

    A product that identifies itself as VOXCO INTERVIEWER 4.3 uses CR-only line ends in this field, rather than the more usual LF-only or CR LF line ends.

  • char padding[3];

    Ignored padding bytes to make the structure a multiple of 32 bits in length. Set to zeros.

Variable Record

There must be one variable record for each numeric variable and each string variable with width 8 bytes or less. String variables wider than 8 bytes have one variable record for each 8 bytes, rounding up. The first variable record for a long string specifies the variable's correct dictionary information. Subsequent variable records for a long string are filled with dummy information: a type of -1, no variable label or missing values, print and write formats that are ignored, and an empty string as name. A few system files have been encountered that include a variable label on dummy variable records, so readers should take care to parse dummy variable records in the same way as other variable records.

The "dictionary index" of a variable is a 1-based offset in the set of variable records, including dummy variable records for long string variables. The first variable record has a dictionary index of 1, the second has a dictionary index of 2, and so on.

The system file format does not directly support string variables wider than 255 bytes. Such very long string variables are represented by a number of narrower string variables. See very long string record for details.

A system file should contain at least one variable and thus at least one variable record, but system files have been observed in the wild without any variables (thus, no data either).

     int32               rec_type;
     int32               type;
     int32               has_var_label;
     int32               n_missing_values;
     int32               print;
     int32               write;
     char                name[8];

     /* Present only if `has_var_label` is 1. */
     int32               label_len;
     char                label[];

     /* Present only if `n_missing_values` is nonzero. */
     flt64               missing_values[];

  • int32 rec_type;

    Record type code. Always set to 2.

  • int32 type;

    Variable type code. Set to 0 for a numeric variable. For a short string variable or the first part of a long string variable, this is set to the width of the string. For the second and subsequent parts of a long string variable, set to -1, and the remaining fields in the structure are ignored.

  • int32 has_var_label;

    If this variable has a variable label, set to 1; otherwise, set to 0.

  • int32 n_missing_values;

    If the variable has no missing values, set to 0. If the variable has one, two, or three discrete missing values, set to 1, 2, or 3, respectively. If the variable has a range for missing variables, set to -2; if the variable has a range for missing variables plus a single discrete value, set to -3.

    A long string variable always has the value 0 here. A separate record indicates missing values for long string variables.

  • int32 print;

    Print format for this variable. See below.

  • int32 write;

    Write format for this variable. See below.

  • char name[8];

    Variable name. The variable name must begin with a capital letter or the at-sign (@). Subsequent characters may also be digits, octothorpes (#), dollar signs ($), underscores (_), or full stops (.). The variable name is padded on the right with spaces.

    The name fields should be unique within a system file. System files written by SPSS that contain very long string variables with similar names sometimes contain duplicate names that are later eliminated by resolving the very long string names. PSPP handles duplicates by assigning them new, unique names.

  • int32 label_len;

    This field is present only if has_var_label is set to 1. It is set to the length, in characters, of the variable label. The documented maximum length varies from 120 to 255 based on SPSS version, but some files have been seen with longer labels. PSPP accepts labels of any length.

  • char label[];

    This field is present only if has_var_label is set to 1. It has length label_len, rounded up to the nearest multiple of 32 bits. The first label_len characters are the variable's variable label.

  • flt64 missing_values[];

    This field is present only if n_missing_values is nonzero. It has the same number of 8-byte elements as the absolute value of n_missing_values. Each element is interpreted as a number for numeric variables (with HIGHEST and LOWEST indicated as described in the introduction). For string variables of width less than 8 bytes, elements are right-padded with spaces.

    For discrete missing values, each element represents one missing value. When a range is present, the first element denotes the minimum value in the range, and the second element denotes the maximum value in the range. When a range plus a value are present, the third element denotes the additional discrete missing value.

Format Types

The print and write members of sysfile_variable are output formats coded into int32 types. The least-significant byte of the int32 represents the number of decimal places, and the next two bytes in order of increasing significance represent field width and format type, respectively. The most-significant byte is not used and should be set to zero.

Format types are defined as follows:

ValueMeaning
0Not used.
1A
2AHEX
3COMMA
4DOLLAR
5F
6IB
7PIBHEX
8P
9PIB
10PK
11RB
12RBHEX
13Not used.
14Not used.
15Z
16N
17E
18Not used.
19Not used.
20DATE
21TIME
22DATETIME
23ADATE
24JDATE
25DTIME
26WKDAY
27MONTH
28MOYR
29QYR
30WKYR
31PCT
32DOT
33CCA
34CCB
35CCC
36CCD
37CCE
38EDATE
39SDATE
40MTIME
41YMDHMS

A few system files have been observed in the wild with invalid write fields, in particular with value 0. Readers should probably treat invalid print or write fields as some default format.

Obsolete Treatment of Long String Missing Values

SPSS and most versions of PSPP write missing values for string variables wider than 8 bytes with a Long String Missing Values Record. Very old versions of PSPP instead wrote these missing values on the variables record, writing only the first 8 bytes of each missing value, with the remainder implicitly all spaces. Any new software should use the Long String Missing Values Record, but it might possibly be worthwhile also to accept the old format used by PSPP.

Value Labels Records

The value label records documented in this section are used for numeric and short string variables only. Long string variables may have value labels, but their value labels are recorded using a different record type.

ReadStat writes value labels that label a single value more than once. In more detail, it emits value labels whose values are longer than string variables' widths, that are identical in the actual width of the variable, e.g. labels for values ABC123 and ABC456 for a string variable with width 3. For files written by this software, PSPP ignores such labels.

Value Label Record for Labels

The value label record has the following format:

     int32               rec_type;
     int32               label_count;

     /* Repeated `n_label` times. */
     char                value[8];
     char                label_len;
     char                label[];

  • int32 rec_type;

    Record type. Always set to 3.

  • int32 label_count;

    Number of value labels present in this record.

The remaining fields are repeated count times. Each repetition specifies one value label.

  • char value[8];

    A numeric value or a short string value padded as necessary to 8 bytes in length. Its type and width cannot be determined until the following value label variables record (see below) is read.

  • char label_len;

    The label's length, in bytes. The documented maximum length varies from 60 to 120 based on SPSS version. PSPP supports value labels up to 255 bytes long.

  • char label[];

    label_len bytes of the actual label, followed by up to 7 bytes of padding to bring label and label_len together to a multiple of 8 bytes in length.

Value Label Record for Variables

The value label record is always immediately followed by a value label variables record with the following format:

  int32               rec_type;
  int32               var_count;
  int32               vars[];

  • int32 rec_type;

    Record type. Always set to 4.

  • int32 var_count;

    Number of variables that the associated value labels from the value label record are to be applied.

  • int32 vars[];

    A list of 1-based dictionary indexes of variables to which to apply the value labels. There are var_count elements.

    String variables wider than 8 bytes may not be specified in this list.

Document Record

The document record, if present, has the following format:

     int32               rec_type;
     int32               n_lines;
     char                lines[][80];

  • int32 rec_type;

    Record type. Always set to 6.

  • int32 n_lines;

    Number of lines of documents present. This should be greater than zero, but ReadStats writes system files with zero n_lines.

  • char lines[][80];

    Document lines. The number of elements is defined by n_lines. Lines shorter than 80 characters are padded on the right with spaces.

Machine Integer Info Record

The integer info record, if present, has the following format:

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Data. */
     int32               version_major;
     int32               version_minor;
     int32               version_revision;
     int32               machine_code;
     int32               floating_point_rep;
     int32               compression_code;
     int32               endianness;
     int32               character_code;

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 3.

  • int32 size;

    Size of each piece of data in the data part, in bytes. Always set to 4.

  • int32 count;

    Number of pieces of data in the data part. Always set to 8.

  • int32 version_major;

    PSPP major version number. In version X.Y.Z, this is X.

  • int32 version_minor;

    PSPP minor version number. In version X.Y.Z, this is Y.

  • int32 version_revision;

    PSPP version revision number. In version X.Y.Z, this is Z.

  • int32 machine_code;

    Machine code. PSPP always set this field to value to -1, but other values may appear.

  • int32 floating_point_rep;

    Floating point representation code. For IEEE 754 systems (the most common) this is 1. IBM 370 is supposed to set this to 2, and DEC VAX E to 3, but neither of these has been observed.

  • int32 compression_code;

    Compression code. Always set to 1, regardless of whether or how the file is compressed.

  • int32 endianness;

    Machine endianness. 1 indicates big-endian, 2 indicates little-endian.

  • int32 character_code;

    Character code. The following values have been actually observed in system files:

    • 1

      EBCDIC. Only one example has been observed.

    • 2

      7-bit ASCII. Old versions of SPSS for Unix and Windows always wrote value 2 in this field, regardless of the encoding in use, so it is not reliable and should be ignored.

    • 3

      8-bit "ASCII".

    • 819

      ISO 8859-1 (IBM AIX code page number).

    • 874
      9066

      The windows-874 code page for Thai.

    • 932

      The windows-932 code page for Japanese (aka Shift_JIS).

    • 936

      The windows-936 code page for simplified Chinese (aka GBK).

    • 949

      Probably ks_c_5601-1987, Unified Hangul Code.

    • 950

      The big5 code page for traditional Chinese.

    • 1250

      The windows-1250 code page for Central European and Eastern European languages.

    • 1251

      The windows-1251 code page for Cyrillic languages.

    • 1252

      The windows-1252 code page for Western European languages.

    • 1253

      The windows-1253 code page for modern Greek.

    • 1254

      The windows-1254 code page for Turkish.

    • 1255

      The windows-1255 code page for Hebrew.

    • 1256

      The windows-1256 code page for Arabic script.

    • 1257

      The windows-1257 code page for Estonian, Latvian, and Lithuanian.

    • 1258

      The windows-1258 code page for Vietnamese.

    • 20127

      US-ASCII.

    • 28591

      ISO 8859-1 (Latin-1).

    • 25592

      ISO 8859-2 (Central European).

    • 28605

      ISO 8895-9 (Latin-9).

    • 51949

      The euc-kr code page for Korean.

    • 65001

      UTF-8.

    The following additional values are known to be defined:

    • 3

      8-bit "ASCII".

    • 4

      DEC Kanji.

    The most common values observed, from most to least common, are 1252, 65001, 2, and 28591.

    Other Windows code page numbers are known to be generally valid.

    Newer versions also write the character encoding as a string.

Machine Floating-Point Info Record

The floating-point info record, if present, has the following format:

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Data. */
     flt64               sysmis;
     flt64               highest;
     flt64               lowest;

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 4.

  • int32 size;

    Size of each piece of data in the data part, in bytes. Always set to 8.

  • int32 count;

    Number of pieces of data in the data part. Always set to 3.

  • flt64 sysmis;
    flt64 highest;
    flt64 lowest;

    The system missing value, the value used for HIGHEST in missing values, and the value used for LOWEST in missing values, respectively. See the introduction for more information.

    The SPSSWriter library in PHP, which identifies itself as FOM SPSS 1.0.0 in the file header record prod_name field, writes unexpected values to these fields, but it uses the same values consistently throughout the rest of the file.

Multiple Response Sets Records

The system file format has two different types of records that represent multiple response sets. The first type of record describes multiple response sets that can be understood by SPSS before version 14. The second type of record, with a closely related format, is used for multiple dichotomy sets that use the CATEGORYLABELS=COUNTEDVALUES feature added in version 14.

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Exactly `count` bytes of data. */
     char                mrsets[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Set to 7 for records that describe multiple response sets understood by SPSS before version 14, or to 19 for records that describe dichotomy sets that use the CATEGORYLABELS=COUNTEDVALUES feature added in version 14.

  • int32 size;

    The size of each element in the mrsets member. Always set to 1.

  • int32 count;

    The total number of bytes in mrsets.

  • char mrsets[];

    Zero or more line feeds (byte 0x0a), followed by a series of multiple response sets, each of which consists of the following:

    • The set's name (an identifier that begins with $), in mixed upper and lower case.

    • An equals sign (=).

    • C for a multiple category set, D for a multiple dichotomy set with CATEGORYLABELS=VARLABELS, or E for a multiple dichotomy set with CATEGORYLABELS=COUNTEDVALUES.

    • For a multiple dichotomy set with CATEGORYLABELS=COUNTEDVALUES, a space, followed by a number expressed as decimal digits, followed by a space. If LABELSOURCE=VARLABEL was specified on MRSETS, then the number is 11; otherwise it is 1.1

    • For either kind of multiple dichotomy set, the counted value, as a positive integer count specified as decimal digits, followed by a space, followed by as many string bytes as specified in the count. If the set contains numeric variables, the string consists of the counted integer value expressed as decimal digits. If the set contains string variables, the string contains the counted string value. Either way, the string may be padded on the right with spaces (older versions of SPSS seem to always pad to a width of 8 bytes; newer versions don't).

    • A space.

    • The multiple response set's label, using the same format as for the counted value for multiple dichotomy sets. A string of length 0 means that the set does not have a label. A string of length 0 is also written if LABELSOURCE=VARLABEL was specified.

    • The short names of the variables in the set, converted to lowercase, each preceded by a single space.

      Even though a multiple response set must have at least two variables, some system files contain multiple response sets with no variables or one variable. The source and meaning of these multiple response sets is unknown. (Perhaps they arise from creating a multiple response set then deleting all the variables that it contains?)

    • One line feed (byte 0x0a). Sometimes multiple, even hundreds, of line feeds are present.

Example: Given appropriate variable definitions, consider the following MRSETS command:

MRSETS /MCGROUP NAME=$a LABEL='my mcgroup' VARIABLES=a b c
       /MDGROUP NAME=$b VARIABLES=g e f d VALUE=55
       /MDGROUP NAME=$c LABEL='mdgroup #2' VARIABLES=h i j VALUE='Yes'
       /MDGROUP NAME=$d LABEL='third mdgroup' CATEGORYLABELS=COUNTEDVALUES
        VARIABLES=k l m VALUE=34
       /MDGROUP NAME=$e CATEGORYLABELS=COUNTEDVALUES LABELSOURCE=VARLABEL
        VARIABLES=n o p VALUE='choice'.

The above would generate the following multiple response set record of subtype 7:

$a=C 10 my mcgroup a b c
$b=D2 55 0  g e f d
$c=D3 Yes 10 mdgroup #2 h i j

It would also generate the following multiple response set record with subtype 19:

$d=E 1 2 34 13 third mdgroup k l m
$e=E 11 6 choice 0  n o p

Extra Product Info Record

This optional record appears to contain a text string that describes the program that wrote the file and the source of the data. (This is redundant with the file label and product info found in the file header record.)

 /* Header. */
 int32               rec_type;
 int32               subtype;
 int32               size;
 int32               count;

 /* Exactly `count` bytes of data. */
 char                info[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 10.

  • int32 size;

    The size of each element in the info member. Always set to 1.

  • int32 count;

    The total number of bytes in info.

  • char info[];

    A text string. A product that identifies itself as VOXCO INTERVIEWER 4.3 uses CR-only line ends in this field, rather than the more usual LF-only or CR LF line ends.

Variable Display Parameter Record

The variable display parameter record, if present, has the following format:

 /* Header. */
 int32               rec_type;
 int32               subtype;
 int32               size;
 int32               count;

 /* Repeated `count` times. */
 int32               measure;
 int32               width;           /* Not always present. */
 int32               alignment;

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 11.

  • int32 size;

    The size of int32. Always set to 4.

  • int32 count;

    The number of sets of variable display parameters (ordinarily the number of variables in the dictionary), times 2 or 3.

The remaining members are repeated count times, in the same order as the variable records. No element corresponds to variable records that continue long string variables. The meanings of these members are as follows:

  • int32 measure;

    The measurement level of the variable:

    ValueLevel
    0Unknown
    1Nominal
    2Ordinal
    3Scale

    An "unknown" measure of 0 means that the variable was created in some way that doesn't make the measurement level clear, e.g. with a COMPUTE transformation. PSPP sets the measurement level the first time it reads the data, so this should rarely appear.

  • int32 width;

    The width of the display column for the variable in characters.

    This field is present if count is 3 times the number of variables in the dictionary. It is omitted if count is 2 times the number of variables.

  • int32 alignment;

    The alignment of the variable for display purposes:

    ValueAlignment
    0Left aligned
    1Right aligned
    2Centre aligned

Variable Sets Record

The SPSS GUI offers users the ability to arrange variables in sets. Users may enable and disable sets individually, and the data editor and analysis dialog boxes only show enabled sets. Syntax does not use variable sets.

The variable sets record, if present, has the following format:

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Exactly `count` bytes of text. */
     char                text[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 5.

  • int32 size;

    Always set to 1.

  • int32 count;

    The total number of bytes in text.

  • char text[];

    The variable sets, in a text-based format.

    Each variable set occupies one line of text, each of which ends with a line feed (byte 0x0a), optionally preceded by a carriage return (byte 0x0d).

    Each line begins with the name of the variable set, followed by an equals sign (=) and a space (byte 0x20), followed by the long variable names of the members of the set, separated by spaces. A variable set may be empty, in which case the equals sign and the space following it are still present.

Long Variable Names Record

If present, the long variable names record has the following format:

 /* Header. */
 int32               rec_type;
 int32               subtype;
 int32               size;
 int32               count;

 /* Exactly `count` bytes of data. */
 char                var_name_pairs[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 13.

  • int32 size;

    The size of each element in the var_name_pairs member. Always set to 1.

  • int32 count;

    The total number of bytes in var_name_pairs.

  • char var_name_pairs[];

    A list of key-value tuples, where each key is the name of a variable, and the value is its long variable name. The key field is at most 8 bytes long and must match the name of a variable which appears in the variable record. The value field is at most 64 bytes long. The key and value fields are separated by a = byte. Each tuple is separated by a byte whose value is 09. There is no trailing separator following the last tuple. The total length is count bytes.

Very Long String Record

Old versions of SPSS limited string variables to a width of 255 bytes. For backward compatibility with these older versions, the system file format represents a string longer than 255 bytes, called a “very long string”, as a collection of strings no longer than 255 bytes each. The strings concatenated to make a very long string are called its “segments”; for consistency, variables other than very long strings are considered to have a single segment.

A very long string with a width of w has n = (w + 251) / 252 segments, that is, one segment for every 252 bytes of width, rounding up. It would be logical, then, for each of the segments except the last to have a width of 252 and the last segment to have the remainder, but this is not the case. In fact, each segment except the last has a width of 255 bytes. The last has width w - (n - 1) * 252; some versions of SPSS make it slightly wider, but not wide enough to make the last segment require another 8 bytes of data.

Data is packed tightly into segments of a very long string, 255 bytes per segment. Because 255 bytes of segment data are allocated for every 252 bytes of the very long string's width (approximately), some unused space is left over at the end of the allocated segments. Data in unused space is ignored.

Example: Consider a very long string of width 20,000. Such a very long string has 20,000 / 252 = 80 (rounding up) segments. The first 79 segments have width 255; the last segment has width 20,000 - 79 * 252 = 92 or slightly wider (up to 96 bytes, the next multiple of 8). The very long string's data is actually stored in the 19,890 bytes in the first 78 segments, plus the first 110 bytes of the 79th segment (19,890 + 110 = 20,000). The remaining 145 bytes of the 79th segment and all 92 bytes of the 80th segment are unused.

The very long string record explains how to stitch together segments to obtain very long string data. For each of the very long string variables in the dictionary, it specifies the name of its first segment's variable and the very long string variable's actual width. The remaining segments immediately follow the named variable in the system file's dictionary.

The very long string record, which is present only if the system file contains very long string variables, has the following format:

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Exactly `count` bytes of data. */
     char                string_lengths[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 14.

  • int32 size;

    The size of each element in the string_lengths member. Always set to 1.

  • int32 count;

    The total number of bytes in string_lengths.

  • char string_lengths[];

    A list of key-value tuples, where key is the name of a variable, and value is its length. The key field is at most 8 bytes long and must match the name of a variable which appears in the variable record. The value field is exactly 5 bytes long. It is a zero-padded, ASCII-encoded string that is the length of the variable. The key and value fields are separated by a = byte. Tuples are delimited by a two-byte sequence {00, 09}. After the last tuple, there may be a single byte 00, or {00, 09}. The total length is count bytes.

Character Encoding Record

This record, if present, indicates the character encoding for string data, long variable names, variable labels, value labels and other strings in the file.

 /* Header. */
 int32               rec_type;
 int32               subtype;
 int32               size;
 int32               count;

 /* Exactly `count` bytes of data. */
 char                encoding[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 20.

  • int32 size;

    The size of each element in the encoding member. Always set to 1.

  • int32 count;

    The total number of bytes in encoding.

  • char encoding[];

    The name of the character encoding. Normally this will be an official IANA character set name or alias. Character set names are not case-sensitive, and SPSS is not consistent, e.g. both windows-1251 and WINDOWS-1252 have both been observed, as have Big5 and BIG5.

This record is not present in files generated by older software. See also character_code in the machine integer info record.

The following character encoding names have been observed. The names are shown in lowercase, even though they were not always in lowercase in the file. Alternative names for the same encoding are, when known, listed together. For each encoding, the character_code values that they were observed paired with are also listed. First, the following are strictly single-byte, ASCII-compatible encodings:

  • (encoding record missing)

    0, 2, 3, 874, 1250, 1251, 1252, 1253, 1254, 1255, 1256, 20127, 28591, 28592, 28605

  • ansi_x3.4-1968
    ascii

    1252

  • cp28605

    2

  • cp874

    9066

  • iso-8859-1

    819

  • windows-874

    874

  • windows-1250

    2, 1250, 1252

  • windows-1251

    2, 1251

  • cp1252
    windows-1252

    2, 1250, 1252, 1253

  • cp1253
    windows-1253

    1253

  • windows-1254

    2, 1254

  • windows-1255

    2, 1255

  • windows-1256

    2, 1252, 1256

  • windows-1257

    2, 1257

  • windows-1258

    1258

The others are multibyte encodings, in which some code points occupy a single byte and others multiple bytes. The following multibyte encodings are "ASCII compatible," that is, they use ASCII values only to indicate ASCII:

  • (encoding record missing)

    65001, 949

  • euc-kr

    2, 51949

  • utf-8

    0, 2, 1250, 1251, 1252, 1256, 65001

The following multibyte encodings are not ASCII compatible, that is, while they encode ASCII characters as their native values, they also use ASCII values as second or later bytes in multibyte sequences:

  • (encoding record missing)

    932, 936, 950

  • big5
    cp950

    2, 950

  • gbk

    936

  • cp932
    windows-31j

    932

As the tables above show, when the character encoding record and the machine integer info record are both present, they can contradict each other. Observations show that, in this case, the character encoding record should be honored.

If, for testing purposes, a file is crafted with different character_code and encoding, it seems that character_code controls the encoding for all strings in the system file before the dictionary termination record, including strings in data (e.g. string missing values), and encoding controls the encoding for strings following the dictionary termination record.

Long String Value Labels Record

This record, if present, specifies value labels for long string variables.

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Repeated up to exactly `count` bytes. */
     int32               var_name_len;
     char                var_name[];
     int32               var_width;
     int32               n_labels;
     long_string_label   labels[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 21.

  • int32 size;

    Always set to 1.

  • int32 count;

    The number of bytes following the header until the next header.

  • int32 var_name_len;
    char var_name[];

    The number of bytes in the name of the variable that has long string value labels, plus the variable name itself, which consists of exactly var_name_len bytes. The variable name is not padded to any particular boundary, nor is it null-terminated.

  • int32 var_width;

    The width of the variable, in bytes, which will be between 9 and 32767.

  • int32 n_labels;
    long_string_label labels[];

    The long string labels themselves. The labels array contains exactly n_labels elements, each of which has the following substructure:

         int32               value_len;
     char                value[];
     int32               label_len;
     char                label[];

    • int32 value_len;
      char value[];

      The string value being labeled. value_len is the number of bytes in value; it is equal to var_width. The value array is not padded or null-terminated.

    • int32 label_len;
      char label[];

      The label for the string value. label_len, which must be between 0 and 120, is the number of bytes in label. The label array is not padded or null-terminated.

Long String Missing Values Record

This record, if present, specifies missing values for long string variables.

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Repeated up to exactly `count` bytes. */
     int32               var_name_len;
     char                var_name[];
     char                n_missing_values;
     int32               value_len;
     char                values[value_len * n_missing_values];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 22.

  • int32 size;

    Always set to 1.

  • int32 count;

    The number of bytes following the header until the next header.

  • int32 var_name_len;
    char var_name[];

    The number of bytes in the name of the long string variable that has missing values, plus the variable name itself, which consists of exactly var_name_len bytes. The variable name is not padded to any particular boundary, nor is it null-terminated.

  • char n_missing_values;

    The number of missing values, either 1, 2, or 3. (This is, unusually, a single byte instead of a 32-bit number.)

  • int32 value_len;

    The length of each missing value string, in bytes. This value should be 8, because long string variables are at least 8 bytes wide (by definition), only the first 8 bytes of a long string variable's missing values are allowed to be non-spaces, and any spaces within the first 8 bytes are included in the missing value here.

  • char values[value_len * n_missing_values]

    The missing values themselves, without any padding or null terminators.

An earlier version of this document stated that value_len was repeated before each of the missing values, so that there was an extra int32 value of 8 before each missing value after the first. Old versions of PSPP wrote data files in this format. Readers can tolerate this mistake, if they wish, by noticing and skipping the extra int32 values, which wouldn't ordinarily occur in strings.

Data File and Variable Attributes Records

The data file and variable attributes records represent custom attributes for the system file or for individual variables in the system file, as defined on the DATAFILE ATTRIBUTE and VARIABLE ATTRIBUTE commands, respectively.

     /* Header. */
     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;

     /* Exactly `count` bytes of data. */
     char                attributes[];

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 17 for a data file attribute record or to 18 for a variable attributes record.

  • int32 size;

    The size of each element in the attributes member. Always set to value 1.

  • int32 count;

    The total number of bytes in attributes.

  • char attributes[];

    The attributes, in a text-based format.

    In record subtype 17, this field contains a single attribute set. An attribute set is a sequence of one or more attributes concatenated together. Each attribute consists of a name, which has the same syntax as a variable name, followed by, inside parentheses, a sequence of one or more values. Each value consists of a string enclosed in single quotes (') followed by a line feed (byte 0x0a). A value may contain single quote characters, which are not themselves escaped or quoted or required to be present in pairs. There is no apparent way to embed a line feed in a value. There is no distinction between an attribute with a single value and an attribute array with one element.

    In record subtype 18, this field contains a sequence of one or more variable attribute sets. If more than one variable attribute set is present, each one after the first is delimited from the previous by /. Each variable attribute set consists of a long variable name, followed by :, followed by an attribute set with the same syntax as on record subtype 17.

    System files written by Stata 14.1/-savespss- 1.77 by S.Radyakin may include multiple records with subtype 18, one per variable that has variable attributes.

    The total length is count bytes.

Example: A system file produced with the following VARIABLE ATTRIBUTE commands in effect:

VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=fred[1]('23') fred[2]('34').
VARIABLE ATTRIBUTE VARIABLES=dummy ATTRIBUTE=bert('123').

will contain a variable attribute record with the following contents:

0000  07 00 00 00 12 00 00 00  01 00 00 00 22 00 00 00  |............"...|
0010  64 75 6d 6d 79 3a 66 72  65 64 28 27 32 33 27 0a  |dummy:fred('23'.|
0020  27 33 34 27 0a 29 62 65  72 74 28 27 31 32 33 27  |'34'.)bert('123'|
0030  0a 29                                             |.)              |

Variable Roles

A variable's role is represented as an attribute named $@Role. This attribute has a single element whose values and their meanings are:

ValueRole
0Input
1Target
2Both
3None
4Partition
5Split

The default and most common role is 0 (input).

Extended Number of Cases Record

ncases in the file header record expresses the number of cases in the system file as an int32. This record allows the number of cases in the system file to be expressed as a 64-bit number.

     int32               rec_type;
     int32               subtype;
     int32               size;
     int32               count;
     int64               unknown;
     int64               ncases64;

  • int32 rec_type;

    Record type. Always set to 7.

  • int32 subtype;

    Record subtype. Always set to 16.

  • int32 size;

    Size of each element. Always set to 8.

  • int32 count;

    Number of pieces of data in the data part. Alway set to 2.

  • int64 unknown;

    Meaning unknown. Always set to 1.

  • int64 ncases64;

    Number of cases in the file as a 64-bit integer. Presumably this could be -1 to indicate that the number of cases is unknown, for the same reason as ncases in the file header record, but this has not been observed in the wild.

Other Informational Records

This chapter documents many specific types of extension records are documented here, but others are known to exist. PSPP ignores unknown extension records when reading system files.

The following extension record subtypes have also been observed, with the following believed meanings:

  • 6

    Date info, probably related to USE (according to Aapi Hämäläinen).

  • 12

    A UUID in the format described in RFC 4122. Only two examples observed, both written by SPSS 13, and in each case the UUID contained both upper and lower case.

  • 24

    XML that describes how data in the file should be displayed on-screen.

Dictionary Termination Record

The dictionary termination record separates all other records from the data records.

     int32               rec_type;
     int32               filler;

  • int32 rec_type;

    Record type. Always set to 999.

  • int32 filler;

    Ignored padding. Should be set to 0.

Data Record

The data record must follow all other records in the system file. Every system file must have a data record that specifies data for at least one case. The format of the data record varies depending on the value of compression in the file header record:

  • 0: no compression

    Data is arranged as a series of 8-byte elements. Each element corresponds to the variable declared in the respective variable record. Numeric values are given in flt64 format; string values are literal characters string, padded on the right when necessary to fill out 8-byte units.

  • 1: bytecode compression

    The first 8 bytes of the data record is divided into a series of 1-byte command codes. These codes have meanings as described below:

    • 0

      Ignored. If the program writing the system file accumulates compressed data in blocks of fixed length, 0 bytes can be used to pad out extra bytes remaining at the end of a fixed-size block.

    • 1 through 251

      A number with value code - bias, where code is the value of the compression code and bias comes from the file header.

      Example: Code 105 with bias 100.0 (the normal value) indicates a numeric variable of value 5.

      A code of 0 (after subtracting the bias) in a string field encodes null bytes. This is unusual, since a string field normally encodes text data, but it exists in real system files.

    • 252

      End of file. This code may or may not appear at the end of the data stream. PSPP always outputs this code but its use is not required.

    • 253

      A numeric or string value that is not compressible. The value is stored in the 8 bytes following the current block of command bytes. If this value appears twice in a block of command bytes, then it indicates the second group of 8 bytes following the command bytes, and so on.

    • 254

      An 8-byte string value that is all spaces.

    • 255

      The system-missing value.

    The end of the 8-byte group of bytecodes is followed by any 8-byte blocks of non-compressible values indicated by code 253. After that follows another 8-byte group of bytecodes, then those bytecodes' non-compressible values. The pattern repeats to the end of the file or a code with value 252.

  • 2: ZLIB compression

    The data record consists of the following, in order:

    • ZLIB data header, 24 bytes long.

    • One or more variable-length blocks of ZLIB compressed data.

    • ZLIB data trailer, with a 24-byte fixed header plus an additional 24 bytes for each preceding ZLIB compressed data block.

    The ZLIB data header has the following format:

         int64               zheader_ofs;
     int64               ztrailer_ofs;
     int64               ztrailer_len;

    • int64 zheader_ofs;

      The offset, in bytes, of the beginning of this structure within the system file. A reader does not need to use this, so it can ignore it (PSPP issues a warning if it does not match its own offset).

    • int64 ztrailer_ofs;

      The offset, in bytes, of the first byte of the ZLIB data trailer.

    • int64 ztrailer_len;

      The number of bytes in the ZLIB data trailer. This and the previous field sum to the size of the system file in bytes.

    The data header is followed by (ztrailer_len - 24) / 24 ZLIB compressed data blocks. Each ZLIB compressed data block begins with a ZLIB header as specified in RFC 1950, e.g. hex bytes 78 01 (the only header yet observed in practice). Each block decompresses to a fixed number of bytes (in practice only 0x3ff000-byte blocks have been observed), except that the last block of data may be shorter. The last ZLIB compressed data block gends just before offset ztrailer_ofs.

    The result of ZLIB decompression is bytecode compressed data as described above for compression format 1.

    The ZLIB data trailer begins with the following 24-byte fixed header:

         int64               bias;
     int64               zero;
     int32               block_size;
     int32               n_blocks;

    • int64 int_bias;

      The compression bias as a negative integer, e.g. if bias in the file header record is 100.0, then int_bias is −100 (this is the only value yet observed in practice).

    • int64 zero;

      Always observed to be zero.

    • int32 block_size;

      The number of bytes in each ZLIB compressed data block, except possibly the last, following decompression. Only 0x3ff000 has been observed so far.

    • int32 n_blocks;

      The number of ZLIB compressed data blocks, always exactly (ztrailer_len - 24) / 24.

    The fixed header is followed by n_blocks 24-byte ZLIB data block descriptors, each of which describes the compressed data block corresponding to its offset. Each block descriptor has the following format:

         int64               uncompressed_ofs;
     int64               compressed_ofs;
     int32               uncompressed_size;
     int32               compressed_size;

    • int64 uncompressed_ofs;

      The offset, in bytes, that this block of data would have in a similar system file that uses compression format 1. This is zheader_ofs in the first block descriptor, and in each succeeding block descriptor it is the sum of the previous desciptor's uncompressed_ofs and uncompressed_size.

    • int64 compressed_ofs;

      The offset, in bytes, of the actual beginning of this compressed data block. This is zheader_ofs + 24 in the first block descriptor, and in each succeeding block descriptor it is the sum of the previous descriptor's compressed_ofs and compressed_size. The final block descriptor's compressed_ofs and compressed_size sum to ztrailer_ofs.

    • int32 uncompressed_size;

      The number of bytes in this data block, after decompression. This is block_size in every data block except the last, which may be smaller.

    • int32 compressed_size;

      The number of bytes in this data block, as stored compressed in this system file.

  1. This part of the format may not be fully understood, because only a single example of each possibility has been examined. ↩

SPSS Viewer File Format

SPSS Viewer or .spv files, here called SPV files, are written by SPSS 16 and later to represent the contents of its output editor. This chapter documents the format, based on examination of a corpus of about 8,000 files from a variety of sources. This description is detailed enough to both read and write SPV files.

SPSS 15 and earlier versions instead use .spo files, which have a completely different output format based on the Microsoft Compound Document Format. This format is not documented here.

An SPV file is a Zip archive that can be read with zipinfo and unzip and similar programs. The final member in the Zip archive is the "manifest", a file named META-INF/MANIFEST.MF. This structure makes SPV files resemble Java "JAR" files (and ODF files), but whereas a JAR manifest contains a sequence of colon-delimited key/value pairs, an SPV manifest contains the string allowPivoting=true, without a new-line. PSPP uses this string to identify an SPV file; it is invariant across the corpus.

SPV files always begin with the 7-byte sequence 50 4b 03 04 14 00 08, but this is not a useful magic number because most Zip archives start the same way.

Checking only for the presence of META-INF/MANIFEST.MF is also not a useful magic number because this file name also appears in every Java JAR archive.

SPSS writes META-INF/MANIFEST.MF to every SPV file, but it does not read it or even require it to exist, so using different contents, e.g. allowPivoting=false, has no effect.

The rest of the members in an SPV file's Zip archive fall into two categories: "structure" and "detail" members. Structure member names take the form with outputViewerNUMBER.xml or outputViewerNUMBER_heading.xml, where NUMBER is an 10-digit decimal number. Each of these members represents some kind of output item (a table, a heading, a block of text, etc.) or a group of them. The member whose output goes at the beginning of the document is numbered 0, the next member in the output is numbered 1, and so on.

Structure members contain XML. This XML is sometimes self-contained, but it often references detail members in the Zip archive, which are named as follows:

  • PREFIX_table.xml and PREFIX_tableData.bin
    PREFIX_lightTableData.bin
    The structure of a table plus its data. Older SPV files pair a PREFIX_table.xml legacy detail XML member that describes the table's structure with a PREFIX_tableData.bin legacy detail binary member that gives its data. Newer SPV files (the majority of those in the corpus) instead include a single PREFIX_lightTableData.bin light detail binary member that incorporates both into a single binary format.

  • PREFIX_warning.xml and PREFIX_warningData.bin
    PREFIX_lightWarningData.bin
    Same format used for tables, with a different name.

  • PREFIX_notes.xml and PREFIX_notesData.bin
    PREFIX_lightNotesData.bin
    Same format used for tables, with a different name.

  • PREFIX_chartData.bin and PREFIX_chart.xml
    The structure of a chart plus its data. Charts do not have a "light" format.

  • PREFIX_Imagegeneric.png
    PREFIX_PastedObjectgeneric.png
    PREFIX_imageData.bin
    A PNG image referenced by an object element (in the first two cases) or an image element (in the final case). See The object and image Elements, for details.

  • PREFIX_pmml.scf
    PREFIX_stats.scf
    PREFIX_model.xml
    Not yet investigated. The corpus contains few examples.

The PREFIX in the names of the detail members is typically an 11-digit decimal number that increases for each item, tending to skip values. Older SPV files use different naming conventions for detail members. Structure member refer to detail members by name, and so their exact names do not matter to readers as long as they are unique.

SPSS tolerates corrupted Zip archives that Zip reader libraries tend to reject. These can be fixed up with zip -FF.

Structure Member Format

A structure member lays out the high-level structure for a group of output items such as heading, tables, and charts. Structure members do not include the details of tables and charts but instead refer to them by their member names.

Structure members' XML files claim conformance with a collection of XML Schemas. These schemas are distributed, under a nonfree license, with SPSS binaries. Fortunately, the schemas are not necessary to understand the structure members. The schemas can even be deceptive because they document elements and attributes that are not in the corpus and do not document elements and attributes that are commonly found in the corpus.

Structure members use a different XML namespace for each schema, but these namespaces are not entirely consistent. In some SPV files, for example, the viewer-tree schema is associated with namespace http://xml.spss.com/spss/viewer-tree and in others with http://xml.spss.com/spss/viewer/viewer-tree (note the additional viewer/). Under either name, the schema URIs are not resolvable to obtain the schemas themselves.

One may ignore all of the above in interpreting a structure member. The actual XML has a simple and straightforward form that does not require a reader to take schemas or namespaces into account. A structure member's root is heading element, which contains heading or container elements (or a mix), forming a tree. In turn, container holds a label and one more child, usually text or table.

Grammar

The following sections document the elements found in structure members in a context-free grammar-like fashion. Consider the following example, which specifies the attributes and content for the container element:

container
   :visibility=(visible | hidden)
   :page-break-before=(always)?
   :text-align=(left | center)?
   :width=dimension
=> label (table | container_text | graph | model | object | image | tree)

Each attribute specification begins with : followed by the attribute's name. If the attribute's value has an easily specified form, then = and its description follows the name. Finally, if the attribute is optional, the specification ends with ?. The following value specifications are defined:

  • (A | B | ...)
    One of the listed literal strings. If only one string is listed, it is the only acceptable value. If OTHER is listed, then any string not explicitly listed is also accepted.

  • bool
    Either true or false.

  • dimension
    A floating-point number followed by a unit, e.g. 10pt. Units in the corpus include in (inch), pt (points, 72/inch), px ("device-independent pixels", 96/inch), and cm. If the unit is omitted then points should be assumed. The number and unit may be separated by white space.

    The corpus also includes localized names for units. A reader must understand these to properly interpret the dimension:

    • inch: 인치, pol., cala, cali
    • point: пт
    • centimeter: см

  • real
    A floating-point number.

  • int
    An integer.

  • color
    A color in one of the forms #RRGGBB or RRGGBB, or the string transparent, or one of the standard Web color names.

  • ref
    ref ELEMENT
    ref(ELEM1 | ELEM2 | ...)
    The name from the id attribute in some element. If one or more elements are named, the name must refer to one of those elements, otherwise any element is acceptable.

All elements have an optional id attribute. If present, its value must be unique. In practice many elements are assigned id attributes that are never referenced.

The content specification for an element supports the following syntax:

  • ELEMENT
    An element.

  • A B
    A followed by B.

  • A | B | C
    One of A or B or C.

  • A?
    Zero or one instances of A.

  • A*
    Zero or more instances of A.

  • B+
    One or more instances of A.

  • (SUBEXPRESSION)
    Grouping for a subexpression.

  • EMPTY
    No content.

  • TEXT
    Text and CDATA.

Element and attribute names are sometimes suffixed by another name in square brackets to distinguish different uses of the same name. For example, structure XML has two text elements, one inside container, the other inside pageParagraph. The former is defined as text[container_text] and referenced as container_text, the latter defined as text[pageParagraph_text] and referenced as pageParagraph_text.

This language is used in the PSPP source code for parsing structure and detail XML members. Refer to src/output/spv/structure-xml.grammar and src/output/spv/detail-xml.grammar for the full grammars.

The following example shows the contents of a typical structure member for a DESCRIPTIVES procedure. A real structure member is not indented. This example also omits most attributes, all XML namespace information, and the CSS from the embedded HTML:

<?xml version="1.0" encoding="utf-8"?>
<heading>
  <label>Output</label>
  <heading commandName="Descriptives">
    <label>Descriptives</label>
    <container>
      <label>Title</label>
      <text commandName="Descriptives" type="title">
        <html lang="en">
<![CDATA[<head><style type="text/css">...</style></head><BR>Descriptives]]>
        </html>
      </text>
    </container>
    <container visibility="hidden">
      <label>Notes</label>
      <table commandName="Descriptives" subType="Notes" type="note">
        <tableStructure>
          <dataPath>00000000001_lightNotesData.bin</dataPath>
        </tableStructure>
      </table>
    </container>
    <container>
      <label>Descriptive Statistics</label>
      <table commandName="Descriptives" subType="Descriptive Statistics"
             type="table">
        <tableStructure>
          <dataPath>00000000002_lightTableData.bin</dataPath>
        </tableStructure>
      </table>
    </container>
  </heading>
</heading>

The heading Element

heading[root_heading]
   :creator-version?
   :creator?
   :creation-date-time?
   :lockReader=bool?
   :schemaLocation?
=> label pageSetup? (container | heading)*

heading
   :creator-version?
   :commandName?
   :visibility[heading_visibility]=(collapsed)?
   :locale?
   :olang?
=> label (container | heading)*

A heading represents a tree of content that appears in an output viewer window. It contains a label text string that is shown in the outline view ordinarily followed by content containers or further nested (sub)-sections of output. Unlike heading elements in HTML and other common document formats, which precede the content that they head, heading contains the elements that appear below the heading.

The root of a structure member is a special heading. The direct children of the root heading elements in all structure members in an SPV file are siblings. That is, the root heading in all of the structure members conceptually represent the same node. The root heading's label is ignored (see the label element). The root heading in the first structure member in the Zip file may contain a pageSetup element.

The schema implies that any heading may contain a sequence of any number of heading and container elements. This does not work for the root heading in practice, which must actually contain exactly one container or heading child element. Furthermore, if the root heading's child is a heading, then the structure member's name must end in _heading.xml; if it is a container child, then it must not.

The following attributes have been observed on both document root and nested heading elements.

  • creator-version
    The version of the software that created this SPV file. A string of the form xxyyzzww represents software version xx.yy.zz.ww, e.g. 21000001 is version 21.0.0.1. Trailing pairs of zeros are sometimes omitted, so that 21, 210000, and 21000000 are all version 21.0.0.0 (and the corpus contains all three of those forms).

The following attributes have been observed on document root heading elements only:

  • creator
    The directory in the file system of the software that created this SPV file.

  • creation-date-time
    The date and time at which the SPV file was written, in a locale-specific format, e.g. Friday, May 16, 2014 6:47:37 PM PDT or lunedì 17 marzo 2014 3.15.48 CET or even Friday, December 5, 2014 5:00:19 o'clock PM EST.

  • lockReader
    Whether a reader should be allowed to edit the output. The possible values are true and false. The value false is by far the most common.

  • schemaLocation
    This is actually an XML Namespace attribute. A reader may ignore it.

The following attributes have been observed only on nested heading elements:

  • commandName
    A locale-invariant identifier for the command that produced the output, e.g. Frequencies, T-Test, Non Par Corr.

  • visibility
    If this attribute is absent, the heading's content is expanded in the outline view. If it is set to collapsed, it is collapsed. (This attribute is never present in a root heading because the root node is always expanded when a file is loaded, even though the UI can be used to collapse it interactively.)

  • locale
    The locale used for output, in Windows format, which is similar to the format used in Unix with the underscore replaced by a hyphen, e.g. en-US, en-GB, el-GR, sr-Cryl-RS.

  • olang
    The output language, e.g. en, it, es, de, pt-BR.

The label Element

label => TEXT

Every heading and container holds a label as its first child. The label text is what appears in the outline pane of the GUI's viewer window. PSPP also puts it into the outline of PDF output. The label text doesn't appear in the output itself.

The text in label describes what it labels, often by naming the statistical procedure that was executed, e.g. "Frequencies" or "T-Test". Labels are often very generic, especially within a container, e.g. "Title" or "Warnings" or "Notes". Label text is localized according to the output language, e.g. in Italian a frequency table procedure is labeled "Frequenze".

The user can edit labels to be anything they want. The corpus contains a few examples of empty labels, ones that contain no text, probably as a result of user editing.

The root heading in an SPV file has a label, like every heading. It normally contains "Output" but its content is disregarded anyway. The user cannot edit it.

The container Element

container
   :visibility=(visible | hidden)
   :page-break-before=(always)?
   :text-align=(left | center)?
   :width=dimension
=> label (table | container_text | graph | model | object | image | tree)

A container serves to contain and label a table, text, or other kind of item.

This element has the following attributes.

  • visibility
    Whether the container's content is displayed. "Notes" tables are often hidden; other data is usually visible.

  • text-align
    Alignment of text within the container. Observed with nested table and text elements.

  • width
    The width of the container, e.g. 1097px.

All of the elements that nest inside container (except the label) have the following optional attribute.

  • commandName
    As on the heading element. The corpus contains one example of where commandName is present but set to the empty string.

The text Element (Inside container)

text[container_text]
  :type[text_type]=(title | log | text | page-title)
  :commandName?
  :creator-version?
=> html

This text element is nested inside a container. There is a different text element that is nested inside a pageParagraph.

This element has the following attributes.

  • commandName
    See the container element. For output not specific to a command, this is simply log.

  • type
    The semantics of the text.

  • creator-version
    As on the heading element.

The html Element

html :lang=(en) => TEXT

The element contains an HTML document as text (or, in practice, as CDATA). In some cases, the document starts with <html> and ends with </html>; in others the html element is implied. Generally the HTML includes a head element with a CSS stylesheet. The HTML body often begins with <BR>.

The HTML document uses only the following elements:

  • html
    Sometimes, the document is enclosed with <html>...</html>.

  • br
    The HTML body often begins with <BR> and may contain it as well.

  • b
    i
    u
    Styling.

  • font
    The attributes face, color, and size are observed. The value of color takes one of the forms #RRGGBB or rgb (R, G, B). The value of size is a number between 1 and 7, inclusive.

The CSS in the corpus is simple. To understand it, a parser only needs to be able to skip white space, <!--, and -->, and parse style only for p elements. Only the following properties matter:

  • color
    In the form RRGGBB, e.g. 000000, with no leading #.

  • font-weight
    Either bold or normal.

  • font-style
    Either italic or normal.

  • text-decoration
    Either underline or normal.

  • font-family
    A font name, commonly Monospaced or SansSerif.

  • font-size
    Values claim to be in points, e.g. 14pt, but the values are actually in "device-independent pixels" (px), at 96/inch.

This element has the following attributes.

  • lang
    This always contains en in the corpus.

The table Element

table
   :VDPId?
   :ViZmlSource?
   :activePageId=int?
   :commandName
   :creator-version?
   :displayFiltering=bool?
   :maxNumCells=int?
   :orphanTolerance=int?
   :rowBreakNumber=int?
   :subType
   :tableId
   :tableLookId?
   :type[table_type]=(table | note | warning)
=> tableProperties? tableStructure

tableStructure => path? dataPath csvPath?

This element has the following attributes.

  • commandName
    See the container element.

  • type
    One of table, note, or warning.

  • subType
    The locale-invariant command ID for the particular kind of output that this table represents in the procedure. This can be the same as commandName e.g. Frequencies, or different, e.g. Case Processing Summary. Generic subtypes Notes and Warnings are often used.

  • tableId
    A number that uniquely identifies the table within the SPV file, typically a large negative number such as -4147135649387905023.

  • creator-version
    As on the heading element. In the corpus, this is only present for version 21 and up and always includes all 8 digits.

This element contains the following:

  • tableProperties: See Legacy Properties, for details.

  • tableStructure, which in turn contains:

    • Both path and dataPath for legacy members.

    • dataPath but not path for light detail binary members.

    • The usage of csvPath is rare and not yet understood.

    See SPSS Viewer File Format for more information on how structure members refer to tables.

The graph Element

graph
   :VDPId?
   :ViZmlSource?
   :commandName?
   :creator-version?
   :dataMapId?
   :dataMapURI?
   :editor?
   :refMapId?
   :refMapURI?
   :csvFileIds?
   :csvFileNames?
=> dataPath? path csvPath?

This element represents a graph. The dataPath and path elements name the Zip members that give the details of the graph. Normally, both elements are present; there is only one counterexample in the corpus.

csvPath only appears in one SPV file in the corpus, for two graphs. In these two cases, dataPath, path, and csvPath all appear. These csvPath name Zip members with names of the form NUMBER_csv.bin, where NUMBER is a many-digit number and the same as the csvFileIds. The named Zip members are CSV text files (despite the .bin extension). The CSV files are encoded in UTF-8 and begin with a U+FEFF byte-order marker.

The model Element

model
   :PMMLContainerId?
   :PMMLId
   :StatXMLContainerId
   :VDPId
   :auxiliaryViewName
   :commandName
   :creator-version
   :mainViewName
=> ViZml? dataPath? path | pmmlContainerPath statsContainerPath

pmmlContainerPath => TEXT

statsContainerPath => TEXT

ViZml :viewName? => TEXT

This element represents a model. The dataPath and path elements name the Zip members that give the details of the model. Normally, both elements are present; there is only one counterexample in the corpus.

The details are unexplored. The ViZml element contains base-64 encoded text, that decodes to a binary format with some embedded text strings, and path names an Zip member that contains XML. Alternatively, pmmlContainerPath and statsContainerPath name Zip members with .scf extension.

The object and image Elements

object
   :commandName?
   :type[object_type]=(unknown)?
   :uri
=> EMPTY

image
   :commandName?
   :VDPId
=> dataPath

These two elements represent an image in PNG format. They are equivalent and the corpus contains examples of both. The only difference is the syntax: for object, the uri attribute names the Zip member that contains a PNG file; for image, the text of the inner dataPath element names the Zip member.

PSPP writes object in output but there is no strong reason to choose this form.

The corpus only contains PNG image files.

The tree Element

tree
   :commandName
   :creator-version
   :name
   :type
=> dataPath path

This element represents a tree. The dataPath and path elements name the Zip members that give the details of the tree. The details are unexplored.

Path Elements

dataPath => TEXT

path => TEXT

csvPath => TEXT

These element contain the name of the Zip members that hold details for a container. For tables:

  • When a "light" format is used, only dataPath is present, and it names a .bin member of the Zip file that has light in its name, e.g. 0000000001437_lightTableData.bin. See Light Detail Member Format for light format details.

  • When the legacy format is used, both are present. In this case, dataPath names a Zip member with a legacy binary format that contains relevant data (see Legacy Detail Member Binary Format), and path names a Zip member that uses an XML format (see Legacy Detail Member XML Member Format).

Graphs normally follow the legacy approach described above. The corpus contains one example of a graph with path but not dataPath. The reason is unexplored.

Models use path but not dataPath. See graph element, for more information.

These elements have no attributes.

The pageSetup Element

pageSetup
   :initial-page-number=int?
   :chart-size=(as-is | full-height | half-height | quarter-height | OTHER)?
   :margin-left=dimension?
   :margin-right=dimension?
   :margin-top=dimension?
   :margin-bottom=dimension?
   :paper-height=dimension?
   :paper-width=dimension?
   :reference-orientation?
   :space-after=dimension?
=> pageHeader pageFooter

pageHeader => pageParagraph?

pageFooter => pageParagraph?

pageParagraph => pageParagraph_text

The pageSetup element has the following attributes.

  • initial-page-number
    The page number to put on the first page of printed output. Usually 1.

  • chart-size
    One of the listed, self-explanatory chart sizes, quarter-height, or a localization (!) of one of these (e.g. dimensione attuale, Wie vorgegeben).

  • margin-left
    margin-right
    margin-top
    margin-bottom
    Margin sizes, e.g. 0.25in.

  • paper-height
    paper-width
    Paper sizes.

  • reference-orientation
    Indicates the orientation of the output page. Either 0deg (portrait) or 90deg (landscape),

  • space-after
    The amount of space between printed objects, typically 12pt.

The text Element (Inside pageParagraph)

text[pageParagraph_text] :type=(title | text) => TEXT

This text element is nested inside a pageParagraph. There is a different text element that is nested inside a container.

The element is either empty, or contains CDATA that holds almost-XHTML text: in the corpus, either an html or p element. It is almost-XHTML because the html element designates the default namespace as http://xml.spss.com/spss/viewer/viewer-tree instead of an XHTML namespace, and because the CDATA can contain substitution variables. The following variables are supported:

  • &[Date]
    &[Time]
    The current date or time in the preferred format for the locale.

  • &[Head1]
    &[Head2]
    &[Head3]
    &[Head4]
    First-, second-, third-, or fourth-level heading.

  • &[PageTitle]
    The page title.

  • &[Filename]
    Name of the output file.

  • &[Page]
    The page number.

Typical contents (indented for clarity):

<html xmlns="http://xml.spss.com/spss/viewer/viewer-tree">
    <head></head>
    <body>
        <p style="text-align:right; margin-top: 0">Page &[Page]</p>
    </body>
</html>

This element has the following attributes.

  • type
    Always text.

Light Detail Member Format

This section describes the format of "light" detail .bin members.

Binary Format Conventions

These members have a binary format which we describe here in terms of a context-free grammar using the following conventions:

  • NonTerminal ⇒ ...
    Nonterminals have CamelCaps names, and ⇒ indicates a production. The right-hand side of a production is often broken across multiple lines. Break points are chosen for aesthetics only and have no semantic significance.

  • 00, 01, ..., ff.
    A bytes with a fixed value, written as a pair of hexadecimal digits.

  • i0, i1, ..., i9, i10, i11, ...
    ib0, ib1, ..., ib9, ib10, ib11, ...
    A 32-bit integer in little-endian or big-endian byte order, respectively, with a fixed value, written in decimal. Prefixed by i for little-endian or ib for big-endian.

  • byte
    A byte.

  • bool
    A byte with value 0 or 1.

  • int16
    be16
    A 16-bit unsigned integer in little-endian or big-endian byte order, respectively.

  • int32
    be32
    A 32-bit unsigned integer in little-endian or big-endian byte order, respectively.

  • int64
    be64
    A 64-bit unsigned integer in little-endian or big-endian byte order, respectively.

  • double
    A 64-bit IEEE floating-point number.

  • float
    A 32-bit IEEE floating-point number.

  • string
    bestring
    A 32-bit unsigned integer, in little-endian or big-endian byte order, respectively, followed by the specified number of bytes of character data. (The encoding is indicated by the Formats nonterminal.)

  • X?
    X is optional, e.g. 00? is an optional zero byte.

  • X*N
    X is repeated N times, e.g. byte*10 for ten arbitrary bytes.

  • X[NAME]
    Gives X the specified NAME. Names are used in textual explanations. They are also used, also bracketed, to indicate counts, e.g. int32[n] byte*[n] for a 32-bit integer followed by the specified number of arbitrary bytes.

  • A | B
    Either A or B.

  • (X)
    Parentheses are used for grouping to make precedence clear, especially in the presence of |, e.g. in 00 (01 | 02 | 03) 00.

  • count(X)
    becount(X)
    A 32-bit unsigned integer, in little-endian or big-endian byte order, respectively, that indicates the number of bytes in X, followed by X itself.

  • v1(X)
    In a version 1 .bin member, X; in version 3, nothing. (The .bin header indicates the version.)

  • v3(X)
    In a version 3 .bin member, X; in version 1, nothing.

PSPP uses this grammar to parse light detail members. See src/output/spv/light-binary.grammar in the PSPP source tree for the full grammar.

Little-endian byte order is far more common in this format, but a few pieces of the format use big-endian byte order.

Light detail members express linear units in two ways: points (pt), at 72/inch, and "device-independent pixels" (px), at 96/inch. To convert from pt to px, multiply by 1.33 and round up. To convert from px to pt, divide by 1.33 and round down.

Top-Level Structure

A "light" detail member .bin consists of a number of sections concatenated together, terminated by an optional byte 01:

Table =>
    Header Titles Footnotes
    Areas Borders PrintSettings TableSettings Formats
    Dimensions Axes Cells
    01?

An SPV light member begins with a 39-byte header:

Header =>
    01 00
    (i1 | i3)[version]
    bool[x0]
    bool[x1]
    bool[rotate-inner-column-labels]
    bool[rotate-outer-row-labels]
    bool[x2]
    int32[x3]
    int32[min-col-heading-width] int32[max-col-heading-width]
    int32[min-row-heading-width] int32[max-row-heading-width]
    int64[table-id]

version is a version number that affects the interpretation of some of the other data in the member. We will refer to "version 1" and "version 3" later on and use v1(...) and v3(...) for version-specific formatting (as described previously).

If rotate-inner-column-labels is 1, then column labels closest to the data are rotated 90° counterclockwise; otherwise, they are shown in the normal way.

If rotate-outer-row-labels is 1, then row labels farthest from the data are rotated 90° counterclockwise; otherwise, they are shown in the normal way.

min-col-heading-width, max-col-heading-width, min-row-heading-width, and max-row-heading-width are measurements in 1/96 inch units (called "device independent pixel" units in Windows) whose values influence column widths. For the purpose of interpreting these values, a table is divided into the three regions shown below:

┌──────────────────┬─────────────────────────────────────────────────┐
│                  │                  column headings                │
│                  ├─────────────────────────────────────────────────┤
│      corner      │                                                 │
│       and        │                                                 │
│   row headings   │                      data                       │
│                  │                                                 │
│                  │                                                 │
└──────────────────┴─────────────────────────────────────────────────┘

min-col-heading-width and max-col-heading-width apply to the columns in the column headings region. min-col-heading-width is the minimum width that any of these columns will be given automatically. In addition, max-col-heading-width is the maximum width that a column will be assigned to accommodate a long label in the column headings cells. These columns will still be made wider to accommodate wide data values in the data region.

min-row-heading-width is the minimum width that a column in the corner and row headings region will be given automatically. max-col-heading-width is the maximum width that a column in this region will be assigned to accomodate a long label. This region doesn't include data, so data values don't affect column widths.

table-id is a binary version of the tableId attribute in the structure member that refers to the detail member. For example, if tableId is -4122591256483201023, then table-id would be 0xc6c99d183b300001.

The meaning of the other variable parts of the header is not known. A writer may safely use version 3, true for x0, false for x1, true for x2, and 0x15 for x3.

Titles

Titles =>
    Value[title] 01?
    Value[subtype] 01? 31
    Value[user-title] 01?
    (31 Value[corner-text] | 58)
    (31 Value[caption] | 58)

The Titles follow the Header and specify the table's title, caption, and corner text.

The user-title reflects any user editing of the title text or style. The title is the title originally generated by the procedure. Both of these are appropriate for presentation and localized to the user's language. For example, for a frequency table, title and user-title normally name the variable and c is simply "Frequencies".

subtype is the same as the subType attribute in the table structure XML element that referred to this member.

The corner-text, if present, is shown in the upper-left corner of the table, above the row headings and to the left of the column headings. It is usually absent. When row dimension labels are displayed in the corner (see show-row-labels-in-corner), corner text is hidden.

The caption, if present, is shown below the table. caption reflects user editing of the caption.

Footnotes

Footnotes => int32[n-footnotes] Footnote*[n-footnotes]
Footnote => Value[text] (58 | 31 Value[marker]) int32[show]

Each footnote has text and an optional custom marker (such as *).

The syntax for Value would allow footnotes (and their markers) to reference other footnotes, but in practice this doesn't work.

show is a 32-bit signed integer. It is positive to show the footnote or negative to hide it. Its magnitude is often 1, and in other cases tends to be the number of references to the footnote. It is safe to write 1 to show a footnote and -1 to hide it.

Areas

Areas => 00? Area*8
Area =>
    byte[index] 31
    string[typeface] float[size] int32[style] bool[underline]
    int32[halign] int32[valign]
    string[fg-color] string[bg-color]
    bool[alternate] string[alt-fg-color] string[alt-bg-color]
    v3(int32[left-margin] int32[right-margin] int32[top-margin] int32[bottom-margin])

Each Area represents the style for a different area of the table. index is the 1-based index of the Area, i.e. 1 for the first Area, through 8 for the final Area. The following table shows the index values and the areas that they represent:

indexArea
1Title
2Caption
3Footer
4Corner
5Column labels
6Row labels
7Data
8Layers

typeface is the string name of the font used in the area. In the corpus, this is SansSerif in over 99% of instances and Times New Roman in the rest.

size is the size of the font, in px. The most common size in the corpus is 12 px. Even though size has a floating-point type, in the corpus its values are always integers.

style is a bit mask. Bit 0 (with value 1) is set for bold, bit 1 (with value 2) is set for italic.

underline is 1 if the font is underlined, 0 otherwise.

halign specifies horizontal alignment:

halignAlignment
0Center
2Left
4Right
64173Mixed

Mixed alignment varies according to type: string data is left-justified, numbers and most other formats are right-justified.

valign specifies vertical alignment:

valignAlignment
0Center
1Top
3Bottom

fg-color and bg-color are the foreground color and background color, respectively. In the corpus, these are always #000000 and #ffffff, respectively.

alternate is 1 if rows should alternate colors, 0 if all rows should be the same color. When alternate is 1, alt-fg-color and alt-bg-color specify the colors for the alternate rows; otherwise they are empty strings.

left-margin, right-margin, top-margin, and bottom-margin are measured in px.

Borders

Borders =>
    count(
        ib1[endian]
        be32[n-borders] Border*[n-borders]
        bool[show-grid-lines]
        00 00 00)

Border =>
    be32[index]
    be32[stroke-type]
    be32[color]

Borders reflects how borders between regions are drawn.

The fixed value of endian can be used to validate the endianness.

show-grid-lines is 1 to draw grid lines, otherwise 0.

Each Border describes one kind of border. n-borders seems to always be 19. Each index appears once (although in an unpredictable order) and correspond to the following borders:

indexBorders
0Title.
1...4Left, top, right, and bottom outer frame.
5...8Left, top, right, and bottom inner frame.
9, 10Left and top of data area.
11, 12Horizontal and vertical dimension rows.
13, 14Horizontal and vertical dimension columns.
15, 16Horizontal and vertical category rows.
17, 18Horizontal and vertical category columns.

stroke-type describes how a border is drawn, as one of:

stroke-typeBorder style
0No line.
1Solid line.
2Dashed line.
3Thick line.
4Thin line.
5Double line.

color is an RGB color. Bits 24-31 are alpha, bits 16-23 are red, 8-15 are green, 0-7 are blue. An alpha of 255 indicates an opaque color, therefore opaque black is 0xff000000.

PrintSettings =>
    count(
        ib1[endian]
        bool[all-layers]
        bool[paginate-layers]
        bool[fit-width]
        bool[fit-length]
        bool[top-continuation]
        bool[bottom-continuation]
        be32[n-orphan-lines]
        bestring[continuation-string])

PrintSettings reflects settings for printing. The fixed value of endian can be used to validate the endianness.

all-layers is 1 to print all layers, 0 to print only the layer designated by current-layer in TableSettings.

paginate-layers is 1 to print each layer at the start of a new page, 0 otherwise. (This setting is honored only all-layers is 1, since otherwise only one layer is printed.)

fit-width and fit-length control whether the table is shrunk to fit within a page's width or length, respectively.

n-orphan-lines is the minimum number of rows or columns to put in one part of a table that is broken across pages.

If top-continuation is 1, then continuation-string is printed at the top of a page when a table is broken across pages for printing; similarly for bottom-continuation and the bottom of a page. Usually, continuation-string is empty.

Table Settings

TableSettings =>
    count(
      v3(
        ib1[endian]
        be32[x5]
        be32[current-layer]
        bool[omit-empty]
        bool[show-row-labels-in-corner]
        bool[show-alphabetic-markers]
        bool[footnote-marker-superscripts]
        byte[x6]
        becount(
          Breakpoints[row-breaks] Breakpoints[column-breaks]
          Keeps[row-keeps] Keeps[column-keeps]
          PointKeeps[row-point-keeps] PointKeeps[column-point-keeps]
        )
        bestring[notes]
        bestring[table-look]
        )...)

Breakpoints => be32[n-breaks] be32*[n-breaks]

Keeps => be32[n-keeps] Keep*[n-keeps]
Keep => be32[offset] be32[n]

PointKeeps => be32[n-point-keeps] PointKeep*[n-point-keeps]
PointKeep => be32[offset] be32 be32

TableSettings reflects display settings. The fixed value of endian can be used to validate the endianness.

current-layer is the displayed layer. Suppose there are \(d\) layers, numbered 1 through \(d\) in the order given in the Dimensions, and that the displayed value of dimension \(i\) is \(d_i, 0 \le x_i < n_i\), where \(n_i\) is the number of categories in dimension \(i\). Then current-layer is the \(k\) calculated by the following algorithm:

let \(k = 0\).
for each \(i\) from \(d\) downto 1:
\(\quad k = (n_i \times k) + x_i\).

If omit-empty is 1, empty rows or columns (ones with nothing in any cell) are hidden; otherwise, they are shown.

If show-row-labels-in-corner is 1, then row labels are shown in the upper left corner; otherwise, they are shown nested.

If show-alphabetic-markers is 1, markers are shown as letters (e.g. a, b, c, ...); otherwise, they are shown as numbers starting from 1.

When footnote-marker-superscripts is 1, footnote markers are shown as superscripts, otherwise as subscripts.

The Breakpoints are rows or columns after which there is a page break; for example, a row break of 1 requests a page break after the second row. Usually no breakpoints are specified, indicating that page breaks should be selected automatically.

The Keeps are ranges of rows or columns to be kept together without a page break; for example, a row Keep with offset 1 and n 10 requests that the 10 rows starting with the second row be kept together. Usually no Keeps are specified.

The PointKeeps seem to be generated automatically based on user-specified Keeps. They seems to indicate a conversion from rows or columns to pixel or point offsets.

notes is a text string that contains user-specified notes. It is displayed when the user hovers the cursor over the table, like text in the title attribute in HTML. It is not printed. It is usually empty.

table-look is the name of a SPSS "TableLook" table style, such as "Default" or "Academic"; it is often empty.

TableSettings ends with an arbitrary number of null bytes. A writer may safely write 82 null bytes.

A writer may safely use 4 for x5 and 0 for x6.

Formats

Formats =>
    int32[n-widths] int32*[n-widths]
    string[locale]
    int32[current-layer]
    bool[x7] bool[x8] bool[x9]
    Y0
    CustomCurrency
    count(
      v1(X0?)
      v3(count(X1 count(X2)) count(X3)))
Y0 => int32[epoch] byte[decimal] byte[grouping]
CustomCurrency => int32[n-ccs] string*[n-ccs]

If n-widths is nonzero, then the accompanying integers are column widths as manually adjusted by the user.

locale is a locale including an encoding, such as en_US.windows-1252 or it_IT.windows-1252. (locale is often duplicated in Y1, described below).

epoch is the year that starts the epoch. A 2-digit year is interpreted as belonging to the 100 years beginning at the epoch. The default epoch year is 69 years prior to the current year; thus, in 2017 this field by default contains 1948. In the corpus, epoch ranges from 1943 to 1948, plus some contain -1.

decimal is the decimal point character. The observed values are . and ,.

grouping is the grouping character. Usually, it is , if decimal is ., and vice versa. Other observed values are ' (apostrophe), (space), and zero (presumably indicating that digits should not be grouped).

n-ccs is observed as either 0 or 5. When it is 5, the following strings are CCA through CCE format strings. Most commonly these are all -,,, but other strings occur.

A writer may safely use false for x7, x8, and x9.

X0

X0 only appears, optionally, in version 1 members.

X0 => byte*14 Y1 Y2
Y1 =>
    string[command] string[command-local]
    string[language] string[charset] string[locale]
    bool[x10] bool[include-leading-zero] bool[x12] bool[x13]
    Y0
Y2 => CustomCurrency byte[missing] bool[x17]

command describes the statistical procedure that generated the output, in English. It is not necessarily the literal syntax name of the procedure: for example, NPAR TESTS becomes "Nonparametric Tests." command-local is the procedure's name, translated into the output language; it is often empty and, when it is not, sometimes the same as command.

include-leading-zero is the LEADZERO setting for the table, where false is OFF (the default) and true is ON.

missing is the character used to indicate that a cell contains a missing value. It is always observed as ..

A writer may safely use false for x10 and x17 and true for x12 and x13.

X1

X1 only appears in version 3 members.

X1 =>
    bool[x14]
    byte[show-title]
    bool[x16]
    byte[lang]
    byte[show-variables]
    byte[show-values]
    int32[x18] int32[x19]
    00*17
    bool[x20]
    bool[show-caption]

lang may indicate the language in use. Some values and their apparent meanings are:

ValueLanguage
0en
1de
2es
3it
5ko
6pl
8zh-tw
10pt_BR
11fr

show-variables determines how variables are displayed by default:

ValueMeaning
0Use global default (the most common value)
1Variable name only
2Variable label only (when available)
3Both (name followed by label, separated by a space)

show-values is a similar setting for values:

ValueMeaning
0Use global default (the most common value)
1Value only
2Value label only (when available)
3Both

show-title is 1 to show the caption, 10 to hide it.

show-caption is true to show the caption, false to hide it.

A writer may safely use false for x14, false for x16, 0 for lang, -1 for x18 and x19, and false for x20.

X2

X2 only appears in version 3 members.

X2 =>
    int32[n-row-heights] int32*[n-row-heights]
    int32[n-style-map] StyleMap*[n-style-map]
    int32[n-styles] StylePair*[n-styles]
    count((i0 i0)?)
StyleMap => int64[cell-index] int16[style-index]

If present, n-row-heights and the accompanying integers are row heights as manually adjusted by the user.

The rest of X2 specifies styles for data cells. At first glance this is odd, because each data cell can have its own style embedded as part of the data, but in practice X2 specifies a style for a cell only if that cell is empty (and thus does not appear in the data at all). Each StyleMap specifies the index of a blank cell, calculated the same was as in the Cells, along with a 0-based index into the accompanying StylePair array.

A writer may safely omit the optional i0 i0 inside the count(...).

X3

X3 only appears in version 3 members.

X3 =>
    01 00 byte[x21] 00 00 00
    Y1
    double[small] 01
    (string[dataset] string[datafile] i0 int32[date] i0)?
    Y2
    (int32[x22] i0 01?)?

small is a small real number. In the corpus, it overwhelmingly takes the value 0.0001, with zero occasionally seen. Nonzero numbers with format 40 (see Value) whose magnitudes are smaller than displayed in scientific notation. (Thus, a small of zero prevents scientific notation from being chosen.)

dataset is the name of the dataset analyzed to produce the output, e.g. DataSet1, and datafile the name of the file it was read from, e.g. C:\Users\foo\bar.sav. The latter is sometimes the empty string.

date is a date, as seconds since the epoch, i.e. since January 1, 1970. Pivot tables within an SPV file often have dates a few minutes apart, so this is probably a creation date for the table rather than for the file.

Sometimes dataset, datafile, and date are present and other times they are absent. The reader can distinguish by assuming that they are present and then checking whether the presumptive dataset contains a null byte (a valid string never will).

x22 is usually 0 or 2000000.

A writer may safely use 4 for x21 and omit x22 and the other optional bytes at the end.

Encoding

Formats contains several indications of character encoding:

  • locale in Formats itself.

  • locale in Y1 (in version 1, Y1 is optionally nested inside X0; in version 3, Y1 is nested inside X3).

  • charset in version 3, in Y1.

  • lang in X1, in version 3.

charset, if present, is a good indication of character encoding, and in its absence the encoding suffix on locale in Formats will work.

A reader may disregard locale in Y1, because it is normally the same as locale in Formats, and it is only present if charset is also.

lang is not helpful and should be ignored for character encoding purposes.

However, the corpus contains many examples of light members whose strings are encoded in UTF-8 despite declaring some other character set. Furthermore, the corpus contains several examples of light members in which some strings are encoded in UTF-8 (and contain multibyte characters) and other strings are encoded in another character set (and contain non-ASCII characters). PSPP treats any valid UTF-8 string as UTF-8 and only falls back to the declared encoding for strings that are not valid UTF-8.

The pspp-output program's strings command can help analyze the encoding in an SPV light member. Use pspp-output --help-dev to see its usage.

Dimensions

A pivot table presents multidimensional data. A Dimension identifies the categories associated with each dimension.

Dimensions => int32[n-dims] Dimension*[n-dims]
Dimension =>
    Value[name] DimProperties
    int32[n-categories] Category*[n-categories]
DimProperties =>
    byte[x1]
    byte[x2]
    int32[x3]
    bool[hide-dim-label]
    bool[hide-all-labels]
    01 int32[dim-index]

name is the name of the dimension, e.g. Variables, Statistics, or a variable name.

The meanings of x1 and x3 are unknown. x1 is usually 0 but many other values have been observed. A writer may safely use 0 for x1 and 2 for x3.

x2 is 0, 1, or 2. For a pivot table with L layer dimensions, R row dimensions, and C column dimensions, x2 is 2 for the first L dimensions, 0 for the next R dimensions, and 1 for the remaining C dimensions. This does not mean that the layer dimensions must be presented first, followed by the row dimensions, followed by the column dimensions--on the contrary, they are frequently in a different order—but x2 must follow this pattern to prevent the pivot table from being misinterpreted.

If hide-dim-label is 00, the pivot table displays a label for the dimension itself. Because usually the group and category labels are enough explanation, it is usually 01.

If hide-all-labels is 01, the pivot table omits all labels for the dimension, including group and category labels. It is usually 00. When hide-all-labels is 01, hide-dim-label is ignored.

dim-index is usually the 0-based index of the dimension, e.g. 0 for the first dimension, 1 for the second, and so on. Sometimes it is -1. There is no visible difference. A writer may safely use the 0-based index.

Categories

Categories are arranged in a tree. Only the leaf nodes in the tree are really categories; the others just serve as grouping constructs.

Category => Value[name] (Leaf | Group)
Leaf => 00 00 bool[x24] i2 int32[leaf-index] i0
Group =>
    bool[merge] 00 01 int32[x23]
    i-1 int32[n-subcategories] Category*[n-subcategories]

name is the name of the category (or group).

A Leaf represents a leaf category. The Leaf's leaf-index is a nonnegative integer unique within the Dimension and less than n-categories in the Dimension. If the user does not sort or rearrange the categories, then leaf-index starts at 0 for the first Leaf in the dimension and increments by 1 with each successive Leaf. If the user does sort or rearrange the categories, then the order of categories in the file reflects that change and leaf-index reflects the original order.

A dimension can have no leaf categories at all. A table that contains such a dimension necessarily has no data at all.

A Group is a group of nested categories. Usually a Group contains at least one Category, so that n-subcategories is positive, but Groups with zero subcategories have been observed.

If a Group's merge is 00, the most common value, then the group is really a distinct group that should be represented as such in the visual representation and user interface. If merge is 01, the categories in this group should be shown and treated as if they were direct children of the group's containing group (or if it has no parent group, then direct children of the dimension), and this group's name is irrelevant and should not be displayed. (Merged groups can be nested!)

Writers need not use merged groups.

A Group's x23 appears to be i2 when all of the categories within a group are leaf categories that directly represent data values for a variable (e.g. in a frequency table or crosstabulation, a group of values in a variable being tabulated) and i0 otherwise. A writer may safely write a constant 0 in this field.

x24 is usually 0. Its meaning is unexplored.

Axes

After the dimensions come assignment of each dimension to one of the axes: layers, rows, and columns.

Axes =>
    int32[n-layers] int32[n-rows] int32[n-columns]
    int32*[n-layers] int32*[n-rows] int32*[n-columns]

The values of n-layers, n-rows, and n-columns each specifies the number of dimensions displayed in layers, rows, and columns, respectively. Any of them may be zero. Their values sum to n-dimensions from Dimensions.

The following n-dimensions integers, in three groups, are a permutation of the 0-based dimension numbers. The first n-layers integers specify each of the dimensions represented by layers, the next n-rows integers specify the dimensions represented by rows, and the final n-columns integers specify the dimensions represented by columns. When there is more than one dimension of a given kind, the inner dimensions are given first. (For the layer axis, this means that the first dimension is at the bottom of the list and the last dimension is at the top when the current layer is displayed.)

Cells

The final part of an SPV light member contains the actual data.

Cells => int32[n-cells] Cell*[n-cells]
Cell => int64[index] v1(00?) Value

A Cell consists of an index and a Value. Suppose there are \(d\) dimensions, numbered 1 through \(d\) in the order given in the Dimensions previously, and that dimension \(i\) has \(n_i\) categories. Consider the cell at coordinates \(x_i, 1 \le i \le d\), and note that \(0 \le x_i < n_i\). Then the index \(k\) is calculated by the following algorithm:

let \(k = 0\).
for each \(i\) from 1 to \(d\):
\(\quad k = (n_i \times k) + x_i\)

For example, suppose there are 3 dimensions with 3, 4, and 5 categories, respectively. The cell at coordinates (1, 2, 3) has index \(k = 5 \times (4 \times (3 \times 0 + 1) + 2) + 3 = 33\). Within a given dimension, the index is the leaf-index in a Leaf.

Value

Value is used throughout the SPV light member format. It boils down to a number or a string.

Value => 00? 00? 00? 00? RawValue
RawValue =>
    01 ValueMod int32[format] double[x]
  | 02 ValueMod int32[format] double[x]
    string[var-name] string[value-label] byte[show]
  | 03 string[local] ValueMod string[id] string[c] bool[fixed]
  | 04 ValueMod int32[format] string[value-label] string[var-name]
    byte[show] string[s]
  | 05 ValueMod string[var-name] string[var-label] byte[show]
  | 06 string[local] ValueMod string[id] string[c]
  | ValueMod string[template] int32[n-args] Argument*[n-args]
Argument =>
    i0 Value
  | int32[x] i0 Value*[x]      /* x > 0 */

There are several possible encodings, which one can distinguish by the first nonzero byte in the encoding.

  • 01
    The numeric value x, intended to be presented to the user formatted according to format, which is about the same as the format described for system files. The exception is that format 40 is not MTIME but instead approximately a synonym for F format with a different rule for whether a value is shown in scientific notation: a value in format 40 is shown in scientific notation if and only if it is nonzero and its magnitude is less than small.

    Values of 0 or 1 or 0x10000 are sometimes seen as format. PSPP interprets these as F40.2.

    Most commonly, format has width 40 (the maximum).

    An x with the maximum negative double value -DBL_MAX represents the system-missing value SYSMIS. (HIGHEST and LOWEST have not been observed.) See System File Format for more about these special values.

  • 02
    Similar to 01, with the additional information that x is a value of variable var-name and has value label value-label. Both var-name and value-label can be the empty string, the latter very commonly.

    show determines whether to show the numeric value or the value label:

    showMeaning
    0Use default specified in show-values
    1Value only
    2Label only
    3Both value and label

  • 03
    A text string, in two forms: c is in English, and sometimes abbreviated or obscure, and local is localized to the user's locale. In an English-language locale, the two strings are often the same, and in the cases where they differ, local is more appropriate for a user interface, e.g. c of "Not a PxP table for MCN..." versus local of "Computed only for a PxP table, where P must be greater than 1."

    c and local are always either both empty or both nonempty.

    id is a brief identifying string whose form seems to resemble a programming language identifier, e.g. cumulative_percent or factor_14. It is not unique.

    fixed is:

    • 00 for text taken from user input, such as syntax fragment, expressions, file names, data set names. id is always the empty string.

    • 01 for fixed text strings such as names of procedures or statistics. id is sometimes empty.

  • 04
    The string value s, intended to be presented to the user formatted according to format. The format for a string is not too interesting, and the corpus contains many clearly invalid formats like A16.39 or A255.127 or A134.1, so readers should probably entirely disregard the format. PSPP only checks format to distinguish AHEX format.

    s is a value of variable var-name and has value label value-label. var-name is never empty but value-label is commonly empty.

    show has the same meaning as in the encoding for 02.

  • 05
    Variable var-name with variable label var-label. In the corpus, var-name is rarely empty and var-label is often empty.

    show determines whether to show the variable name or the variable label. A value of 1 means to show the name, 2 to show the label, 3 to show both, and 0 means to use the default specified in show-variables.

  • 06
    Similar to type 03, with fixed assumed to be true.

  • otherwise
    When the first byte of a RawValue is not one of the above, the RawValue starts with a ValueMod, whose syntax is described in the next section. (A ValueMod always begins with byte 31 or 58.)

    This case is a template string, analogous to printf, followed by one or more Arguments, each of which has one or more values. The template string is copied directly into the output except for the following special syntax:

    • \%
      \:
      \[
      \]
      Each of these expands to the character following \\, to escape characters that have special meaning in template strings. These are effective inside and outside the [...] syntax forms described below.

    • \n
      Expands to a new-line, inside or outside the [...] forms described below.

    • ^I
      Expands to a formatted version of argument I, which must have only a single value. For example, ^1 expands to the first argument's value.

    • [:A:]I
      Expands A for each of the values in I. A should contain one or more ^J conversions, which are drawn from the values for argument I in order. Some examples from the corpus:

      • [:^1:]1
        All of the values for the first argument, concatenated.

      • [:^1\n:]1
        Expands to the values for the first argument, each followed by a new-line.

      • [:^1 = ^2:]2
        Expands to X = Y where X is the second argument's first alue and Y is its second value. (This would be used only if the argument has two values. If there were more values, the second and third values would be directly concatenated, which would look funny.)

    • [A:B:]I
      This extends the previous form so that the first values are expanded using A and later values are expanded using B. For an unknown reason, within A the ^J conversions are instead written as %J. Some examples from the corpus:

      • [%1:*^1:]1
        Expands to all of the values for the first argument, separated by *.

      • [%1 = %2:, ^1 = ^2:]1
        Given appropriate values for the first argument, expands to X = 1, Y = 2, Z = 3.

      • [%1:, ^1:]1
        Given appropriate values, expands to 1, 2, 3.

    The template string is localized to the user's locale.

A writer may safely omit all of the optional 00 bytes at the beginning of a Value, except that it should write a single 00 byte before a templated Value.

ValueMod

A ValueMod can specify special modifications to a Value.

ValueMod =>
    58
  | 31
    int32[n-refs] int16*[n-refs]
    int32[n-subscripts] string*[n-subscripts]
    v1(00 (i1 | i2) 00? 00? int32 00? 00?)
    v3(count(TemplateString StylePair))

TemplateString => count((count((i0 (58 | 31 55))?) (58 | 31 string[id]))?)

StylePair =>
    (31 FontStyle | 58)
    (31 CellStyle | 58)

FontStyle =>
    bool[bold] bool[italic] bool[underline] bool[show]
    string[fg-color] string[bg-color]
    string[typeface] byte[size]

CellStyle =>
    int32[halign] int32[valign] double[decimal-offset]
    int16[left-margin] int16[right-margin]
    int16[top-margin] int16[bottom-margin]

A ValueMod that begins with 31 specifies special modifications to a Value.

Each of the n-refs integers is a reference to a Footnote by a 0-based index. Footnote markers are shown appended to the main text of the Value, as superscripts or subscripts.

The subscripts, if present, are strings to append to the main text of the Value, as subscripts. Each subscript text is a brief indicator, e.g. a or b, with its meaning indicated by the table caption. When multiple subscripts are present, they are displayed separated by commas.

The id inside the TemplateString, if present, is a template string for substitutions using the syntax explained previously. It appears to be an English-language version of the localized template string in the Value in which the Template is nested. A writer may safely omit the optional fixed data in TemplateString.

FontStyle and CellStyle, if present, change the style for this individual Value. In FontStyle, bold, italic, and underline control the particular style. show is ordinarily 1; if it is 0, then the cell data is not shown. fg-color and bg-color are strings in the format #rrggbb, e.g. #ff0000 for red or #ffffff for white. The empty string is occasionally observed also. The size is a font size in units of 1/128 inch.

In CellStyle, halign specified horizontal alignment:

halignMeaning
0Center
2Left
4Right
6Decimal
0xffffffadMixed

For decimal alignment, decimal-offset is the decimal point's offset from the right side of the cell, in pt.

valign specifies vertical alignment:

valignMeaning
0Center
1Top
3Bottom

left-margin, right-margin, top-margin, and bottom-margin are in pt.

Legacy Detail Member Binary Format

Whereas the light binary format represents everything about a given pivot table, the legacy binary format conceptually consists of a number of named sources, each of which consists of a number of named variables, each of which is a 1-dimensional array of numbers or strings or a mix. Thus, the legacy binary member format is quite simple.

This section uses the same context-free grammar notation as in the previous section, with the following additions:

  • vAF(X)
    In a version 0xaf legacy member, X; in other versions, nothing. (The legacy member header indicates the version; see below.)

  • vB0(X)
    In a version 0xb0 legacy member, X; in other versions, nothing.

A legacy detail member .bin has the following overall format:

LegacyBinary =>
    00 byte[version] int16[n-sources] int32[member-size]
    Metadata*[n-sources]
    #Data*[n-sources]
    #Strings?

version is a version number that affects the interpretation of some of the other data in the member. Versions 0xaf and 0xb0 are known. We will refer to "version 0xaf" and "version 0xb0" members later on.

A legacy member consists of n-sources data sources, each of which has Metadata and Data.

member-size is the size of the legacy binary member, in bytes.

The Data and Strings above are commented out because the Metadata has some oddities that mean that the Data sometimes seems to start at an unexpected place. The following section goes into detail.

Metadata

Metadata =>
    int32[n-values] int32[n-variables] int32[data-offset]
    vAF(byte*28[source-name])
    vB0(byte*64[source-name] int32[x])

A data source has n-variables variables, each with n-values data values.

source-name is a 28- or 64-byte string padded on the right with 0-bytes. The names that appear in the corpus are very generic: usually tableData for pivot table data or source0 for chart data.

A given Metadata's data-offset is the offset, in bytes, from the beginning of the member to the start of the corresponding Data. This allows programs to skip to the beginning of the data for a particular source. In every case in the corpus, the Data follow the Metadata in the same order, but it is important to use data-offset instead of reading sequentially through the file because of the exception described below.

One SPV file in the corpus has legacy binary members with version 0xb0 but a 28-byte source-name field (and only a single source). In practice, this means that the 64-byte source-name used in version 0xb0 has a lot of 0-bytes in the middle followed by the variable-name of the following Data. As long as a reader treats the first 0-byte in the source-name as terminating the string, it can properly interpret these members.

The meaning of x in version 0xb0 is unknown.

Numeric Data

Data => Variable*[n-variables]
Variable => byte*288[variable-name] double*[n-values]

Data follow the Metadata in the legacy binary format, with sources in the same order (but readers should use the data-offset in Metadata records, rather than reading sequentially). Each Variable begins with a variable-name that generally indicates its role in the pivot table, e.g. "cell", "cellFormat", "dimension0categories", "dimension0group0", followed by the numeric data, one double per datum. A double with the maximum negative double -DBL_MAX represents the system-missing value SYSMIS.

String Data

Strings => SourceMaps[maps] Labels

SourceMaps => int32[n-maps] SourceMap*[n-maps]

SourceMap => string[source-name] int32[n-variables] VariableMap*[n-variables]
VariableMap => string[variable-name] int32[n-data] DatumMap*[n-data]
DatumMap => int32[value-idx] int32[label-idx]

Labels => int32[n-labels] Label*[n-labels]
Label => int32[frequency] string[label]

Each variable may include a mix of numeric and string data values. If a legacy binary member contains any string data, Strings is present; otherwise, it ends just after the last Data element.

The string data overlays the numeric data. When a variable includes any string data, its Variable represents the string values with a SYSMIS or NaN placeholder. (Not all such values need be placeholders.)

Each SourceMap provides a mapping between SYSMIS or NaN values in source source-name and the string data that they represent. n-variables is the number of variables in the source that include string data. More precisely, it is the 1-based index of the last variable in the source that includes any string data; thus, it would be 4 if there are 5 variables and only the fourth one includes string data.

A VariableMap repeats its variable's name, but variables are always present in the same order as the source, starting from the first variable, without skipping any even if they have no string values. Each VariableMap contains DatumMap nonterminals, each of which maps from a 0-based index within its variable's data to a 0-based label index, e.g. pair value-idx = 2, label-idx = 3, means that the third data value (which must be SYSMIS or NaN) is to be replaced by the string of the fourth Label.

The labels themselves follow the pairs. The valuable part of each label is the string label. Each label also includes a frequency that reports the number of DatumMaps that reference it (although this is not useful).

Legacy Detail XML Member Format

The design of the detail XML format is not what one would end up with for describing pivot tables. This is because it is a special case of a much more general format ("visualization XML" or "VizML") that can describe a wide range of visualizations. Most of this generality is overkill for tables, and so we end up with a funny subset of a general-purpose format.

An XML Schema for VizML is available, distributed with SPSS binaries, under a nonfree license. It contains documentation that is occasionally helpful.

This section describes the detail XML format using the same notation already used for the structure XML format. See src/output/spv/detail-xml.grammar in the PSPP source tree for the full grammar that it uses for parsing.

The important elements of the detail XML format are:

  • Variables.

  • Assignment of variables to axes. A variable can appear as columns, or rows, or layers. The faceting element and its sub-elements describe this assignment.

  • Styles and other annotations.

This description is not detailed enough to write legacy tables. Instead, write tables in the light binary format.

The visualization Element

visualization
   :creator
   :date
   :lang
   :name
   :style[style_ref]=ref style
   :type
   :version
   :schemaLocation?
=> visualization_extension?
   userSource
   (sourceVariable | derivedVariable)+
   categoricalDomain?
   graph
   labelFrame[lf1]*
   container?
   labelFrame[lf2]*
   style+
   layerController?

extension[visualization_extension]
   :numRows=int?
   :showGridline=bool?
   :minWidthSet=(true)?
   :maxWidthSet=(true)?
=> EMPTY

userSource :missing=(listwise | pairwise)? => EMPTY

categoricalDomain => variableReference simpleSort

simpleSort :method[sort_method]=(custom) => categoryOrder

container :style=ref style => container_extension? location+ labelFrame*

extension[container_extension] :combinedFootnotes=(true) => EMPTY

layerController
   :source=(tableData)
   :target=ref label?
=> EMPTY

The visualization element is the root of detail XML member. It has the following attributes:

  • creator
    The version of the software that created this SPV file, as a string of the form xxyyzz, which represents software version xx.yy.zz, e.g. 160001 is version 16.0.1. The corpus includes major versions 16 through 19.

  • date
    The date on the which the file was created, as a string of the form YYYY-MM-DD.

  • lang
    The locale used for output, in Windows format, which is similar to the format used in Unix with the underscore replaced by a hyphen, e.g. en-US, en-GB, el-GR, sr-Cryl-RS.

  • name
    The title of the pivot table, localized to the output language.

  • style
    The base style for the pivot table. In every example in the corpus, the style element has no attributes other than id.

  • type
    A floating-point number. The meaning is unknown.

  • version
    The visualization schema version number. In the corpus, the value is one of 2.4, 2.5, 2.7, and 2.8.

The userSource element has no visible effect.

The extension element as a child of visualization has the following attributes.

  • numRows
    An integer that presumably defines the number of rows in the displayed pivot table.

  • showGridline
    Always set to false in the corpus.

  • minWidthSet

  • maxWidthSet
    Always set to true in the corpus.

The extension element as a child of container has the following attribute

  • combinedFootnotes
    Meaning unknown.

The categoricalDomain and simpleSort elements have no visible effect.

The layerController element has no visible effect.

Variable Elements

A "variable" in detail XML is a 1-dimensional array of data. Each element of the array may, independently, have string or numeric content. All of the variables in a given detail XML member either have the same number of elements or have zero elements.

Two different elements define variables and their content:

  • sourceVariable
    These variables' data comes from the associated tableData.bin member.

  • derivedVariable
    These variables are defined in terms of a mapping function from a source variable, or they are empty.

A variable named cell always exists. This variable holds the data displayed in the table.

Variables in detail XML roughly correspond to the dimensions in a light detail member. Each dimension has the following variables with stylized names, where N is a number for the dimension starting from 0:

  • dimensionNcategories
    The dimension's leaf categories.

  • dimensionNgroup0
    Present only if the dimension's categories are grouped, this variable holds the group labels for the categories. Grouping is inferred through adjacent identical labels. Categories that are not part of a group have empty-string data in this variable.

  • dimensionNgroup1
    Present only if the first-level groups are further grouped, this variable holds the labels for the second-level groups. There can be additional variables with further levels of grouping.

  • dimensionN
    An empty variable.

    Determining the data for a (non-empty) variable is a multi-step process:

  1. Draw initial data from its source, for a sourceVariable, or from another named variable, for a derivedVariable.

  2. Apply mappings from valueMapEntry elements within the derivedVariable element, if any.

  3. Apply mappings from relabel elements within a format or stringFormat element in the sourceVariable or derivedVariable element, if any.

  4. If the variable is a sourceVariable with a labelVariable attribute, and there were no mappings to apply in previous steps, then replace each element of the variable by the corresponding value in the label variable.

A single variable's data can be modified in two of the steps, if both valueMapEntry and relabel are used. The following example from the corpus maps several integers to 2, then maps 2 in turn to the string "Input":

<derivedVariable categorical="true" dependsOn="dimension0categories"
                 id="dimension0group0map" value="map(dimension0group0)">
  <stringFormat>
    <relabel from="2" to="Input"/>
    <relabel from="10" to="Missing Value Handling"/>
    <relabel from="14" to="Resources"/>
    <relabel from="0" to=""/>
    <relabel from="1" to=""/>
    <relabel from="13" to=""/>
  </stringFormat>
  <valueMapEntry from="2;3;5;6;7;8;9" to="2"/>
  <valueMapEntry from="10;11" to="10"/>
  <valueMapEntry from="14;15" to="14"/>
  <valueMapEntry from="0" to="0"/>
  <valueMapEntry from="1" to="1"/>
  <valueMapEntry from="13" to="13"/>
</derivedVariable>

The sourceVariable Element

sourceVariable
   :id
   :categorical=(true)
   :source
   :domain=ref categoricalDomain?
   :sourceName
   :dependsOn=ref sourceVariable?
   :label?
   :labelVariable=ref sourceVariable?
=> variable_extension* (format | stringFormat)?

This element defines a variable whose data comes from the tableData.bin member that corresponds to this .xml.

This element has the following attributes.

  • id
    An id is always present because this element exists to be referenced from other elements.

  • categorical
    Always set to true.

  • source
    Always set to tableData, the source-name in the corresponding tableData.bin member (see Metadata).

  • sourceName
    The name of a variable within the source, corresponding to the variable-name in the tableData.bin member (see Numeric Data).

  • label
    The variable label, if any.

  • labelVariable
    The variable-name of a variable whose string values correspond one-to-one with the values of this variable and are suitable for use as value labels.

  • dependsOn
    This attribute doesn't affect the display of a table.

The derivedVariable Element

derivedVariable
   :id
   :categorical=(true)
   :value
   :dependsOn=ref sourceVariable?
=> variable_extension* (format | stringFormat)? valueMapEntry*

Like sourceVariable, this element defines a variable whose values can be used elsewhere in the visualization. Instead of being read from a data source, the variable's data are defined by a mathematical expression.

This element has the following attributes.

  • id
    An id is always present because this element exists to be referenced from other elements.

  • categorical
    Always set to true.

  • value
    An expression that defines the variable's value. In theory this could be an arbitrary expression in terms of constants, functions, and other variables, e.g. (VAR1 + VAR2) / 2. In practice, the corpus contains only the following forms of expressions:

    • constant(0)
      constant(VARIABLE)
      All zeros. The reason why a variable is sometimes named is unknown. Sometimes the "variable name" has spaces in it.

    • map(VARIABLE)
      Transforms the values in the named VARIABLE using the valueMapEntrys contained within the element.

  • dependsOn
    This attribute doesn't affect the display of a table.

The valueMapEntry Element

valueMapEntry :from :to => EMPTY

A valueMapEntry element defines a mapping from one or more values of a source expression to a target value. (In the corpus, the source expression is always just the name of a variable.) Each target value requires a separate valueMapEntry. If multiple source values map to the same target value, they can be combined or separate.

In the corpus, all of the source and target values are integers.

valueMapEntry has the following attributes.

  • from
    A source value, or multiple source values separated by semicolons, e.g. 0 or 13;14;15;16.

  • to
    The target value, e.g. 0.

The extension Element

This is a general-purpose "extension" element. Readers that don't understand a given extension should be able to safely ignore it. The attributes on this element, and their meanings, vary based on the context. Each known usage is described separately below. The current extensions use attributes exclusively, without any nested elements.

container Parent Element

extension[container_extension] :combinedFootnotes=(true) => EMPTY

With container as its parent element, extension has the following attributes.

  • combinedFootnotes
    Always set to true in the corpus.

sourceVariable and derivedVariable Parent Element

extension[variable_extension] :from :helpId => EMPTY

With sourceVariable or derivedVariable as its parent element, extension has the following attributes. A given parent element often contains several extension elements that specify the meaning of the source data's variables or sources, e.g.

<extension from="0" helpId="corrected_model"/>
<extension from="3" helpId="error"/>
<extension from="4" helpId="total_9"/>
<extension from="5" helpId="corrected_total"/>

More commonly they are less helpful, e.g.

<extension from="0" helpId="notes"/>
<extension from="1" helpId="notes"/>
<extension from="2" helpId="notes"/>
<extension from="5" helpId="notes"/>
<extension from="6" helpId="notes"/>
<extension from="7" helpId="notes"/>
<extension from="8" helpId="notes"/>
<extension from="12" helpId="notes"/>
<extension from="13" helpId="no_help"/>
<extension from="14" helpId="notes"/>

  • from
    An integer or a name like "dimension0".

  • helpId
    An identifier.

The graph Element

graph
   :cellStyle=ref style
   :style=ref style
=> location+ coordinates faceting facetLayout interval

coordinates => EMPTY

graph has the following attributes.

  • cellStyle
    style
    Each of these is the id of a style element. The former is the default style for individual cells, the latter for the entire table.

The location Element

location
   :part=(height | width | top | bottom | left | right)
   :method=(sizeToContent | attach | fixed | same)
   :min=dimension?
   :max=dimension?
   :target=ref (labelFrame | graph | container)?
   :value?
=> EMPTY

Each instance of this element specifies where some part of the table frame is located. All the examples in the corpus have four instances of this element, one for each of the parts height, width, left, and top. Some examples in the corpus add a fifth for part bottom, even though it is not clear how all of top, bottom, and height can be honored at the same time. In any case, location seems to have little importance in representing tables; a reader can safely ignore it.

  • part
    The part of the table being located.

  • method
    How the location is determined:

    • sizeToContent
      Based on the natural size of the table. Observed only for parts height and width.

    • attach
      Based on the location specified in target. Observed only for parts top and bottom.

    • fixed
      Using the value in value. Observed only for parts top, bottom, and left.

    • same
      Same as the specified target. Observed only for part left.

  • min
    Minimum size. Only observed with value 100pt. Only observed for part width.

  • target
    Required when method is attach or same, not observed otherwise. This identifies an element to attach to. Observed with the ID of title, footnote, graph, and other elements.

  • value
    Required when method is fixed, not observed otherwise. Observed values are 0%, 0px, 1px, and 3px on parts top and left, and 100% on part bottom.

The faceting Element

faceting => layer[layers1]* cross layer[layers2]*

cross => (unity | nest) (unity | nest)

unity => EMPTY

nest => variableReference[vars]+

variableReference :ref=ref (sourceVariable | derivedVariable) => EMPTY

layer
   :variable=ref (sourceVariable | derivedVariable)
   :value
   :visible=bool?
   :method[layer_method]=(nest)?
   :titleVisible=bool?
=> EMPTY

The faceting element describes the row, column, and layer structure of the table. Its cross child determines the row and column structure, and each layer child (if any) represents a layer. Layers may appear before or after cross.

The cross element describes the row and column structure of the table. It has exactly two children, the first of which describes the table's columns and the second the table's rows. Each child is a nest element if the table has any dimensions along the axis in question, otherwise a unity element.

A nest element contains of one or more dimensions listed from innermost to outermost, each represented by variableReference child elements. Each variable in a dimension is listed in order. See Variable Elements, for information on the variables that comprise a dimension.

A nest can contain a single dimension, e.g.:

<nest>
  <variableReference ref="dimension0categories"/>
  <variableReference ref="dimension0group0"/>
  <variableReference ref="dimension0"/>
</nest>

A nest can contain multiple dimensions, e.g.:

<nest>
  <variableReference ref="dimension1categories"/>
  <variableReference ref="dimension1group0"/>
  <variableReference ref="dimension1"/>
  <variableReference ref="dimension0categories"/>
  <variableReference ref="dimension0"/>
</nest>

A nest may have no dimensions, in which case it still has one variableReference child, which references a derivedVariable whose value attribute is constant(0). In the corpus, such a derivedVariable has row or column, respectively, as its id. This is equivalent to using a unity element in place of nest.

A variableReference element refers to a variable through its ref attribute.

Each layer element represents a dimension, e.g.:

<layer value="0" variable="dimension0categories" visible="true"/>
<layer value="dimension0" variable="dimension0" visible="false"/>

layer has the following attributes.

  • variable
    Refers to a sourceVariable or derivedVariable element.

  • value
    The value to select. For a category variable, this is always 0; for a data variable, it is the same as the variable attribute.

  • visible
    Whether the layer is visible. Generally, category layers are visible and data layers are not, but sometimes this attribute is omitted.

  • method
    When present, this is always nest.

The facetLayout Element

facetLayout => tableLayout setCellProperties[scp1]*
               facetLevel+ setCellProperties[scp2]*

tableLayout
   :verticalTitlesInCorner=bool
   :style=ref style?
   :fitCells=(ticks both)?
=> EMPTY

The facetLayout element and its descendants control styling for the table.

Its tableLayout child has the following attributes

  • verticalTitlesInCorner
    If true, in the absence of corner text, row headings will be displayed in the corner.

  • style
    Refers to a style element.

  • fitCells
    Meaning unknown.

The facetLevel Element

facetLevel :level=int :gap=dimension? => axis

axis :style=ref style => label? majorTicks

majorTicks
   :labelAngle=int
   :length=dimension
   :style=ref style
   :tickFrameStyle=ref style
   :labelFrequency=int?
   :stagger=bool?
=> gridline?

gridline
   :style=ref style
   :zOrder=int
=> EMPTY

Each facetLevel describes a variableReference or layer, and a table has one facetLevel element for each such element. For example, an SPV detail member that contains four variableReference elements and two layer elements will contain six facetLevel elements.

In the corpus, facetLevel elements and the elements that they describe are always in the same order. The correspondence may also be observed in two other ways. First, one may use the level attribute, described below. Second, in the corpus, a facetLevel always has an id that is the same as the id of the element it describes with _facetLevel appended. One should not formally rely on this, of course, but it is usefully indicative.

  • level
    A 1-based index into the variableReference and layer elements, e.g. a facetLayout with a level of 1 describes the first variableReference in the SPV detail member, and in a member with four variableReference elements, a facetLayout with a level of 5 describes the first layer in the member.

  • gap
    Always observed as 0pt.

Each facetLevel contains an axis, which in turn may contain a label for the facetLevel and does contain a majorTicks element.

  • labelAngle
    Normally 0. The value -90 causes inner column or outer row labels to be rotated vertically.

  • style

  • tickFrameStyle
    Each refers to a style element. style is the style of the tick labels, tickFrameStyle the style for the frames around the labels.

The label Element

label
   :style=ref style
   :textFrameStyle=ref style?
   :purpose=(title | subTitle | subSubTitle | layer | footnote)?
=> text+ | descriptionGroup

descriptionGroup
   :target=ref faceting
   :separator?
=> (description | text)+

description :name=(variable | value) => EMPTY

text
   :usesReference=int?
   :definesReference=int?
   :position=(subscript | superscript)?
   :style=ref style
=> TEXT

This element represents a label on some aspect of the table.

  • style
    textFrameStyle
    Each of these refers to a style element. style is the style of the label text, textFrameStyle the style for the frame around the label.

  • purpose
    The kind of entity being labeled.

A descriptionGroup concatenates one or more elements to form a label. Each element can be a text element, which contains literal text, or a description element that substitutes a value or a variable name.

  • target
    The id of an element being described. In the corpus, this is always faceting.

  • separator
    A string to separate the description of multiple groups, if the target has more than one. In the corpus, this is always a new-line.

Typical contents for a descriptionGroup are a value by itself:

<description name="value"/>

or a variable and its value, separated by a colon:

<description name="variable"/><text>:</text><description name="value"/>

A description is like a macro that expands to some property of the target of its parent descriptionGroup. The name attribute specifies the property.

The setCellProperties Element

setCellProperties
   :applyToConverse=bool?
=> (setStyle | setFrameStyle | setFormat | setMetaData)* union[union_]?

The setCellProperties element sets style properties of cells or row or column labels.

Interpreting setCellProperties requires answering two questions: which cells or labels to style, and what styles to use.

Which Cells?

union => intersect+

intersect => where+ | intersectWhere | alternating | EMPTY

where
   :variable=ref (sourceVariable | derivedVariable)
   :include
=> EMPTY

intersectWhere
   :variable=ref (sourceVariable | derivedVariable)
   :variable2=ref (sourceVariable | derivedVariable)
=> EMPTY

alternating => EMPTY

When union is present with intersect children, each of those children specifies a group of cells that should be styled, and the total group is all those cells taken together. When union is absent, every cell is styled. One attribute on setCellProperties affects the choice of cells:

  • applyToConverse
    If true, this inverts the meaning of the cell selection: the selected cells are the ones not designated. This is confusing, given the additional restrictions of union, but in the corpus applyToConverse is never present along with union.

An intersect specifies restrictions on the cells to be matched. Each where child specifies which values of a given variable to include. The attributes of intersect are:

  • variable
    Refers to a variable, e.g. dimension0categories. Only "categories" variables make sense here, but other variables, e.g. dimension0group0map, are sometimes seen. The reader may ignore these.

  • include
    A value, or multiple values separated by semicolons, e.g. 0 or 13;14;15;16.

PSPP ignores setCellProperties when intersectWhere is present.

What Styles?

setStyle
   :target=ref (labeling | graph | interval | majorTicks)
   :style=ref style
=> EMPTY

setMetaData :target=ref graph :key :value => EMPTY

setFormat
   :target=ref (majorTicks | labeling)
   :reset=bool?
=> format | numberFormat | stringFormat+ | dateTimeFormat | elapsedTimeFormat

setFrameStyle
   :style=ref style
   :target=ref majorTicks
=> EMPTY

The set* children of setCellProperties determine the styles to set.

When setCellProperties contains a setFormat whose target references a labeling element, or if it contains a setStyle that references a labeling or interval element, the setCellProperties sets the style for table cells. The format from the setFormat, if present, replaces the cells' format. The style from the setStyle that references labeling, if present, replaces the label's font and cell styles, except that the background color is taken instead from the interval's style, if present.

When setCellProperties contains a setFormat whose target references a majorTicks element, or if it contains a setStyle whose target references a majorTicks, or if it contains a setFrameStyle element, the setCellProperties sets the style for row or column labels. In this case, the setCellProperties always contains a single where element whose variable designates the variable whose labels are to be styled. The format from the setFormat, if present, replaces the labels' format. The style from the setStyle that references majorTicks, if present, replaces the labels' font and cell styles, except that the background color is taken instead from the setFrameStyle's style, if present.

When setCellProperties contains a setStyle whose target references a graph element, and one that references a labeling element, and the union element contains alternating, the setCellProperties sets the alternate foreground and background colors for the data area. The foreground color is taken from the style referenced by the setStyle that targets the graph, the background color from the setStyle for labeling.

A reader may ignore a setCellProperties that only contains setMetaData, as well as setMetaData within other setCellProperties.

A reader may ignore a setCellProperties whose only set* child is a setStyle that targets the graph element.

The setStyle Element

setStyle
   :target=ref (labeling | graph | interval | majorTicks)
   :style=ref style
=> EMPTY

This element associates a style with the target.

  • target
    The id of an element whose style is to be set.

  • style
    The id of a style element that identifies the style to set on the target.

The setFormat Element

setFormat
   :target=ref (majorTicks | labeling)
   :reset=bool?
=> format | numberFormat | stringFormat+ | dateTimeFormat | elapsedTimeFormat

This element sets the format of the target, "format" in this case meaning the SPSS print format for a variable.

The details of this element vary depending on the schema version, as declared in the root visualization element's version attribute. A reader can interpret the content without knowing the schema version.

The setFormat element itself has the following attributes.

  • target
    Refers to an element whose style is to be set.

  • reset
    If this is true, this format replaces the target's previous format. If it is false, the modifies the previous format.

The numberFormat Element

numberFormat
   :minimumIntegerDigits=int?
   :maximumFractionDigits=int?
   :minimumFractionDigits=int?
   :useGrouping=bool?
   :scientific=(onlyForSmall | whenNeeded | true | false)?
   :small=real?
   :prefix?
   :suffix?
=> affix*

Specifies a format for displaying a number. The available options are a superset of those available from PSPP print formats. PSPP chooses a print format type for a numberFormat as follows:

  1. If scientific is true, uses E format.

  2. If prefix is $, uses DOLLAR format.

  3. If suffix is %, uses PCT format.

  4. If useGrouping is true, uses COMMA format.

  5. Otherwise, uses F format.

For translating to a print format, PSPP uses maximumFractionDigits as the number of decimals, unless that attribute is missing or out of the range [0,15], in which case it uses 2 decimals.

  • minimumIntegerDigits
    Minimum number of digits to display before the decimal point. Always observed as 0.

  • maximumFractionDigits
    minimumFractionDigits
    Maximum or minimum, respectively, number of digits to display after the decimal point. The observed values of each attribute range from 0 to 9.

  • useGrouping
    Whether to use the grouping character to group digits in large numbers.

  • scientific
    This attribute controls when and whether the number is formatted in scientific notation. It takes the following values:

    • onlyForSmall
      Use scientific notation only when the number's magnitude is smaller than the value of the small attribute.

    • whenNeeded
      Use scientific notation when the number will not otherwise fit in the available space.

    • true
      Always use scientific notation. Not observed in the corpus.

    • false
      Never use scientific notation. A number that won't otherwise fit will be replaced by an error indication (see the errorCharacter attribute). Not observed in the corpus.

  • small
    Only present when the scientific attribute is onlyForSmall, this is a numeric magnitude below which the number will be formatted in scientific notation. The values 0 and 0.0001 have been observed. The value 0 seems like a pathological choice, since no real number has a magnitude less than 0; perhaps in practice such a choice is equivalent to setting scientific to false.

  • prefix
    suffix
    Specifies a prefix or a suffix to apply to the formatted number. Only suffix has been observed, with value %.

The stringFormat Element

stringFormat => relabel* affix*

relabel :from=real :to => EMPTY

The stringFormat element specifies how to display a string. By default, a string is displayed verbatim, but relabel can change it.

The relabel element appears as a child of stringFormat (and of format, when it is used to format strings). It specifies how to display a given value. It is used to implement value labels and to display the system-missing value in a human-readable way. It has the following attributes:

  • from
    The value to map. In the corpus this is an integer or the system-missing value -1.797693134862316E300.

  • to
    The string to display in place of the value of from. In the corpus this is a wide variety of value labels; the system-missing value is mapped to ..

The dateTimeFormat Element

dateTimeFormat
   :baseFormat[dt_base_format]=(date | time | dateTime)
   :separatorChars?
   :mdyOrder=(dayMonthYear | monthDayYear | yearMonthDay)?
   :showYear=bool?
   :yearAbbreviation=bool?
   :showQuarter=bool?
   :quarterPrefix?
   :quarterSuffix?
   :showMonth=bool?
   :monthFormat=(long | short | number | paddedNumber)?
   :showWeek=bool?
   :weekPadding=bool?
   :weekSuffix?
   :showDayOfWeek=bool?
   :dayOfWeekAbbreviation=bool?
   :dayPadding=bool?
   :dayOfMonthPadding=bool?
   :hourPadding=bool?
   :minutePadding=bool?
   :secondPadding=bool?
   :showDay=bool?
   :showHour=bool?
   :showMinute=bool?
   :showSecond=bool?
   :showMillis=bool?
   :dayType=(month | year)?
   :hourFormat=(AMPM | AS_24 | AS_12)?
=> affix*

This element appears only in schema version 2.5 and earlier.

Data to be formatted in date formats is stored as strings in legacy data, in the format yyyy-mm-ddTHH:MM:SS.SSS and must be parsed and reformatted by the reader.

The following attribute is required.

  • baseFormat
    Specifies whether a date and time are both to be displayed, or just one of them.

Many of the attributes' meanings are obvious. The following seem to be worth documenting.

  • separatorChars
    Exactly four characters. In order, these are used for: decimal point, grouping, date separator, time separator. Always .,-:.

  • mdyOrder
    Within a date, the order of the days, months, and years. dayMonthYear is the only observed value, but one would expect that monthDayYear and yearMonthDay to be reasonable as well.

  • showYear

  • yearAbbreviation
    Whether to include the year and, if so, whether the year should be shown abbreviated, that is, with only 2 digits. Each is true or false; only values of true and false, respectively, have been observed.

  • showMonth

  • monthFormat
    Whether to include the month (true or false) and, if so, how to format it. monthFormat is one of the following:

    • long
      The full name of the month, e.g. in an English locale, September.

    • short
      The abbreviated name of the month, e.g. in an English locale, Sep.

    • number
      The number representing the month, e.g. 9 for September.

    • paddedNumber
      A two-digit number representing the month, e.g. 09 for September.

    Only values of true and short, respectively, have been observed.

  • dayType
    This attribute is always month in the corpus, specifying that the day of the month is to be displayed; a value of year is supposed to indicate that the day of the year, where 1 is January 1, is to be displayed instead.

  • hourFormat
    hourFormat, if present, is one of:

    • AMPM
      The time is displayed with an am or pm suffix, e.g. 10:15pm.

    • AS_24
      The time is displayed in a 24-hour format, e.g. 22:15.

      This is the only value observed in the corpus.

    • AS_12
      The time is displayed in a 12-hour format, without distinguishing morning or evening, e.g. 10;15.

    hourFormat is sometimes present for elapsedTime formats, which is confusing since a time duration does not have a concept of AM or PM. This might indicate a bug in the code that generated the XML in the corpus, or it might indicate that elapsedTime is sometimes used to format a time of day.

For a baseFormat of date, PSPP chooses a print format type based on the following rules:

  1. If showQuarter is true: QYR.

  2. Otherwise, if showWeek is true: WKYR.

  3. Otherwise, if mdyOrder is dayMonthYear:

    a. If monthFormat is number or paddedNumber: EDATE.

    b. Otherwise: DATE.

  4. Otherwise, if mdyOrder is yearMonthDay: SDATE.

  5. Otherwise, ADATE.

For a baseFormat of dateTime, PSPP uses YMDHMS if mdyOrder is yearMonthDay and DATETIME otherwise. For a baseFormat of time, PSPP uses DTIME if showDay is true, otherwise TIME if showHour is true, otherwise MTIME.

For a baseFormat of date, the chosen width is the minimum for the format type, adding 2 if yearAbbreviation is false or omitted. For other base formats, the chosen width is the minimum for its type, plus 3 if showSecond is true, plus 4 more if showMillis is also true. Decimals are 0 by default, or 3 if showMillis is true.

The elapsedTimeFormat Element

elapsedTimeFormat
   :baseFormat[dt_base_format]=(date | time | dateTime)
   :dayPadding=bool?
   :hourPadding=bool?
   :minutePadding=bool?
   :secondPadding=bool?
   :showYear=bool?
   :showDay=bool?
   :showHour=bool?
   :showMinute=bool?
   :showSecond=bool?
   :showMillis=bool?
=> affix*

This element specifies the way to display a time duration.

Data to be formatted in elapsed time formats is stored as strings in legacy data, in the format H:MM:SS.SSS, with additional hour digits as needed for long durations, and must be parsed and reformatted by the reader.

The following attribute is required.

  • baseFormat
    Specifies whether a day and a time are both to be displayed, or just one of them.

The remaining attributes specify exactly how to display the elapsed time.

For baseFormat of time, PSPP converts this element to print format type DTIME; otherwise, if showHour is true, to TIME; otherwise, to MTIME. The chosen width is the minimum for the chosen type, adding 3 if showSecond is true, adding 4 more if showMillis is also true. Decimals are 0 by default, or 3 if showMillis is true.

The format Element

format
   :baseFormat[f_base_format]=(date | time | dateTime | elapsedTime)?
   :errorCharacter?
   :separatorChars?
   :mdyOrder=(dayMonthYear | monthDayYear | yearMonthDay)?
   :showYear=bool?
   :showQuarter=bool?
   :quarterPrefix?
   :quarterSuffix?
   :yearAbbreviation=bool?
   :showMonth=bool?
   :monthFormat=(long | short | number | paddedNumber)?
   :dayPadding=bool?
   :dayOfMonthPadding=bool?
   :showWeek=bool?
   :weekPadding=bool?
   :weekSuffix?
   :showDayOfWeek=bool?
   :dayOfWeekAbbreviation=bool?
   :hourPadding=bool?
   :minutePadding=bool?
   :secondPadding=bool?
   :showDay=bool?
   :showHour=bool?
   :showMinute=bool?
   :showSecond=bool?
   :showMillis=bool?
   :dayType=(month | year)?
   :hourFormat=(AMPM | AS_24 | AS_12)?
   :minimumIntegerDigits=int?
   :maximumFractionDigits=int?
   :minimumFractionDigits=int?
   :useGrouping=bool?
   :scientific=(onlyForSmall | whenNeeded | true | false)?
   :small=real?
   :prefix?
   :suffix?
   :tryStringsAsNumbers=bool?
   :negativesOutside=bool?
=> relabel* affix*

This element is the union of all of the more-specific format elements. It is interpreted in the same way as one of those format elements, using baseFormat to determine which kind of format to use.

There are a few attributes not present in the more specific formats:

  • tryStringsAsNumbers
    When this is true, it is supposed to indicate that string values should be parsed as numbers and then displayed according to numeric formatting rules. However, in the corpus it is always false.

  • negativesOutside
    If true, the negative sign should be shown before the prefix; if false, it should be shown after.

The affix Element

affix
   :definesReference=int
   :position=(subscript | superscript)
   :suffix=bool
   :value
=> EMPTY

This defines a suffix (or, theoretically, a prefix) for a formatted value. It is used to insert a reference to a footnote. It has the following attributes:

  • definesReference
    This specifies the footnote number as a natural number: 1 for the first footnote, 2 for the second, and so on.

  • position
    Position for the footnote label. Always superscript.

  • suffix
    Whether the affix is a suffix (true) or a prefix (false). Always true.

  • value
    The text of the suffix or prefix. Typically a letter, e.g. a for footnote 1, b for footnote 2, ... The corpus contains other values: *, **, and a few that begin with at least one comma: ,b, ,c, ,,b, and ,,c.

The interval Element

interval :style=ref style => labeling footnotes?

labeling
   :style=ref style?
   :variable=ref (sourceVariable | derivedVariable)
=> (formatting | format | footnotes)*

formatting :variable=ref (sourceVariable | derivedVariable) => formatMapping*

formatMapping :from=int => format?

footnotes
   :superscript=bool?
   :variable=ref (sourceVariable | derivedVariable)
=> footnoteMapping*

footnoteMapping :definesReference=int :from=int :to => EMPTY

The interval element and its descendants determine the basic formatting and labeling for the table's cells. These basic styles are overridden by more specific styles set using setCellProperties.

The style attribute of interval itself may be ignored.

The labeling element may have a single formatting child. If present, its variable attribute refers to a variable whose values are format specifiers as numbers, e.g. value 0x050802 for F8.2. However, the numbers are not actually interpreted that way. Instead, each number actually present in the variable's data is mapped by a formatMapping child of formatting to a format that specifies how to display it.

The labeling element may also have a footnotes child element. The variable attribute of this element refers to a variable whose values are comma-delimited strings that list the 1-based indexes of footnote references. (Cells without any footnote references are numeric 0 instead of strings.)

Each footnoteMapping child of the footnotes element defines the footnote marker to be its to attribute text for the footnote whose 1-based index is given in its definesReference attribute.

The style Element

style
   :color=color?
   :color2=color?
   :labelAngle=real?
   :border-bottom=(solid | thick | thin | double | none)?
   :border-top=(solid | thick | thin | double | none)?
   :border-left=(solid | thick | thin | double | none)?
   :border-right=(solid | thick | thin | double | none)?
   :border-bottom-color?
   :border-top-color?
   :border-left-color?
   :border-right-color?
   :font-family?
   :font-size?
   :font-weight=(regular | bold)?
   :font-style=(regular | italic)?
   :font-underline=(none | underline)?
   :margin-bottom=dimension?
   :margin-left=dimension?
   :margin-right=dimension?
   :margin-top=dimension?
   :textAlignment=(left | right | center | decimal | mixed)?
   :labelLocationHorizontal=(positive | negative | center)?
   :labelLocationVertical=(positive | negative | center)?
   :decimal-offset=dimension?
   :size?
   :width?
   :visible=bool?
=> EMPTY

A style element has an effect only when it is referenced by another element to set some aspect of the table's style. Most of the attributes are self-explanatory. The rest are described below.

  • color
    In some cases, the text color; in others, the background color.

  • color2
    Not used.

  • labelAngle
    Normally 0. The value -90 causes inner column or outer row labels to be rotated vertically.

  • labelLocationHorizontal
    Not used.

  • labelLocationVertical
    The value positive corresponds to vertically aligning text to the top of a cell, negative to the bottom, center to the middle.

The labelFrame Element

labelFrame :style=ref style => location+ label? paragraph?

paragraph :hangingIndent=dimension? => EMPTY

A labelFrame element specifies content and style for some aspect of a table. Only labelFrame elements that have a label child are important. The purpose attribute in the label determines what the labelFrame affects:

  • title
    The table's title and its style.

  • subTitle
    The table's caption and its style.

  • footnote
    The table's footnotes and the style for the footer area.

  • layer
    The style for the layer area.

  • subSubTitle
    Ignored.

The style attribute references the style to use for the area.

The label, if present, specifies the text to put into the title or caption or footnotes. For footnotes, the label has two text children for every footnote, each of which has a usesReference attribute identifying the 1-based index of a footnote. The first, third, fifth, ... text child specifies the content for a footnote; the second, fourth, sixth, ... child specifies the marker. Content tends to end in a new-line, which the reader may wish to trim; similarly, markers tend to end in ..

The paragraph, if present, may be ignored, since it is always empty.

Legacy Properties

The detail XML format has features for styling most of the aspects of a table. It also inherits defaults for many aspects from structure XML, which has the following tableProperties element:

tableProperties
   :name?
=> generalProperties footnoteProperties cellFormatProperties borderProperties printingProperties

generalProperties
   :hideEmptyRows=bool?
   :maximumColumnWidth=dimension?
   :maximumRowWidth=dimension?
   :minimumColumnWidth=dimension?
   :minimumRowWidth=dimension?
   :rowDimensionLabels=(inCorner | nested)?
=> EMPTY

footnoteProperties
   :markerPosition=(superscript | subscript)?
   :numberFormat=(alphabetic | numeric)?
=> EMPTY

cellFormatProperties => cell_style+

any[cell_style]
   :alternatingColor=color?
   :alternatingTextColor=color?
=> style

style
   :color=color?
   :color2=color?
   :font-family?
   :font-size?
   :font-style=(regular | italic)?
   :font-weight=(regular | bold)?
   :font-underline=(none | underline)?
   :labelLocationVertical=(positive | negative | center)?
   :margin-bottom=dimension?
   :margin-left=dimension?
   :margin-right=dimension?
   :margin-top=dimension?
   :textAlignment=(left | right | center | decimal | mixed)?
   :decimal-offset=dimension?
=> EMPTY

borderProperties => border_style+

any[border_style]
   :borderStyleType=(none | solid | dashed | thick | thin | double)?
   :color=color?
=> EMPTY

printingProperties
   :printAllLayers=bool?
   :rescaleLongTableToFitPage=bool?
   :rescaleWideTableToFitPage=bool?
   :windowOrphanLines=int?
   :continuationText?
   :continuationTextAtBottom=bool?
   :continuationTextAtTop=bool?
   :printEachLayerOnSeparatePage=bool?
=> EMPTY

The name attribute appears only in standalone .stt files.

SPSS TableLook File Formats

SPSS has a concept called a TableLook to control the styling of pivot tables in output. SPSS 15 and earlier used .tlo files with a special binary format to save TableLooks to disk; SPSS 16 and later use .stt files in an XML format to save them. Both formats expose roughly the same features, although the older .tlo format does have some features that .stt does not.

This chapter describes both formats.

The .stt Format

The .stt file format is an XML file that contains a subset of the SPV structure member format. Its root element is a tableProperties element.

The .tlo Format

A .tlo file has a custom binary format. This section describes it using the binary format conventions used for SPV binary members. There is one new convention: TLO files express colors as int32 values in which the low 8 bits are the red component, the next 8 bits are green, and next 8 bits are blue, and the high bits are zeros.

TLO files support various features that SPV files do not. PSPP implements the SPV feature set, so it mostly ignores the added TLO features. The details of this mapping are explained below.

At the top level, a TLO file consists of five sections. The first four are always present and the last one is optional:

TableLook =>
   PTTableLook[tl]
   PVSeparatorStyle[ss]
   PVCellStyle[cs]
   PVTextStyle[ts]
   V2Styles?

Each section is described below.

PTTableLook

PTTableLook =>
   ff ff 00 00 "PTTableLook" (00|02)[version]
   int16[flags]
   00 00
   bool[nested-row-labels] 00
   bool[footnote-marker-subscripts] 00
   i54 i18

In PTTableLook, version is 00 or 02. The only difference is that version 00 lacks V2Styles and that version 02 includes it. Both TLO versions are seen in the wild.

flags is a bit-mapped field. Its bits have the following meanings:

  • 0x2: If set to 1, hide empty rows and columns; otherwise, show them.

  • 0x4: If set to 1, use numeric footnote markers; otherwise, use alphabetic footnote markers.

  • 0x8: If set to 1, print all layers; otherwise, print only the current layer.

  • 0x10: If set to 1, scale the table to fit the page width; otherwise, break it horizontally if necessary.

  • 0x20: If set to 1, scale the table to fit the page length; otherwise, break it vertically if necessary.

  • 0x40: If set to 1, print each layer on a separate page (only if all layers are being printed); otherwise, paginate layers naturally.

  • 0x80: If set to 1, print a continuation string at the top of a table that is split between pages.

  • 0x100: If set to 1, print a continuation string at the bottom of a table that is split between pages.

When nested-row-labels is 1, row dimension labels appear nested; otherwise, they are put into the upper-left corner of the pivot table.

When footnote-marker-subscripts is 1, footnote markers are shown as subscripts; otherwise, they are shown as superscripts.

PVSeparatorStyle

PVSeparatorStyle =>
   ff ff 00 00 "PVSeparatorStyle" 00
   Separator*4[sep1]
   03 80 00
   Separator*4[sep2]

Separator =>
   case(
       00 00
     | 01 00 int32[color] int16[style] int16[width]
   )[type]

PVSeparatorStyle contains eight Separators, in two groups. Each Separator represents a border between pivot table elements. TLO and SPV files have the same concepts for borders. See Light Member Borders, for the treatment of borders in SPV files.

A Separator's type is 00 if the border is not drawn, 01 otherwise. For a border that is drawn, color is the color that it is drawn in. style and width have the following meanings:

  • style = 0 and 0 ≤ width ≤ 3
    An increasingly thick single line. SPV files only have three line thicknesses. PSPP treats width 0 as a thin line, width 1 as a solid (normal width) line, and width 2 or 3 as a thick line.

  • style = 1 and 0 ≤ width ≤ 1
    A doubled line, composed of normal-width (0) or thick (1) lines. SPV files only have "normal" width double lines, so PSPP maps both variants the same way.

  • style = 2
    A dashed line.

The first group, sep1, represents the following borders within the pivot table, by index:

  1. Horizontal dimension rows
  2. Vertical dimension rows
  3. Horizontal category rows
  4. Vertical category rows

The second group, sep2, represents the following borders within the pivot table, by index:

  1. Horizontal dimension columns
  2. Vertical dimension columns
  3. Horizontal category columns
  4. Vertical category columns

PVCellStyle and PVTextStyle

PVCellStyle =>
   ff ff 00 00 "PVCellStyle"
   AreaColor[title-color]

PVTextStyle =>
   ff ff 00 00 "PVTextStyle" 00
   AreaStyle[title-style] MostAreas*7[most-areas]

MostAreas =>
   06 80
   AreaColor[color] 08 80 00 AreaStyle[style]

These sections hold the styling and coloring for each of the 8 areas in a pivot table. They are conceptually similar to the Areas style information in SPV light members.

The styling and coloring for the title area is split between PVCellStyle and PVTextStyle: the former holds title-color, the latter holds title-style. The style for the remaining 7 areas is in most-areas in PVTextStyle, in the following order: layers, corner, row labels, column labels, data, caption, and footer.

AreaColor =>
   00 01 00 int32[color10] int32[color0] byte[shading] 00

AreaColor represents the background color of an area. TLO files, but not SPV files, describe backgrounds that are a shaded combination of two colors: shading of 0 is pure color0, shading of 10 is pure color10, and value in between mix pixels of the two different colors in linear degree. PSPP does not implement shading, so for 1 ≤ shading ≤ 9 it interpolates RGB values between colors to arrive at an intermediate shade.

AreaStyle =>
   int16[valign] int16[halign] int16[decimal-offset]
   int16[left-margin] int16[right-margin] int16[top-margin] int16[bottom-margin]
   00 00 01 00
   int32[font-size] int16[stretch]
   00*2
   int32[rotation-angle]
   00*4
   int16[weight]
   00*2
   bool[italic] bool[underline] bool[strikethrough]
   int32[rtf-charset-number]
   byte[x]
   byte[font-name-len] byte*[font-name-len][font-name]
   int32[text-color]
   00*2

AreaStyle represents style properties of an area.

valign has the following values:

valignVertical Alignment
0Top
1Bottom
2Center

halign has the following values:

halignHorizontal Alignment
0Left
1Right
2Center
3Mixed
4Decimal

For decimal alignment, decimal-offset is the offset of the decimal point, in 20ths of a point.

left-margin, right-margin, top-margin, and bottom-margin are also measured in 20ths of a point.

font-size is negative 96ths of an inch, e.g. 9 point is -12 or 0xfffffff3.

stretch has something to do with font size or stretch. The usual value is 01 and values larger than that do weird things. A reader can safely ignore it.

rotation-angle is a font rotation angle. A reader can safely ignore it.

weight is 400 for a normal-weight font, 700 indicates bold. (This is a Windows API convention.)

italic and underline have the obvious meanings. So does strikethrough, which PSPP ignores.

rtf-charset-number is a character set number from RTF. A reader can safely ignore it.

The meaning of x is unknown. Values 12, 22, 31, and 32 have been observed.

The font-name is the name of a font, such as Arial. Only US-ASCII characters have been observed here.

text-color is the color of the text itself.

V2Styles

V2Styles =>
   Separator*11[sep3]
   byte[continuation-len] byte*[continuation-len][continuation]
   int32[min-col-width] int32[max-col-width]
   int32[min-row-height] int32[max-row-height]

This final, optional, part of the TLO file format contains some additional style information. It begins with sep3, which represents the following borders within the pivot table, by index:

  • 0: Title.
  • 1...4: Left, right, top, and bottom inner frame.
  • 5...8: Left, right, top, and bottom outer frame.
  • 9, 10: Left and top of data area.

When V2Styles is absent, the inner frame borders default to a solid line and the others listed above to no line.

continuation is the string that goes at the top or bottom of a table broken across pages. When V2Styles is absent, the default is (Cont.).

min-col-width is the minimum width that a column will be assigned automatically. max-col-width is the maximum width that a column will be assigned to accommodate a long column label. min-row-width and max-row-width are a similar range for the width of row labels. All of these measurements are in points. When V2Styles is absent, the defaults are 36 for min-col-width and min-row-height, 72 for max-col-width, and 120 for max-row-height.

Encrypted File Wrappers

SPSS 21 and later can package multiple kinds of files inside an encrypted wrapper. The wrapper has a common format, regardless of the kind of the file that it contains.

⚠️ Warning: The SPSS encryption wrapper is poorly designed. When the password is unknown, it is much cheaper and faster to decrypt a file encrypted this way than if a well designed alternative were used. If you must use this format, use a 10-byte randomly generated password.

Common Wrapper Format

An encrypted file wrapper begins with the following 36-byte header, where xxx identifies the type of file encapsulated: SAV for a system file, SPS for a syntax file, SPV for a viewer file. PSPP code for identifying these files just checks for the ENCRYPTED keyword at offset 8, but the other bytes are also fixed in practice:

0000  1c 00 00 00 00 00 00 00  45 4e 43 52 59 50 54 45  |........ENCRYPTE|
0010  44 xx xx xx 15 00 00 00  00 00 00 00 00 00 00 00  |Dxxx............|
0020  00 00 00 00                                       |....|

Following the fixed header is essentially the regular contents of the encapsulated file in its usual format, with each 16-byte block encrypted with AES-256 in ECB mode.

To make the plaintext an even multiple of 16 bytes in length, the encryption process appends PKCS #7 padding, as specified in RFC 5652 section 6.3. Padding appends 1 to 16 bytes to the plaintext, in which each byte of padding is the number of padding bytes added. If the plaintext is, for example, 2 bytes short of a multiple of 16, the padding is 2 bytes with value 02; if the plaintext is a multiple of 16 bytes in length, the padding is 16 bytes with value 0x10.

The AES-256 key is derived from a password in the following way:

  1. Start from the literal password typed by the user. Truncate it to at most 10 bytes, then append as many null bytes as necessary until there are exactly 32 bytes. Call this password.

  2. Let constant be the following 73-byte constant:

    0000  00 00 00 01 35 27 13 cc  53 a7 78 89 87 53 22 11
0010  d6 5b 31 58 dc fe 2e 7e  94 da 2f 00 cc 15 71 80
0020  0a 6c 63 53 00 38 c3 38  ac 22 f3 63 62 0e ce 85
0030  3f b8 07 4c 4e 2b 77 c7  21 f5 1a 80 1d 67 fb e1
0040  e1 83 07 d8 0d 00 00 01  00

  3. Compute CMAC-AES-256(password, constant). Call the 16-byte result cmac.

  4. The 32-byte AES-256 key is cmac || cmac, that is, cmac repeated twice.

Example

Consider the password pspp. password is:

0000  70 73 70 70 00 00 00 00  00 00 00 00 00 00 00 00  |pspp............|
0010  00 00 00 00 00 00 00 00  00 00 00 00 00 00 00 00  |................|

cmac is:

0000  3e da 09 8e 66 04 d4 fd  f9 63 0c 2c a8 6f b0 45

The AES-256 key is:

0000  3e da 09 8e 66 04 d4 fd  f9 63 0c 2c a8 6f b0 45
0010  3e da 09 8e 66 04 d4 fd  f9 63 0c 2c a8 6f b0 45

Checking Passwords

A program reading an encrypted file may wish to verify that the password it was given is the correct one. One way is to verify that the PKCS #7 padding at the end of the file is well formed. However, any plaintext that ends in byte 01 is well formed PKCS #7, meaning that about 1 in 256 keys will falsely pass this test. This might be acceptable for interactive use, but the false positive rate is too high for a brute-force search of the password space.

A better test requires some knowledge of the file format being wrapped, to obtain a "magic number" for the beginning of the file.

  • The plaintext of system files begins with $FL2@(#) or $FL3@(#).

  • Before encryption, a syntax file is prefixed with a line at the beginning of the form * Encoding: ENCODING., where ENCODING is the encoding used for the rest of the file, e.g. windows-1252. Thus, * Encoding may be used as a magic number for system files.

  • The plaintext of viewer files begins with 50 4b 03 04 14 00 08 (50 4b is PK).

Password Encoding

SPSS also supports what it calls "encrypted passwords."

⚠️ Warning: SPSS "encrypted passwords" are not encrypted. They are encoded with a simple, fixed scheme and can be decoded to the original password using the rules described below.

An encoded password is always a multiple of 2 characters long, and never longer than 20 characters. The characters in an encoded password are always in the graphic ASCII range 33 through 126. Each successive pair of characters in the password encodes a single byte in the plaintext password.

Use the following algorithm to decode a pair of characters:

  1. Let a be the ASCII code of the first character, and b be the ASCII code of the second character.

  2. Let ah be the most significant 4 bits of a. Find the line in the table below that has ah on the left side. The right side of the line is a set of possible values for the most significant 4 bits of the decoded byte.

    2  ⇒ 2367
3  ⇒ 0145
47 ⇒ 89cd
56 ⇒ abef

  3. Let bh be the most significant 4 bits of b. Find the line in the second table below that has bh on the left side. The right side of the line is a set of possible values for the most significant 4 bits of the decoded byte. Together with the results of the previous step, only a single possibility is left.

    2  ⇒ 139b
3  ⇒ 028a
47 ⇒ 46ce
56 ⇒ 57df

  4. Let al be the least significant 4 bits of a. Find the line in the table below that has al on the left side. The right side of the line is a set of possible values for the least significant 4 bits of the decoded byte.

    03cf ⇒ 0145
12de ⇒ 2367
478b ⇒ 89cd
569a ⇒ abef

  5. Let bl be the least significant 4 bits of b. Find the line in the table below that has bl on the left side. The right side of the line is a set of possible values for the least significant 4 bits of the decoded byte. Together with the results of the previous step, only a single possibility is left.

    03cf ⇒ 028a
12de ⇒ 139b
478b ⇒ 46ce
569a ⇒ 57df

Example

Consider the encoded character pair -|. a is 0x2d and b is 0x7c, so ah is 2, bh is 7, al is 0xd, and bl is 0xc. ah means that the most significant four bits of the decoded character is 2, 3, 6, or 7, and bh means that they are 4, 6, 0xc, or 0xe. The single possibility in common is 6, so the most significant four bits are 6. Similarly, al means that the least significant four bits are 2, 3, 6, or 7, and bl means they are 0, 2, 8, or 0xa, so the least significant four bits are 2. The decoded character is therefore 0x62, the letter b.

Portable File Format

These days, most computers use the same internal data formats for integer and floating-point data, if one ignores little differences like big- versus little-endian byte ordering. This has not always been true, particularly in the 1960s or 1970s, when the portable file format originated as a way to exchange data between systems with incompatible data formats.

At the time, even bytes being 8 bits each was not a given. For that reason, the portable file format is a text format, because text files could be exchanged portably among systems slightly more freely. On the other hand, character encoding was not standardized, so exchanging data in portable file format required recoding it from the origin system's character encoding to the destination's.

Some contemporary systems represented text files as sequences of fixed-length (typically 80-byte) records, without new-line sequences. These operating systems padded lines shorter lines with spaces and truncated longer lines. To tolerate files copied from such systems, which might drop spaces at the ends of lines, the portable file format treats lines less than 80 bytes long as padded with spaces to that length.

The portable file format self-identifies the character encoding on the system that produced it at the very beginning, in the header. Since portable files are normally recoded when they are transported from one system to another, this identification can be wrong on its face: a file that was started in EBCDIC, and is then recoded to ASCII, will still say EBCDIC SPSS PORT FILE at the beginning, just in ASCII instead of EBCDIC.

The portable file header also contains a table of all of the characters that it supports. Readers use this to translate each byte of the file into its local encoding. Like the rest of the portable file, the character table is recoded when the file is moved to a system with a different character set so that it remains correct, or at least consistent with the rest of the file.

The portable file format is mostly obsolete. System files are a better alternative.

Sources

The information in this chapter is drawn from documentation and source code, including:

  • pff.tar.Z, a Fortran program from the 1980s that reads and writes portable files. This program contains translation tables from the portable character set to EBCDIC and to ASCII.

  • A document, now lost, that describes portable file syntax.

It is further informed by a corpus of about 1,400 portable files. The plausible creation dates in the corpus range from 1986 to 2025, in addition to 131 files with alleged creation dates between 1900 and 1907 and 21 files with an invalid creation date.

Portable File Characters

Portable files are arranged as a series of lines of 80 characters each. Each line is terminated by a carriage-return, line-feed sequence ("new-lines"). New-lines are only used to avoid line length limits imposed by some OSes; they are not meaningful.

Most lines in portable files are exactly 80 characters long. The only exception is a line that ends in one or more spaces, in which the spaces may optionally be omitted. Thus, a portable file reader must act as though a line shorter than 80 characters is padded to that length with spaces.

The file must be terminated with a Z character. In addition, if the final line in the file does not have exactly 80 characters, then it is padded on the right with Z characters. (The file contents may be in any character set; the file contains a description of its own character set, as explained in the next section. Therefore, the Z character is not necessarily an ASCII Z.)

For the rest of the description of the portable file format, new-lines and the trailing Zs will be ignored, as if they did not exist, because they are not an important part of understanding the file contents.

Portable File Structure

Every portable file consists of the following records, in sequence:

  • Splash strings.

  • Version and date info.

  • Product identification.

  • Author identification (optional).

  • Subproduct identification (optional).

  • Variable count.

  • Case weight variable (optional).

  • Variables. Each variable record may optionally be followed by a missing value record and a variable label record.

  • Value labels (optional).

  • Documents (optional).

  • Data.

Most records are identified by a single-character tag code. The file header and version info record do not have a tag.

Other than these single-character codes, there are three types of fields in a portable file: floating-point, integer, and string. Floating-point fields have the following format:

  • Zero or more leading spaces.

  • Optional asterisk (*), which indicates a missing value. The asterisk must be followed by a single character, generally a period (.), but it appears that other characters may also be possible. This completes the specification of a missing value.

  • Optional minus sign (-) to indicate a negative number.

  • A whole number, consisting of one or more base-30 digits: 0 through 9 plus capital letters A through T.

  • Optional fraction, consisting of a radix point (.) followed by one or more base-30 digits.

  • Optional exponent, consisting of a plus or minus sign (+ or -) followed by one or more base-30 digits.

  • A forward slash (/).

Integer fields take a form identical to floating-point fields, but they may not contain a fraction.

String fields take the form of a integer field having value N, followed by exactly N characters, which are the string content.

Strings longer than 255 bytes exist in the corpus.

Splash Strings

Every portable file begins with 200 bytes of splash strings that serve to identify the file's type and its original character set. The 200 bytes are divided into five 40-byte sections, each of which is supposed to represent the string <CHARSET> SPSS PORT FILE in a different character set encoding1, where <CHARSET> is the name of the character set used in the file, e.g. ASCII or EBCDIC. Each string is padded on the right with spaces in its respective character set.

It appears that these strings exist only to inform those who might view the file on a screen, letting them know what character set the file is in regardless of how they are viewing it, and that they are not parsed by SPSS products. Thus, they can be safely ignored. It is reasonable to simply write out ASCII SPSS PORT FILE five times, each time padded to 40 bytes.

Translation Table

The splash strings are followed by a 256-byte character set translation table. This segment describes a mapping from the character set used in the portable file to a "portable character set" that does not correspond to any known single-byte character set or code page. Each byte in the table reports the byte value that corresponds to the character represented by its position. The following section lists the character at each position.

For example, position 0x4a (decimal 74) in the portable character set is uppercase letter A (as shown in the table in the following section), so the 75th byte in the table is the value that represents A in the file.

Any real character set will not necessarily include all of the characters in the portable character set. In the translation table, omitted characters are written as digit 02.

For example, in practice, all of the control character positions are always written as 0.

The following section describes how the translation table is supposed to act based on looking at the sources, and then the section after that describes what it actually contains in practice.

Theory

The table below shows the portable character set. The columns in the table are:

  • "Pos", a position within the portable character set, in hex, from 00 to FF.

  • "EBCDIC", the translation for the given position to EBCDIC, as written in pff.tar.Z.

  • "ASCII", the translation for the given position to ASCII, as written in pff.tar.Z.

  • "Unicode", a suggestion for the best translation from this position to Unicode.

  • "Notes", which links to additional information for some characters.

In addition to the sources previously cited, some of the information below is drawn from RFC 183, from 1971. This RFC shows many of the "EBCDIC" hex codes in pff.tar.Z as corresponding to the descriptions in the document, even though no known EBCDIC codepage contains those characters with those codes.

PosEBCDICASCIIUnicodeNotes
0000———3
0101———3
0202———3
0303———3
0404———3
0505—U+0009 CHARACTER TABULATION—3
0606———3
0707———3
0808———3
0909———3
0A0A———3
0B0B—U+000B LINE TABULATION—3
0C0C—U+000C FORM FEED—3
0D0D—U+000D CARRIAGE RETURN—3
0E0E———3
0F0F———3
1010———3
1111———3
1212———3
1313———3
143C———3
1515—U+000A LINE FEED—3
1616—U+0008 BACKSPACE—3
1717———3
1818———3
1919———3
1A1A———3
1B1B———3
1C1C———3
1D1D———3
1E1E———3
1F2A———3
2020———3
2121———3
2222———3
2323———3
242B———3
2525—U+000A LINE FEED—3
2626———3
2727———3
281F———3
2924———3
2A14———3
2B2D———3
2C2E———3
2D2F—U+0007 BELL—3
2E32———3
2F33———3
3034———3
3135———3
3236———3
3337———3
3438———3
3539———3
363A———3
373B———3
383D———3
393F———3
3A28———3
3B29———3
3C2C———3
3D————4
3E————4
3F————4
40F030U+0030 DIGIT ZERO0
...
49F939U+0039 DIGIT NINE9
4AC141U+0041 LATIN CAPITAL LETTER AA
...
52C949U+0049 LATIN CAPITAL LETTER II
53D14AU+004A LATIN CAPITAL LETTER JJ
...
5BD952U+0052 LATIN CAPITAL LETTER RR
5CE253U+0053 LATIN CAPITAL LETTER SS
...
63E95AU+005A LATIN CAPITAL LETTER ZZ
648161U+0061 LATIN SMALL LETTER Aa
...
7D8969U+0069 LATIN SMALL LETTER Ii
64916AU+006A LATIN SMALL LETTER Jj
...
7D9972U+0072 LATIN SMALL LETTER Rr
64A273U+0073 LATIN SMALL LETTER Ss
...
7DA97AU+007A LATIN SMALL LETTER Zz
7E4020U+0020 SPACE
7F4B2EU+002E FULL STOP.
804C3CU+003C LESS-THAN SIGN<
814D28U+0028 LEFT PARENTHESIS(
824E2BU+002B PLUS SIGN+
8359—U+007C VERTICAL LINE|5
845026U+0026 AMPERSAND&
85AD5BU+005B LEFT SQUARE BRACKET[
86BD5DU+005D RIGHT SQUARE BRACKET]
875A21U+0021 EXCLAMATION MARK!
885B24U+0024 DOLLAR SIGN$
895C2AU+002A ASTERISK*
8A5D29U+0029 RIGHT PARENTHESIS)
8B5E3BU+003B SEMICOLON;
8C5F5EU+005E CIRCUMFLEX ACCENT^
8D602DU+002D HYPHEN-MINUS-
8E612FU+002F SOLIDUS/
8F6A76U+00A6 BROKEN BAR¦5
906B2CU+002C COMMA,
916C25U+0025 PERCENT SIGN%
926D5FU+005F LOW LINE_
936E3EU+003E GREATER-THAN SIGN>
946F3FU+003F QUESTION MARK?
957960U+0060 GRAVE ACCENT`
967A3AU+003A COLON:
977B23U+0023 NUMBER SIGN#
987C40U+0040 COMMERCIAL AT@
997D27U+0027 APOSTROPHE'
9A7E3DU+003D EQUALS SIGN=
9B7F22U+0022 QUOTATION MARK"
9C8C—U+2264 LESS-THAN OR EQUAL TO≤
9D9C—U+25A1 WHITE SQUARE□6
9E9E—U+00B1 PLUS-MINUS SIGN±
9F9F—U+25A0 BLACK SQUARE■7
A0——U+00B0 DEGREE SIGN°
A18F—U+2020 DAGGER†
A2A17EU+007E TILDE~
A3A0—U+2013 EN DASH–
A4AB—U+2514 BOX DRAWINGS LIGHT UP AND RIGHT└8
A5AC—U+250C BOX DRAWINGS LIGHT DOWN AND RIGHT┌8
A6AE—U+2265 GREATER-THAN OR EQUAL TO≥
A7B0—U+2070 SUPERSCRIPT ZERO⁰8
...
B0B9—U+2079 SUPERSCRIPT NINE⁹8
B1BB—U+2518 BOX DRAWINGS LIGHT UP AND LEFT┘8
B2BC—U+2510 BOX DRAWINGS LIGHT DOWN AND LEFT┐8
B3BE—U+2260 NOT EQUAL TO≠
B4BF—U+2014 EM DASH—
B58D—U+2070 SUPERSCRIPT LEFT PARENTHESIS⁽
B69D—U+207E SUPERSCRIPT RIGHT PARENTHESIS⁾
B7BE—U+207A SUPERSCRIPT PLUS SIGN⁺9
B8C07BU+007B LEFT CURLY BRACKET{
B9D07DU+007D RIGHT CURLY BRACKET}
BAE05CU+005C REVERSE SOLIDUS\
BB4A—0+00A2 CENT SIGN¢
BCAF—U+00B7 MIDDLE DOT·10
BD————4
...
FF————4

Summary:

RangeCharacters
40...4F0123456789ABCDEF
50...5FGHIJKLMNOPQRSTUV
60...6FWXYZabcdefghijkl
70...7Fmnopqrstuvwxyz .
80...8F<(+|&[]!$*);^-/¦
90...9F,%_>?`:#@'="≤□±■
A0...AF°†~–└┌≥⁰ⁱ⁲⁳⁴⁵⁶⁷⁸
B0...BC⁹┘┐≠—⁽⁾⁺{}\¢·

Practice: Character Set

The previous section described the translation table in theory. This section describes what it contains in the corpus.

Every file in the corpus is encoded in (extended) ASCII, although 31 of them indicate in their splash strings that they were recoded from EBCDIC. This also means that ASCII 0 indicates an unmapped character, that is, one not in the character set represented by the table.

The files are encoded in different ASCII extension. Some appear to be encoded in windows-1252, others in code page 437, others in unidentified character sets. The particular code page in use does not matter to a reader that uses the table for mapping.

  • There are some invariants across the translation tables for every file in the corpus:

    • All control codes (in the range 0 to 63) are unmapped.

      One consequence is that strings in the corpus can never contain new-lines. New-lines encoded literally would be problematic anyhow because readers must ignore them.

    • Digits 0 to 9 and letters A to Z and a to z are correctly mapped.

    • Punctuation for space as well as (+&$*);-/,%_?`:@'=\ are correctly mapped.

  • Characters <!^>\"~{} are mapped correctly in almost every file in the corpus, with a few outliers.

  • Characters [] are mostly correct with a few problems.

  • Position 97 is correctly # in most files, and wrongly $ in some.

  • The characters at positions 83 | and 8F ¦ have lots of issues, stemming from the history described on Wikipedia. In particular, EBCDIC and Unicode have separate characters for | and ¦, but ASCII does not.

    Most of the corpus leaves 83 | unmapped. Most of the rest map it correctly to |. The remainder map it to !.

    Most of the corpus maps 8F ¦ to |. Only a few map it correctly to ¦ in windows-1252 or (creatively) to ║ in code page 437.

  • Characters at the following positions are almost always wrong. The table shows:

    • "Character", the character and its position in the portable character set.

    • "Unmapped", the number of files in the corpus that leave the character unmapped (that is, set to 0).

    • "windows-1252", the number of files that map the character correctly in windows-1252. If there is more than one plausible mapping, or if the mapping doesn't exactly match the preferred Unicode, the entry shows the mapped character.

    • "cp437", the number of files that map the character correctly in code page 437.

      In a few cases, a plausible mapping in the "windows-1252" column is an ASCII character. Those aren't separately counted in the "cp437" column, even though ASCII maps the same way in both encodings.

    • "Wrong", the number of files that map the character to nothing that makes sense in a known encoding.

    CharacterUnmappedwindows-1252cp437Wrong
    9C ≤136601028
    A6 ≥137301021
    9F ■137301021
    9E ±1353151523
    A3 – (en dash)1302as -: 65as ─: 532
    B4 — (em dash)1308as -: 65as ─: 1021
    A4 └136701522
    A5 ┌136701522
    B1 ┘136701522
    B2 ┐136701522
    A8 ¹1286as ¹: 15; as 1: 65038
    A9 ²1286as ²: 15; as 2: 651523
    AA ³1286as ³: 15; as 3: 65038
    AB ⁴1308as 4: 65031
    ...............
    B0 ⁹1308as 9: 65031
    B3 ≠13730as ╪: 1021
    B6 ⁽13080096
    B7 ⁾13730031
    BB ¢1351161027
    BC ·1357as ·: 16; as ×: 1as ∙: 1020
    A0 °1382as °: 15; as º: 156

  • Characters at the following positions are always unmapped or wrong:

    CharacterUnmappedwindows-1252cp437Wrong
    9D □13730as ╬: 1021
    A1 †13640as ┼: 1030
    A7 ⁰1373as Ø: 1030
    B7 ⁺13730031

  • Sometimes the reserved characters are mapped (not in any obviously useful way).

Practice: Characters in Use

The previous section reported on the character sets defined in the translation table in the corpus. This section reports on the characters actually found in the corpus.

In practice, characters in the corpus are in ISO-8859-1, with very few exceptions. The exceptions are a handful of files that either use reserved characters from the portable character set, for unclear reasons, or declare surprising encodings for bytes in the normal ASCII range. These exceptions might be file corruption; they do not appear to be useful.

As a result, a portable file reader could reasonably ignore the translation table and simply interpret all portable files as ISO-8859-1 or windows-1252.

There is no visible distinction in practice between portable files in "communication" versus "tape" format. Neither kind contains control characters.

Files in the corpus have a mix of CRLF and LF-only line ends.

Tag String

The translation table is followed by an 8-byte tag string that consists of the exact characters SPSSPORT in the portable file's character set. This can be used to verify that the file is indeed a portable file.

Since every file in the corpus is encoded in (extended) ASCII, this string always appears in ASCII too.

Version and Date Info Record

This record does not have a tag code. It has the following structure:

  • A single character identifying the file format version. It is always A.

  • An 8-character string field giving the file creation date in the format YYYYMMDD.

  • A 6-character string field giving the file creation time in the format HHMMSS.

In the corpus, there is some variation for file creation dates and times by product:

  • STAT/TRANSFER often writes dates that are invalid (e.g. 20040931) or obviously wrong (e.g. 19040823, 19000607).

  • STAT/TRANSFER often writes the time as all spaces.

  • IBM SPSS Statistics 19.0 (and probably other versions) writes HH as H for single-digit hours.

  • SPSS 6.1 for the Power Macintosh writes invalid dates such as 19:11010.

Identification Records

The product identification record has tag code 1. It consists of a single string field giving the name of the product that wrote the portable file.

The author identification record has tag code 2. It is optional and usually omitted. If present, it consists of a single string field giving the name of the person who caused the portable file to be written.

The corpus contains a few different kinds of authors:

  • Organizational names, such as the names of companies or universities or their departments.

  • Product names, such as SPSS for HP-UX.

  • Internet host names, such as icpsr.umich.edu.

The subproduct identification record has tag code 3. It is optional and usually omitted. If present, it consists of a single string field giving additional information on the product that wrote the portable file.

The corpus contains a few different kinds of subproduct:

  • x86_64-w64-mingw32 or another target triple (written by PSPP).

  • A file name for a .sav file.

  • SPSS/PC+ Studentware+ written by SPSS for MS WINDOWS Release 7.0 in 1996.

  • FILE BUILT VIA IMPORT written by SPSS RELEASE 4.1 FOR VAX/VMS in 1998.

  • SPSS/PC+ written by SPSS for MS WINDOWS Release 7.0 in 1996.

  • Multiple instances of SPSS/PC+ written by SPSS/PC+ on IBM PC, but with several spaces padding out both product and subproduct fields.

  • PFF TEST FILE written by SPSS-X RELEASE 2.1 FOR IBM VM/CMS in 1986.

Variable Count Record

The variable count record has tag code 4. It consists of a single integer field giving the number of variables in the file dictionary.

Precision Record

The precision record has tag code 5. It consists of a single integer field specifying the maximum number of base-30 digits used in data in the file.

Case Weight Variable Record

The case weight variable record is optional. If it is present, it indicates the variable used for weighting cases; if it is absent, cases are unweighted. It has tag code 6. It consists of a single string field that names the weighting variable.

Variable Records

Each variable record represents a single variable. Variable records have tag code 7. They have the following structure:

  • Width (integer). This is 0 for a numeric variable. For portability to old versions of SPSS, it should be between 1 and 255 for a string variable.

    Portable files in the corpus contain strings as wide as 32000 bytes. None of these was written by SPSS itself, but by a variety of third-party products: STAT/TRANSFER, inquery export tool (c) inworks GmbH, QDATA Data Entry System for the IBM PC. The creation dates in the files range from 2016 to 2024.

  • Name (string). 1-8 characters long. Must be in all capitals.

    A few portable files that contain duplicate variable names have been spotted in the wild. PSPP handles these by renaming the duplicates with numeric extensions: VAR001, VAR002, and so on.

  • Print format. This is a set of three integer fields:

    • Format type encoded the same as in system files.

    • Format width. 1-40.

    • Number of decimal places. 1-40.

    A few portable files with invalid format types or formats that are not of the appropriate width or decimals for their variables have been spotted in the wild. PSPP assigns a default F or A format to a variable with an invalid format.

  • Write format. Same structure as the print format described above.

Each variable record can optionally be followed by a missing value record, which has tag code 8. A missing value record has one field, the missing value itself (a floating-point or string, as appropriate). Up to three of these missing value records can be used.

There are also records for missing value ranges:

  • Tag code B for X THRU Y ranges. It is followed by two floating-point values representing X and Y.

  • Tag code 9 for LO THRU Y ranges, followed by a floating-point number representing Y.

  • Tag code A for X THRU HI ranges, followed by a floating-point number representing X.

If a missing value range is present, it may be followed by a single missing value record.

In addition, each variable record can optionally be followed by a variable label record, which has tag code C. A variable label record has one field, the variable label itself (string).

Value Label Records

Value label records have tag code D. They have the following format:

  • Variable count (integer).

  • List of variables (strings). The variable count specifies the number in the list. Variables are specified by their names. All variables must be of the same type (numeric or string), but string variables do not necessarily have the same width.

  • Label count (integer).

  • List of (value, label) tuples. The label count specifies the number of tuples. Each tuple consists of a value, which is numeric or string as appropriate to the variables, followed by a label (string).

The corpus contains a few portable files that specify duplicate value labels, that is, two different labels for a single value of a single variable. PSPP uses the last value label specified in these cases.

Document Record

One document record may optionally follow the value label record. The document record consists of tag code E, following by the number of document lines as an integer, followed by that number of strings, each of which represents one document line. Document lines must be 80 bytes long or shorter.

Portable File Data

The data record has tag code F. There is only one tag for all the data; thus, all the data must follow the dictionary. The data is terminated by the end-of-file marker Z, which is not valid as the beginning of a data element.

Data elements are output in the same order as the variable records describing them. String variables are output as string fields, and numeric variables are output as floating-point fields.

  1. The strings are supposed to be in EBCDIC, 7-bit ASCII, CDC 6-bit ASCII, 6-bit ASCII, and Honeywell 6-bit ASCII. (It is somewhat astonishing that anyone considered the possibility of 6-bit "ASCII", or that there were at least three incompatible version of it.) ↩

  2. Character 0, not NUL or byte zero. ↩

  3. From the EBCDIC translation table in pff.tar.Z. The ASCII translation table leaves all of them undefined. Code points are only listed for common control characters with some modern relevance. ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7 ↩8 ↩9 ↩10 ↩11 ↩12 ↩13 ↩14 ↩15 ↩16 ↩17 ↩18 ↩19 ↩20 ↩21 ↩22 ↩23 ↩24 ↩25 ↩26 ↩27 ↩28 ↩29 ↩30 ↩31 ↩32 ↩33 ↩34 ↩35 ↩36 ↩37 ↩38 ↩39 ↩40 ↩41 ↩42 ↩43 ↩44 ↩45 ↩46 ↩47 ↩48 ↩49 ↩50 ↩51 ↩52 ↩53 ↩54 ↩55 ↩56 ↩57 ↩58 ↩59 ↩60 ↩61

  4. Reserved ↩ ↩2 ↩3 ↩4 ↩5

  5. The document describes 83 as "a solid vertical pipe" and 8F as "a broken vertical pipe". Even though the ASCII translation table in pff.tar.Z leaves position 83 undefined and translates 8F to U+007C VERTICAL LINE, using U+007C VERTICAL LINE and U+00A6 BROKEN BAR, respectively, seem more accurate in a Unicode environment. ↩ ↩2

  6. Unicode inferred from document description as "empty box". ↩

  7. Unicode inferred from document description as "filled box". ↩

  8. These characters are as described in the document. Some of these don't appear in any known EBCDIC code page, but the EBCDIC translations given in pff.tar.Z match the graphics shown in RFC 183 with those hex codes. ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  9. Described in document as "horizontal dagger", which doesn't appear in Unicode or any known code page. This interpretation from RFC 183 seems more likely. ↩

  10. Unicode inferred from document description as "centered dot, or bullet" ↩

SPSS/PC+ System File Format

SPSS/PC+, first released in 1984, was a simplified version of SPSS for IBM PC and compatible computers. It used a data file format related to the one described in the previous chapter, but simplified and incompatible. The SPSS/PC+ software became obsolete in the 1990s, so files in this format are rarely encountered today. Nevertheless, for completeness, and because it is not very difficult, it seems worthwhile to support at least reading these files. This chapter documents this format, based on examination of a corpus of about 60 files from a variety of sources.

System files use four data types: 8-bit characters, 16-bit unsigned integers, 32-bit unsigned integers, and 64-bit floating points, called here char, uint16, uint32, and flt64, respectively. Data is not necessarily aligned on a word or double-word boundary.

SPSS/PC+ ran only on IBM PC and compatible computers. Therefore, values in these files are always in little-endian byte order. Floating-point numbers are always in IEEE 754 format.

SPSS/PC+ system files represent the system-missing value as -1.66e308, or f5 1e 26 02 8a 8c ed ff expressed as hexadecimal. (This is an unusual choice: it is close to, but not equal to, the largest negative 64-bit IEEE 754, which is about -1.8e308.)

Text in SPSS/PC+ system file is encoded in ASCII-based 8-bit MS DOS codepages. The corpus used for investigating the format were all ASCII-only.

An SPSS/PC+ system file begins with the following 256-byte directory:

uint32              two;
uint32              zero;
struct {
    uint32          ofs;
    uint32          len;
} records[15];
char                filename[128];

  • uint32 two;
    uint32 zero;
    Always set to 2 and 0, respectively.

    These fields could be used as a signature for the file format, but the product field in record 0 seems more likely to be unique.

  • struct { ... } records[15];
    Each of the elements in this array identifies a record in the system file. The ofs is a byte offset, from the beginning of the file, that identifies the start of the record. len specifies the length of the record, in bytes. Many records are optional or not used. If a record is not present, ofs and len for that record are both are zero.

  • char filename[128];
    In most files in the corpus, this field is entirely filled with spaces or null bytes. In others, it contains a filename, which generally contains doubled backslashes, e.g. c:\\doli\\altm\\f_sum94.sys. The unusual extension (_) is also common, e.g. DER56.(_).

The following sections describe the contents of each record, identified by the index into the records array.

Record 0: Main Header Record

All files in the corpus have this record at offset 0x100 with length 0xb0 (but readers should find this record, like the others, via the records table in the directory). Its format is:

uint16              one0;
char                family[2];
char                product[60];
flt64               sysmis;
uint32              zero0;
uint32              zero1;
uint16              one1;
uint16              compressed;
uint16              nominal_case_size;
uint16              n_cases0;
uint16              weight_index;
uint16              unknown;
uint16              n_cases1;
uint16              zero2;
char                creation_date[8];
char                creation_time[8];
char                file_label[64];

  • uint16 one0;
    uint16 one1;
    Always set to 1.

  • uint32 zero0;
    uint32 zero1;
    uint16 zero2;
    Always set to 0.

  • uint16 unknown; Unknown meaning. Usually set to 0.

  • char family[2];
    Identifies the product family that created the file. This is either PC for SPSS/PC+ and related software, or DE for SPSS Data Entry and related software.

  • char product[60];
    Name of the program that created the file. Only the following unique values have been observed, in each case padded on the right with spaces:

    SPSS/PC+ System File Written by Data Entry II
SPSS SYSTEM FILE.  IBM PC DOS, SPSS/PC+
SPSS SYSTEM FILE.  IBM PC DOS, SPSS/PC+ V3.0
SPSS SYSTEM FILE.  IBM PC DOS, SPSS for Windows

    Thus, it is reasonable to use the presence of the string SPSS at offset 0x104 as a simple test for an SPSS/PC+ data file.

  • flt64 sysmis;
    The system-missing value, as described previously.

  • uint16 compressed;
    Set to 0 if the data in the file is not compressed, 1 if the data is compressed with simple bytecode compression.

    The corpus contains a mix of compressed and uncompressed files.

  • uint16 nominal_case_size;
    Number of data elements per case. This is the number of variables, except that long string variables add extra data elements (one for every 8 bytes after the first 8). String variables in SPSS/PC+ system files are limited to 255 bytes.

  • uint16 n_cases0;
    uint16 n_cases1;
    The number of cases in the data record. Both values are the same.

    Readers must use these case counts because some files in the corpus contain garbage that somewhat resembles data after the specified number of cases.

  • uint16 weight_index;
    0, if the file is unweighted, otherwise a 1-based index into the data record of the weighting variable, e.g. 4 for the first variable after the 3 system-defined variables.

  • char creation_date[8];
    The date that the file was created, in mm/dd/yy format.

    Single-digit days and months are not prefixed by zeros. The string is padded with spaces on right or left or both, e.g. _2/4/93_, 10/5/87_, and _1/11/88 (with _ standing in for a space) are all actual examples from the corpus.

  • char creation_time[8];
    The time that the file was created, in HH:MM:SS format.

    Single-digit hours are padded on the left with a space. Minutes and seconds are always written as two digits.

  • char file_label[64];
    File label declared by the user, if any. Padded on the right with spaces.

Record 1: Variables Record

The variables record most commonly starts at offset 0x1b0, but it can be placed elsewhere. The record contains instances of the following 32-byte structure:

uint32              value_label_start;
uint32              value_label_end;
uint32              var_label_ofs;
uint32              format;
char                name[8];
union {
    flt64           f;
    char            s[8];
} missing;

The number of instances is the nominal_case_size specified in the main header record. There is one instance for each numeric variable and each string variable with width 8 bytes or less. String variables wider than 8 bytes have one instance for each 8 bytes, rounding up. The first instance for a long string specifies the variable's correct dictionary information. Subsequent instances for a long string are generally filled with all-zero bytes, although the missing field contains the numeric system-missing value, and some writers also fill in var_label_ofs, format, and name, sometimes filling the latter with the numeric system-missing value rather than a text string. Regardless of the values used, readers should ignore the contents of these additional instances for long strings.

  • uint32 value_label_start;
    uint32 value_label_end;
    These specify offsets into the label record of the start and end of value labels for this variable. They are zero if there are no value labels. See the labels record, for more information. A long string variable may not have value labels.

    Sometimes the data is, instead of value labels, some form of data validation rules for SPSS Data Entry. There is no known way to distinguish, except that data validation rules often cannot be interpreted as valid value labels because the label length field makes them not fit exactly in the allocated space.

    It appears that SPSS products cannot properly read these either. All the files in the corpus with these problems are closely related, so it's also possible that they are corrupted in some way.

  • uint32 var_label_ofs;
    For a variable with a variable label, this specifies an offset into the label record. See the labels record, for more information.

    For a variable without a variable label, this is zero.

  • uint32 format;
    The variable's output format, in the format used in system files. SPSS/PC+ system files only use format types 5 (F, for numeric variables) and 1 (A, for string variables).

  • char name[8];
    The variable's name, padded on the right with spaces.

  • union { ... } missing;
    A user-missing value. For numeric variables, missing.f is the variable's user-missing value. For string variables, missing.s is a string missing value. A variable without a user-missing value is indicated with missing.f set to the system-missing value, even for string variables (!). A long string variable may not have a missing value.

In addition to the user-defined variables, every SPSS/PC+ system file contains, as its first three variables, the following system-defined variables, in the following order. The system-defined variables have no variable label, value labels, or missing values. PSPP renames these variables to start with @ when it reads an SPSS/PC+ system file.

  • $CASENUM
    A numeric variable with format F8.0. Most of the time this is a sequence number, starting with 1 for the first case and counting up for each subsequent case. Some files skip over values, which probably reflects cases that were deleted.

  • $DATE
    A string variable with format A8. Same format (including varying padding) as the creation_date field in the main header record. The actual date can differ from creation_date and from record to record. This may reflect when individual cases were added or updated.

  • $WEIGHT
    A numeric variable with format F8.2. This represents the case's weight. If weighting has not been enabled, every case has value 1.0.

Record 2: Labels Record

The labels record holds value labels and variable labels. Unlike the other records, it is not meant to be read directly and sequentially. Instead, this record must be interpreted one piece at a time, by following pointers from the variables record.

The value_label_start, value_label_end, and var_label_ofs fields in a variable record are all offsets relative to the beginning of the labels record, with an additional 7-byte offset. That is, if the labels record starts at byte offset labels_ofs and a variable has a given var_label_ofs, then the variable label begins at byte offset labels_ofs + var_label_ofs + 7 in the file.

A variable label, starting at the offset indicated by var_label_ofs, consists of a one-byte length followed by the specified number of bytes of the variable label string, like this:

uint8               length;
char                s[length];

A set of value labels, extending from value_label_start to value_label_end (exclusive), consists of a numeric or string value followed by a string in the format just described. String values are padded on the right with spaces to fill the 8-byte field, like this:

union {
    flt64           f;
    char            s[8];
} value;
uint8               length;
char                s[length];

The labels record begins with a pair of uint32 values. The first of these is always 3. The second is between 8 and 16 less than the number of bytes in the record. Neither value is important for interpreting the file.

Record 3: Data Record

The format of the data record varies depending on the value of compressed in the file header record:

  • 0: no compression
    Data is arranged as a series of 8-byte elements, one per variable instance variable in the variable record. Numeric values are given in flt64 format; string values are literal characters string, padded on the right with spaces when necessary to fill out 8-byte units.

  • 1: bytecode compression
    The first 8 bytes of the data record is divided into a series of 1-byte command codes. These codes have meanings as described below:

    • 0
      The system-missing value.

    • 1
      A numeric or string value that is not compressible. The value is stored in the 8 bytes following the current block of command bytes. If this value appears twice in a block of command bytes, then it indicates the second group of 8 bytes following the command bytes, and so on.

    • 2 through 255
      A number with value CODE - 100, where CODE is the value of the compression code. For example, code 105 indicates a numeric variable of value 5.

    The end of the 8-byte group of command codes is followed by any 8-byte blocks of non-compressible values indicated by code 1. After that follows another 8-byte group of command codes, then those command codes' non-compressible values. The pattern repeats up to the number of cases specified by the main header record have been seen.

    The corpus does not contain any files with command codes 2 through 95, so it is possible that some of these codes are used for special purposes.

Cases of data often, but not always, fill the entire data record. Readers should stop reading after the number of cases specified in the main header record. Otherwise, readers may try to interpret garbage following the data as additional cases.

Records 4 and 5: Data Entry

Records 4 and 5 appear to be related to SPSS/PC+ Data Entry.