Abstract

This is a brief guide to the essentials needed to know to write an SPSS program. Although this document was written for those using SPSS on the Cunix cluster, most, if not all commands, should work on other operating systems. See our brief handout, SPSS on the Cunix Cluster or the SPSS manual specific to your particular operating system for instructions on running an SPSS program.

Note to SPSS for Windows users: If you are using SPSS for Windows you probably will never need to write a program. However, you may receive data with a syntax file. This handout can help you understand it. If you have trouble running that file on your PC, please contact us at EDS. We can help.

An SPSS Program Has 4 Parts:

Note: Steps #1 and #2 take most of the time, so plan accordingly.

1. Defining and reading the data (FILE HANDLE and DATA LIST or GET FILE).
2. Selecting and/or modifying the data (SELECT IF, RECODE, COMPUTE, etc.).
3. Statistical procedure(s) (FREQUENCIES, CROSSTABS, REGRESSION etc.).
4. Saving a "save file" [optional but recommended for repeated runs]. "Save files" used to be called "system files". You may see that usage in old codebooks.

Basic Rules for Writing Commands:

• A command must begin in column 1.
• Continuation lines must be indented at least one space.
• A command ends with a "." (period).
• Case does not matter except in the name of the file.
• Quotes can be single or double but they must match.

Part 1 - Defining and Reading in the Data:

First, you need to know what type of data are you working with. Is it "raw data", an "SPSS save file", an "SPSS portable file" or something else?

SPSS "save files" (.sav) are special format files that SPSS creates. In the past SPSS save files could only be read on the same type of computer/operating system where they were created. Newer versions of SPSS save files can usually be moved to different operating systems without problem.

SPSS "portable files" (.por) are also special format files that SPSS creates. These are actually ASCII files and can be transported to other computers and operating systems without problems. These are, however, seldom seen anymore. SPSS save and portable files are easy to use in that all the variables are already defined. They can be read directly into Spss without a program.

If you have "something else", e.g., a EXCEL file or a Stata file and you have Spss version 15 or later, you can also read data directly into Spss without a program. Otherwise the file will have to be "transformed" to SPSS format. Stat/transfer is available for this purpose on cunix and at EDS. You can also buy your own copy. See this web page for more information.

If you have raw data, you will have to do all the work of defining the variables yourself, i.e, write a program.

Second, you need to know where the data is, i.e., you need a "filename" and possibly a "path" if the file is not in your home directory. Examples:

• If you are reading an SPSS .sav file you need:
 
1. Documentation listing the variables you want and their mnemonic names in the system file.
 
2. An SPSS program with
   • a GET FILE = command with the name of the save file in quotes,
   • (optional) a /KEEP (or /DROP) subcommand with the names of the variables you want to use (or not use). If you need all the variables, leave this out.
   • a period to end the command.
   
   

        Example:   GET FILE="/u/9/s/me2000/surveys/file1.sav"
                     /KEEP zodiac sex.
     

• If you are reading raw (ASCII) data you need:
 
1. Documentation describing the variables.
 
2. Mnemonic names for the variables you want (up to 8 characters each only) with
   • Each variable's column position(s) in the file, e.g. 1-4, and
   • Each variable's type, i.e., integer, decimal, or alphanumeric.
   You make up the names. You can use V[n] to V[m], e.g., V1 to V100, if you want, but using mnemonic names is a lot easier in the long run. You don't have to define all the variables in your file, just the ones you need.
 
3. The length of the records in the file (the LRECL).
 
4. An SPSS program with
   • A FILE HANDLE command with
     • a "handle" (" be indented at least one space.
     • A command ends with a "IN" in the example below),
     • a "/" (slash),
     • a NAME= subcommand with the name of the raw data file in quotes,
     • the LRECL= subcommand, and
     • a period to end the command.
   
    
   • A DATA LIST command with
     • a FILE=handle subcommand,
     • a "/" (slash),
     • and then the list of the mnemonic variable names, each followed by its column positions, and its type if it has a decimal place or is an alphanumeric, and
     • a period to end the command.
     
     

     Example:   FILE HANDLE MYFILE/NAME="file1.dat" lrecl=1200.
                DATA LIST FILE=MYFILE /
                       PERSONID   1-4
                       SEX        6
                       BIRTHYR    7-10
                       INCOME     15-21 (2)
                       STATE      55-56 (A).
     

     In the example above, INCOME has 2 decimal places and STATE is a 2 column character variable.
     Note that you don't have to define all the variables in your file, just the ones you need.
    
   • Some raw data files can have multiple line of data for each case.
     This frequently happens with opinion surveys where the responses from one respondent are reported on two or three lines (in the documentation often referred to as "records" or "cards"). Use the subcommand, RECORDS= following the DATA LIST command.
     • a records=# subcommand placed after the file handle, with # = the number of records per case,
     • a "/" (slash) marking the start of each record followed by an interger that indicates which record it is.

     Example:   FILE HANDLE MYFILE/NAME="file3.dat".
                DATA LIST FILE=MYFILE records=3
                     /1
                     P-ID-REC1  1-4
                     SEX        6
                     BIRTHYR    7-10
                     INCOME     15-21 (2)
                     STATE      55-56 (A)
                     /3
                     P-ID-REC3  1-4
                     industry   5-8
                     occup      9-11.
     

     In the example above, there are three records per case. Note that no variables are defined for record type=2. You only need to define the variables you need.
    
5. If you are reading an SPSS portable file you need:
    
   1. Documentation listing the variables you want and their mnemonic names in the portable file.
    
   2. An SPSS program with
      • an IMPORT FILE= command with the name of the portable file in quotes,
      • (optional) a /KEEP (or /DROP) subcommand with the names of the variables you want to use (or not use). If you need all the variables, leave this out.
      • a period to end the command.

      Example:  
        IMPORT FILE="/eds/datasets/gss/data/gss94-all.por"
                   /KEEP zodiac sex.
      

Part 2 - Selecting and Modifying the Data:

This part is optional. You may not need to select cases or modify or create new variables. But if you do, these are the most common commands.

1. SELECT IF - This command selects whole CASES, usually people.
   

   Examples:   SELECT IF (sex = 1).
               SELECT IF (STATE = "NJ").
               select if (any(racegrp,4,5,6,8)).
   

   Warning! The effect of multiple SELECT IF statements is cumulative. See the manual on using the TEMPORARY command if you don't want this.
    
2. COMPUTE - Create a new variable.

   Examples:  COMPUTE NEWAGE=0.
              COMPUTE YRRETIRE=BIRTHYR+65.
              COMPUTE INCOME=salary+interest+divdnds.
   

    
3. RECODE - Change the values of a variable. It is best to do this on a new variable created from an old one so you don't lose the old values. You never know when you may have to backup and use them again. The default format for new integer variables is F8.2. It's worth making this more efficient with the FORMAT command.

Example:   RECODE AGE (MISSING=9)(18 thru HI=1)(LOW thru 18=0) into VOTER.
           RECODE PLACE (1=1)(2 thru 7=2)(else=0) into CITYTOWN.
           RECODE MONTH (" "=99) (CONVERT) ("-"=11)("&"=12) into NEWMONTH
           FORMATS VOTER CITYTOWN (F1.0) NEWMONTH (F2.0)

 
4. IF - Conditional change. This is useful for cleaning data as well as recoding (3rd example below).



Examples:  COMPUTE WORKWK=0.
           IF (WORK GT 0 and WORK LE 35) WORKWK=1.
           COMPUT PLRTY = 1.
           IF RANGE(VALUE(PLURALTY),2,8) PLRTY = 2.
           IF (STATE EQ "JN") STATE="NJ".

 
5. MISSING VALUES - Declare some values of a variable "missing" so they won't be used in statistical calculations.

Example:   MISSING VALUES AGE (0) 
                          Score1 to Score10 (999) 
                          STATE ("XX").


Warning! Missing Values affect RECODE and COMPUTE statements and can have unexpected results. When you create or modify any variable be sure to check very carefully what happened with the Missing Values. For example:

           MISSING VALUES PLURALTY (2 THRU 8).
           COMPUTE PLRTY=PLURALTY.
           RECODE PLRTY (2 THRU 8 = 2).


won't work. Cases coded 2 through 8 are Missing and won't be recoded. (See the Manual for the VALUE function to get around this.)

    TITLE  "A very simple program".
    set width 80.
    file handle in /name="/p/us/sue/zspssx/famous.dat".
    DATA LIST FILE=in /
         idnum    1-3 (N) 
         fname    4-15 (A) 
         lname   16-27 (A)
         age     28-29 
         sex     30 
         byear   31-34 
         dyear   35-38 
         status  39.
    VAR LABELS
         idnum   "Case Number"
         fname   "First Name"
         lname   "Last Name"
         age     "Age at Death"
         sex     "Sex"
         byear   "Year of Birth"
         dyear   "Year of Death"
         status  "Status".
    VALUE LABELS
         sex
             1 "Male"
             2 "Female" /
         status
             1 "Real"
             2 "Fictional"
             3 "Possibly Real".
    missing values
         status (3).
    select if (sex eq 2).
    freq vars=status.
    save outfile="famous_females.sav".