Mastering SPSS Syntax for Advanced Data Analysis

SPSS Syntax to the Next Level

Presented by Christine R. Wells, Ph.D.

Statistical Methods and Data Analytics

UCLA Office of Advanced Research Computing

Introductory topics: Setting options

•

Using SPSS version 28.0.0.0

•

Setting options (Edit -> Options)

•

General tab: Mode, Variable Lists, Output

•

Viewer tab: Syntax echoed in output

•

Viewer tab and Pivot Tables tab: Font sizes

•

Output tab: Outline labeling

•

Charts tab:  Chart Template Optional settings -> APA style

•

File Locations tab: Journal file and Startup Folders

•

Syntax Editor: Auto-Complete Settings

•

Privacy: your choice!

•

Click on Apply and then OK

•

The SPSS Command Syntax Reference is your new best friend!

Introductory topics:  SPSS syntax

•

This workshop focuses on using SPSS syntax rather than point-and-

click

•

The 

comment

 command

•

Other ways to add comments (* and /* */)

•

SPSS is not case sensitive

•

The period at the end of the command is the end-of-command

marker

•

Commands can span multiple lines, even if a new subcommand is not

being specified

Introductory topics:  SPSS syntax

•

Commands and subcommands

•

Editor coloring

•

Shortened names of commands (may not get editor coloring)

•

SPSS keywords

•

Two types of variables: numeric and string (more on these later)

•

Will not be discussing dates, but dates can be stored as either

numeric or string

Introductory topics:

When do commands execute?

•

SPSS commands are executed by going down the data file row by row

•

If multiple commands are submitted simultaneously, the commands

are executed in the order in which they are encountered

•

Except for very complicated analyses, the slowest part of executing a

command is reading through the data file

•

Because of this, SPSS tries to limit the number of times is must read

the active dataset or “make a pass through the data”

Introductory topics:

When do commands execute?

•

Pages 37-40 lists the 1) commands that take effect immediately without

reading the active dataset or executing pending transformations and 2)

commands that are stored pending execution.  Procedures (AKA things that

produce output) are executed immediately and force SPSS to read the

active dataset.

•

Many of the data transformation commands covered in this workshop are

on the list of commands that are stored pending execution.

•

Pending command or commands that do not force SPSS to read the data

can be executed with 

execute

 command, often shorted to 

exe.

•

Procedure commands can also be used to execute pending data

transformation commands.

•

You will know if commands are pending execution by looking in the lower

right-hand corner of the Data Editor window (transformations pending).

Introductory topics:

SPSS Command Syntax Reference

•

The ultimate source for information regarding the built-in SPSS

commands.

•

Familiarizing yourself with the first 92 or so pages is a very good use

of time.

•

Have a look at the entry for the 

aggregate

 command.

•

Notice that multiple subcommands can appear on one line of the

syntax diagram.

•

Bold means default if subcommand or keyword is omitted.

Getting data into SPSS:  The get command

•

An SPSS data file as one of the following extensions: sav, zsav, por

•

Syntax files have an extension of sps; syntax files are just text files

•

Output files have an extension of spv (spo is the old extension)

•

Use the 

get file

 command

get file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo.sav".

•

The file extension is needed; otherwise, an error is put in the output

saying that the file is not found.

•

Include the file extension every time you read or save a file.

Getting data into SPSS:  Dataset activate

•

SPSS will allow you to have many data files open at once.

•

While this may be handy, it can also be problematic when executing syntax,

because the syntax will execute on the active dataset.

•

Hence, a command is needed to control which open dataset is the active

dataset.

•

First, name the open dataset with 

dataset name

.

•

The command to make an open dataset active is 

dataset activate

.

•

If you run syntax and get strange error messages about variables not found,

etc., you probably ran the syntax on the wrong data file.

•

Everyone does this!  Just activate the dataset you want and run the syntax

again (click on big green arrow or Control-R or click on Run…).

dataset name hsbdemo.

Getting data into SPSS: The get sas command

get sas data =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo_sas.sas7bdat".

dataset name sas.

get sas data =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo_sas.sas7bdat"

    /formats = "D:\data\seminars\SPSS_syntax_2022\formats.sas7bcat".

dataset name saswithformats.

Getting data into SPSS: The get stata command

get stata file =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo_stata.dta".

dataset name stata.

•

Notice that with the 

get sas

 command the keyword is 

data

, but with

the 

get stata

 command, the keyword is 

file

•

SPSS can usually read the latest version of Stata data files, unless the

latest release of Stata is more recent than the latest version of SPSS

Getting data into SPSS:  The get data

command

get data

    /type = xlsx

    /file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo_excel.xlsx"

    /sheet = name "hsbdemo"

    /readnames = on

    /assumedstrwidth = 500

    /hidden ignore = no.

dataset name excel.

Getting data into SPSS:  The get data command

get data

    /type = txt

    /file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo_csv.dat"

    /delimiters = ","

    /firstcase = 1

    /variables = id f2.0 female f1.0 ses f1.0 schtyp f1.0 progtype f1.0

    read f1.0 write f1.0 math f1.0 science f1.0 socst f1.0 honros f1.0

    awards f1.0 cid f1.0.

dataset name csv.

Getting data into SPSS: The get data command

get data

/type = txt

/file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo_tab.dat"

/delimiters = "\t"

/firstcase = 2

/variables = id f2 female f1 ses f1 schtyp f1 prog f1 read f2 write f2

math f2 science f2 socst f2 honors f1 awards f1 cid f2.

dataset name tab.

Getting data into SPSS:  Doing it yourself!

data list list

/id (f2.0) v1 (f2.2) v2 (f1.0) v3 (f5) stringvar1 (a5).

begin data.

12 .63 5 12548 abcde

16 .98 7 98745 jklmn

22 .01 2 15963 fdsaq

55 .00 6 35741 poiuy

79 .33 1 75321 qwert

end data.

dataset name littletest.

list.

Dataset commands

•

dataset name

: names the active dataset

•

dataset activate

: activates the dataset

•

dataset declare

: creates a new dataset that is not associated with any open

dataset (helpful when you need a temporary dataset)

•

dataset display

:  displays a list of the currently available datasets

•

dataset copy

:  creates a new dataset that captures the current state of the

active dataset (the current state of the active dataset may be different than

the state of saved dataset).  The copy is not saved to your computer, but

you can do that if you wish

•

dataset close

:  closes the named dataset. If the keyword 

all

 is used, all but

the active dataset are closed.

Examples using the dataset commands

dataset close sas.

dataset display.

dataset activate hsbdemo.

dataset close all.

dataset display.

dataset name hsbdemo.

dataset display.

Example datasets

•

We will mostly be using the 

hsbdemo

 dataset.

•

Based on real data  but heavily edited so that our examples work

(don’t do that with your data!!!).

•

200 cases representing students in school who took tests and

provided demographic information.

•

We will input small datasets as needed.

Detour:  The temporary command

•

We will use the 

temporary

 command in several of the examples in

this workshop.

•

The 

temporary

 command signals the beginning of temporary

transformations that are in effect only for the next procedure.

•

The temporary command does not read the active dataset; rather, it

is stored pending execution with the next command that reads the

dataset.

•

The 

temporary

 command can be used with 

compute

, 

recode

, 

if

,

count

, 

do repeat

, 

loop

, 

do if

, 

select if

, 

sample

, 

filter

, 

formats

,

numeric

, 

string

, 

split file

, 

variable labels

, 

value labels

, 

missing value

s

and 

weight

 (and a few other commands!).

Dataset manipulation commands

•

flip

: transposes rows and columns; don’t use with string variables

•

sample

:  samples cases from the active dataset

•

n of cases

: uses the first n cases

•

sort cases

: sorts the rows in the active dataset

•

sort variables

: sorts the variables in the active dataset

Dataset manipulation: The flip command

•

The 

flip

 command restructures the active dataset such the rows

become columns and the columns become rows.

•

Use the 

casestovars

 or 

varstocases

 commands to reshape data.

•

The 

flip

 command read the active dataset and will cause the

execution of any pending transformations.

•

The 

flip

 command assigns system missing values to string variables in

the active dataset.

•

The 

flip

 command does not respect the 

temporary

 command.

•

We will make a small example dataset to use with this command so

that the change is easy to see.

Dataset manipulation: The flip command

data list list

/id (f2.0) v1 (f2.2) v2 (f1.0) v3 (f5).

begin data.

12 .63 5 12548

16 .98 7 98745

22 .01 2 15963

55 .00 6 35741

79 .33 1 75321

end data.

list.

dataset name little.

list.

flip.

list.

dataset close all.

Dataset manipulation: The sample command

•

The 

sample

 command draws a random sample of cases from the

active dataset

•

The command does not read the active dataset; rather, it is stored

pending execution.

•

Sample

 is a permanent transformation.

•

Sample

 is based on a pseudo-random-number generator that

depends on a seed value that is set by the program.

•

Often used with the temporary command so that the change to the

dataset is not permanent.

Dataset manipulation: The sample command

get file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo.sav".

dataset name hsbdemo.

dataset activate hsbdemo.

dataset copy hsbdemo1.* sample is a permanent transformation!.

* may want to set the seed before doing this so that the results are

replicable.

set seed 3698521.

sample .5.

* notice the "Transformations pending" in the lower right corner.

exe.

dataset close hsbdemo1.

Dataset manipulation: The sample command

dataset activate hsbdemo.

dataset copy hsbdemo2.

sample 50 from 200.

exe.

dataset close hsbdemo2.

Data manipulation: The n of cases command

•

The 

n of cases

 command limits the analyses to the n cases of the

active dataset.

•

The 

n of cases

 command is often combined with the 

temporary

command.

•

This can be useful if the data file has many cases and therefore takes

a long to run.

•

Remember that the effect of the 

temporary

 command ends when the

next procedure is executed.

Data manipulation: The n of cases command

get file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo.sav".

dataset name hsbdemo.

dataset activate hsbdemo.

freq var = female.

temporary.

n of cases 100.

freq var = female.

Data manipulation: The sort cases command

•

The 

sort cases

 command reorders the cases in the active dataset based on the

values of one or more variables.

•

If more than one 

by

 variable is provided, the data are sorted based on the first

variable listed, and the sorted within each value of the first variable by the

second variable.

•

The data can be sorted in ascending or descending order.

•

The keyword 

by

 is optional.

•

The 

by

 variables can be numeric or string, but not system, scratch or temporary

variables.

•

The sorted data can be saved to a new file using the 

outfile

 subcommand.

•

There is a 

passprotect

 subcommand.

•

You cannot sort by more than 64 variables at once!

Data manipulation: The sort cases command

sort cases by id.

sort cases by id (d).

sort cases by cid (a) id (d).

•

The sort order of the variable 

id

 within each value of 

cid

 depends on

the locale-defined order.  The sort order of rows with cid may be

different with each value of cid if the variable id does not uniquely

identify the rows.  This may be a problem when creating variables

based on the sort order of the data.

Data manipulation:

The sort variables command

•

The 

sort variables

 command rearranges the order of the variables in the active dataset.

•

Only one dictionary attribute can be specified.

•

The keyword 

by

 is optional.

•

Variables can be sorted in ascending order using (a) or (up).

•

Variables can be sorted in descending order using (d) or (down).

•

The variables can be ordered by the following:

•

Name

•

Type

•

Format

•

Label

•

Values

•

Missing

•

Measure

•

Role

•

Columns

•

Alignment

•

Attribute

name

Data manipulation:

The sort variables command

sort variables by name.

sort variables by type.

sort variables by role.

Data manipulation:

The delete variables command

•

The 

delete variables

 command deletes the specified variables from

the active dataset.

•

The 

delete variables 

command takes effect immediately, but it does

not read the data or execute pending transformations.

•

The 

delete variables

 command cannot be executed when there are

pending transformations.

•

The 

delete variables

 command cannot be used to delete all of the

variables from the active dataset.

•

The 

delete variables

 command cannot be used with the 

temporary

command.

Data manipulation:

The delete variables command

delete variables awards.

Creating variables: Two types of variables

•

There are two types of variables in SPSS:  numeric and string.

•

Numeric variables can contain only numbers.

•

String variables may contain numbers, letters or characters (e.g., @,

#, $ %, <, +, etc.).

•

The maximum length of a string variable is 32,767 characters as of

version 13 (but you need to use a work-around in versions 13 and 14).

•

A null string is considered a valid value for a string variable unless it

has been declared as a user-defined missing value.

Creating variables:

more about string variables

•

String variables may be used in logical expressions, but they may not be

compared to numeric variables.

•

If string variables are of different lengths, the shorter string is right-padded

with blanks to equal the length of the longer string.

•

The magnitude of strings can be compared using LT, GT, etc., but the

outcome depends on the sorting sequence of the computer, so use with

caution.

•

User-defined missing string values are treated the same as nonmissing

string values when evaluating string variables in logical expressions.  This

means that all string values are treated as valid, nonmissing missing values

in logical expressions.

Creating variables: System variables

•

System variables are special variables created during a working

session to keep system-required information.

•

The names of system variables begin with a dollar sign ($).

•

System variables cannot be modified, nor can its print or write format

be altered.

•

System variables cannot be used in procedures, but they can be

useful in creating new variables.

•

There are eight system variables (although some are much more

useful than others).

Creating variables: System variables

•

$casenum: current case sequence number.

•

$sysmis: system-missing value.

•

$jdate:  current date in number of days from October 14, 1582.

•

Question: Why is October 14, 1582 important?.

•

$date:  current date in international date format with two-digit year (format A9, dd-

mmm-yy).

•

$date11:  current date in international date format with four-digit year (format A9, dd-

mmm-yyyy).

•

$time: current date and time; $time represents the number of seconds from midnight,

Oct. 14, 1582 to the date and time when the transformation command is executed.

format F20.

•

$length: current page length; format is F11.0 (see 

set

 for more info).

•

$width: current page length; format is F3.0 (see 

set

 for more info).

Creating variables: System variables

compute newid = $casenum.

compute newvar = $sysmis.

compute currentdate = $jdate.

exe.

Creating variables: Scratch variables

•

Scratch variables are temporary variables whose name starts with #.

•

Scratch variables can be either numeric or string.

•

Scratch variables are initialized to 0 for numeric variables and blank for

string variables.

•

Scratch variables cannot be used in procedures and cannot be saved to a

dataset.

•

Scratch variables are not reinitialized when a new case is read.

•

Scratch variables cannot be assigned missing values, variable names or

value labels.

•

Scratch variables are discarded when a procedure begins or when the

temporary

 command is encountered.

Creating variables: Scratch variables

NOTE:  The data must be listed in a single column (not row) in order for the data file

to be correctly entered.

data list list / a.

begin data.

1 2 3 1 2 3 4 1 2 3 4 5 6 1 2 1 2 3

end data.

compute #x = #x + 1.

if a ne 1 #x = lag(#x).

compute x = #x.

exe.

list.

dataset name scratchex.

dataset close scratchex.

Creating variables: Relational operators

•

eq

 or 

=

 : equal to

•

ne

 or 

~=

 or 

<>

: not equal to

•

lt

 or 

<

: less than

•

le

 or 

<=

: less than or equal to

•

gt

 or 

>

: greater than

•

ge

 or 

>=

: greater than or equal to

•

and

 or 

&

: both must be true

•

or

 or 

|

: either relation can be true

•

not

: reverses the outcome of an expression

Creating variables: Order of evaluation

•

When arithmetic operators and functions are used in a logical expression,

the order of operations is functions and arithmetic operations first, then

relational operators, and then logical operators.

•

When more than one logical operator is used, 

not

 is evaluated first, then

and

, and then 

or

.

•

To change the order of evaluation, use parentheses.

•

Each argument to a logical function (expression, variable name, or

constant) must be separated by a comma.

•

The target variable for a logical function must be numeric.

•

The functions 

range

 and 

any

 can be useful shortcuts to more complicated

specifications on the 

if

, 

do if

, and other conditional commands.

Creating variables: Keywords

•

All

•

To

•

Thru

•

Hi or highest

•

Lo or lowest

•

By

•

With

Creating variables: The numeric and string

commands

dataset activate hsbdemo.

numeric v1 to v6 (f4.0)

    /v7 v8 (f1.0).

string county (a20).

string a1 to a4 (a1)

    /a5 to a10 (a2).

Creating variables: The compute and if

commands

•

The 

compute

 and 

if

 commands are the two main commands for

creating new numeric variables.

compute var1 = 5.

exe.

•

Need to use exe. after the 

compute

 command to execute

immediately.

•

There is no "then" in if-then logic in SPSS.

if female = 1 var1 = 6.

freq var = var1.

Creating variables:  “and” and “or”

if prog = 1 and female = 0 and id lt 100 var2 = 0.

•

Be careful with "or".

if prog = 2 or female = 1 or id gt 190 var2 = 1.

freq var = var2.

if (prog = 3 and female = 1) or id gt 180 var2 = 3.

if prog = 3 and (female = 1 or id gt 180) var2 = 4.

freq var = var2.

•

Functions can be used as part of the logical expression.

if abs(read - write) gt 7 var2 = 5.

freq var = var2.

Creating variables: Enumerating cases by

group

sort cases by cid.

compute npergroup = 1.

if cid = lag(cid) id = lag(npergroup) + 1.

exe.

Creating variables: Creating dummy variables

freq var = ses.

compute ses1 = (ses = 1).

compute ses2 = (ses = 2).

compute ses3 = (ses = 3).

•

Warning:  the table is not easy to read!.

crosstabs

    /tables = ses by ses3 by ses2 by ses1.

Creating variables:  Using numeric functions

compute dvar = read/write.

compute rndvar = rnd(dvar).

compute truncvar = trunc(dvar).

compute sumvar = sum(read to socst).

means dvar rndvar truncvar sumvar.

Creating variables:  Using numeric functions

•

The 

normal

 function creates a new numeric variable with a mean of 0

and a standard deviation of the value given in parentheses.

compute normrand = normal(1).

means tables =  normrand.

Question:  How can you use the 

normal

 function to create a simple

random sample of your data?

Creating variables: String variables

string stringf (a6).

if female = 1 stringf = "female".

if female = 0 stringf = "male".

exe.

string strings (a6).

if ses = 1 strings = "low".

if ses = 2 strings = "medium".

if ses = 3 strings = "high".

if id gt 4 and id lt 11 strings = "".

exe.

Creating variables: Using string functions

compute flength = char.length(stringf).

•

There is a problem here - what is it?

string subf (a2).

compute subf = char.substr(strings, 1, 3).

freq var = flength subf.

string subf3 (a3).

compute subf3 = char.substr(strings, 1, 3).

freq var = flength subf3.

Creating variables:

The clear transformations command

•

This is a super silly example for a reason!!

freq var = ses female prog.

recode ses, female, prog (1 = 2) (1 = 2) (3 = 4).

compute vsum = ses + female + prog.

variable labels vsum "sum of items".

clear transformations.

display dictionary.

freq var ses female prog.

Creating variables:  The aggregate command

•

The functions that can be used with the 

aggregate

 command are shown on

page 128.

•

This syntax adds the new variables to the active dataset.

dataset activate hsbdemo.

aggregate

    /outfile = * mode = addvariables

    /break = cid

    /sumofread = sum(read)

    /minoftwrite = min(write)

    /maxofmath = max(math)

    /nuofscience = nu(science).

Creating variables:  The aggregate command

•

This syntax creates a new SPSS dataset.

aggregate outfile *

    /break cid

    /sumofread = sum(read)

    /minoftwrite = min(write)

    /maxofmath = max(math)

    /nuofscience = nu(science).

Creating variables:  The aggregate command

•

This syntax creates a new dataset that is saved to the location

specified.

aggregate outfile =

"D:\data\seminars\SPSS_syntax_2022\hsbdemonew.sav"

    /break cid

    /sumofread = sum(read)

    /minoftwrite = min(write)

    /maxofmath = max(math)

    /nuofscience = nu(science).

Creating variables:  The count command

•

The 

count

 command creates a new numeric variable that counts the

number of times a specified value occurs in the listed variables.

•

This syntax creates two new numeric variables called 

lowvalues

 and

numsysmisvalues

.

dataset activate hsbdemo.

count lowvalues = read write math science (lo thru 60)

    /numsysmisvalues = read write math science (sysmis).

Creating variables:  The shift values command

•

The 

shift values

 command creates new numeric variables that contain the

values of existing variables from preceding or subsequent cases.

•

Notice that lag = 1 is the same as shift = -1.

shift values variable = write result = newwrite lead = 1.

shift values variable = read result = newread lag = 1.

shift values variable = math result = newmath shift = 1.

shift values variable = science result = newscience lag = 2.

shift values variable = read result = newread1 shift = -1.

correlations variables = read newread newread1.

Creating variables: The rank command

•

The 

ran

k command creates new numeric variables that contain ranks, normal

scores or Savage scores.

•

You can use (a) and (d) to order variable in ascending or descending order.  All

variables preceding (a) or (d) will be ranked in that way.

•

The default name for the new variable is R+variable name.

rank variables = read write.

rank variables = math (a) science socst (d).

rank variables = read

    /normal into readnorm

    /ntiles(4) into readquart

    /rank into readrank

    /ties = mean.

BREAK TIME!!!

•

If you want to practice some of what we have just covered, here are a few

ideas of things to try:

1.

Create a small dataset that contains at least one string variable of length

5, one ordinal variable with 3 unique values, and any other variables that

you want.

2.

Create dummy variables for the ordinal in as many different ways as you

can.

3.

Create a new string variable that contains the values of the original string

variable starting at the second value and ending at the fourth value.

4.

Use your new favorite command to create a new numeric variable.

Modifying variables: Practice dataset

data list list

    /v1 (a1) v2 (a3) v3 (a3) v4 (a1).

begin data.

0 ab ab  *

1 cd cd &

2 ab ef !

3 cd de #

4 fg gh

end data.

dataset name modex.

list.

Modifying variables:

The autorecode command

•

The 

autorecode

 command recodes the values of string and numeric

variables to consecutive integers and puts the recoded into a new

numeric variable.  The value labels or the values of the orginial

variable are used as the value labels for the new variable.

•

The values of the new variable always start with 1.  If you want the

new variable to have values that start with 0, you need to create yet

another new variable by subtracting 1 from the variable that

autorecode

 created.

•

A variable cannot be recoded into itself; rather, a new variable must

be created.

Modifying variables:

The autorecode command

•

The subcommand order matters in this procedure!

•

The 

variables

 subcommand must be specified first.

•

The 

into

 subcommand must immediately follow the variables

subcommand.

•

All other subcommands can be specified in any order.

•

Notice the structure of the subcommands that create the new

variables.

Modifying variables:

The autorecode command

•

In this example, we are converting a string variable that contains only

numbers into a new numeric variable.

•

Notice that the value of the first case of the variable v1 is 0 but the

label is 1.  You may want to remove the value labels or alter them so

that the match the actual value of the variable.

autorecode variables = v1

    /into v1num

    /print.

Modifying variables:

The autorecode command

In this example, the 

group

 subcommand is used.

The 

group

 subcommand specifies that a single scheme should be

created for all of the specified variables so that a consistent coding

system is used for all of the variables.

autorecode variables = v2 v3

    /into v2num v3num

    /group

    /print.

Modifying variables: The numeric function

•

The 

numeric

 function can be used with the 

compute

 command to

create a new numeric variable from a string value.

•

Of course, this only works if the string variable contains only numeric

values.

•

A format for the new numeric variable must be given.

compute v1a = numeric(v1, f1).

exe.

Modifying variables: The recode command

•

In this example, we will use the 

convert

 option to recode a string

variable into a numeric variable.

•

Of course, this only works if the string variable contains only numeric

values.

•

Null strings (AKA blanks) in the string variable become system missing

in the new numeric variable.

recode v1 (convert) into v1num.

exe.

Modifying variables: The recode command

•

The 

recode

 command can be specified in a few different ways.

dataset activate hsbdemo.

recode ses (1 = 4) (2 = 5) (3 = 6).

recode prog, female (1 = 5) into progrec femalerec.

recode math to read (40 = 45)

/science socst (35 = 45).

exe.

Modifying variables: The recode command

•

In these examples, we will use SPSS keywords.

•

If this was syntax for a real work project, the recoded variables should

have values labels associated with them (i.e., the next command

should be value labels), but we will get to that later.

recode read (sysmis = -99) (lo thru 60 = 1) (61 thru hi = 2) (else = copy)

into readrec.

recode read (sysmis = -99) (lo thru 60 = 1) (else = copy) into readrec1.

freq var = read readrec readrec1.

Modifying variables: The string function

•

The 

string

 function can be used to populate a string variable with the

values of an existing numeric variable.

•

The string variable must already be in the active dataset.  The 

string

command can be used to do this.

•

The format of the numeric variable is the second argument given in

the 

string

 function.

dataset activate modex.

string v1string (a1).

compute v1string = string(v1num, f1).

exe.

Modifying variables:

The recode command with string variables

•

The 

recode

 command can also be used with string variables.

•

Note that all string values must be enclosed in quotation marks.

string v4new (a1).

recode v4 ('  ' = 'm') ('-' = '1') ('&' = 'a') into v4new.

list.

Modifying variables:  The valuelabel function

•

The 

valuelabel

 function is used to create a new string variable that

contains the value labels from another variable.

dataset activate hsbdemo.

string progstring (a8).

compute progstring = valuelabel(prog).

codebook progstring.

Modifying variables:  The alter type command

•

The 

alter type

 command can be used to make string variable numeric,

numeric variables string, to change the length of a string variable, or the

format of a variable.

•

The SPSS keywords 

to

 and 

all

 may be used.

•

In this example, the 

alter type

 command is used to change the length of

the variables 

v2

 and 

v3

.

dataset activate modex.

codebook v2 v3.

alter type v2 v3 (a4).

codebook v2 v3.

Modifying varibles:  The rename command

•

The 

rename

 command does exactly what you think it does:  it renames

variables in the active dataset.

•

One or more variables can be renamed with a single 

rename

 command.

•

The maximum length of a variable name is 64 characters.

dataset activate hsbdemo.

rename variables a1 = a1ren.

rename variables (a2 a3 = a2ren a3ren).

rename variables (a4 = a4ren) (a5 = a5ren).

exe.

BREAK TIME!!!!

•

If you want to practice some of what we have just covered, here are a

few ideas of things to try:

•

Convert a string variable to numeric using the 

autorecode

 command.

•

Recode a string variable using the 

recode

 command.

•

Create a new numeric variable using the function of your choice.

•

Use the 

alter type

 command to change the length of a string variable.

•

Rename two variables using the 

rename

 command.

Looping:  The do repeat command

•

The 

do repeat

 – 

end repeat

 commands is one set of commands that can be

used to create loops in SPSS.

•

The purpose of loops is to reduce the number of commands necessary to

complete a task by repeating the same transformations on the specified

variable.

•

The 

do repeat

 command uses a stand-in variable to represent a

replacement list of variables or values.

•

When the program repeats the transformation commands, the stand-in

variable is replaced by each variable or value in the replacement list.

•

The 

do repeat

 – 

end repeat

 commands do not read the active dataset;

rather, they are stored pending execution with the next command that

reads the dataset.

Looping:  The do repeat command

•

The following commands can be used with the 

do repeat

 command:

•

compute

, 

recode

, 

if

, 

count

, 

select if

•

vector

, 

string

, 

numeric

•

data list

, 

missing values

 (but not 

variable labels

 or 

value labels

)

•

loop

 (and 

end loop

), 

break

•

do if

, 

else if

, 

else

, and 

end if

•

formats

Looping:  The do repeat command

dataset activate hsbdemo.

do repeat a = prog1 prog2 prog3

    /b = 1 2 3.

compute a=(prog = b).

end repeat.

freq var = prog1 prog2 prog3.

compute ses1 = (ses = 1).

compute ses2 = (ses = 2).

compute ses3 = (ses = 3).

exe.

Looping:  The do repeat command

do repeat existvar = read to socst

    /newvar = new1 to new5

    /value = 1 to 5.

compute newvar = existvar + value.

end repeat print.

list var = read write math science socst new1 to new5.

Looping: The do if command

•

The 

do if

 – 

end if 

commands are used to do conditional

transformations on a subset of cases based on one or more logical

expressions.

•

The commands 

else

 and 

else if

 are available for additional control.

•

The 

do if

 – 

end if

 commands do not read the active dataset; rather,

they are stored pending execution with the next command that reads

the dataset.

Looping:  The do if command

do if female = 1.

compute newvar = 0.

else.

compute newvar = 1.

end if.

freq var = newvar.

Looping:  The do if command

do if prog = 1.

compute newvar2 = 0.

else if prog = 2.

compute newvar2 = 1.

else.

compute newvar2 = 2.

end if.

freq var = newvar2.

Looping: The loop command

•

The 

loop

 – 

end loop

 commands perform repeated transformations

specified by the commands within the loop until a specified cut off is

reached.

•

The cut off can be specified by an indexing clause, an if clause, or a

break

 command, or by the maximum number of iterations set by

mxloops. (The default number of loops is 40.)

•

The 

loop

 – 

end loop

 commands do not read the active dataset;

rather, they are stored pending execution with the next command

that reads the dataset.

Looping:  The loop command

•

The following example is taken from

https://www.spsstools.net/en/resources/spss-programming-book/

•

second edition of the book

•

page 144

•

Written by Raynald Levesque

Looping:  The loop command

**create sample data, 4 vars = 0.

DATA LIST FREE /var1 var2 var3 var4 var5.

BEGIN DATA

0 0 0 0 0

END DATA.

Looping:  The loop command

***Loops start here***.

*Loop that repeats until MXLOOPS value reached.

SET MXLOOPS=10.

LOOP.

- COMPUTE var1=var1+1.

END LOOP.

Looping:  The loop command

*Loop that repeats 9 times, based on indexing clause.

LOOP #I = 1 to 9.

- COMPUTE var2=var2+1.

END LOOP.

Looping:  The loop command

*Loop while condition not encountered.

LOOP IF (var3 < 8).

- COMPUTE var3=var3+1.

END LOOP.

Looping:  The loop command

*Loop until condition encountered.

LOOP.

- COMPUTE var4=var4+1.

END LOOP IF (var4 >= 7).

Looping:  The loop command

*Loop until BREAK condition.

LOOP.

- DO IF (var5 < 6).

- COMPUTE var5=var5+1.

- ELSE.

- BREAK.

- END IF.

END LOOP.

EXECUTE.

list.

Looping:  The loop command

dataset activate hsbdemo.

n of cases 20.

compute newvar1 = 10.

loop if newvar1 lt 100.

compute newvar1 = newvar1 + 10.

end loop.

list newvar1.

Looping:

The preserve and restore commands

•

The 

preserve

 command stores all current 

set

 specifications and the current

working directory setting.

•

The 

restore

 command reestablishes the 

set

 specifications and the working

directory that were in effect prior to when preserve was specified.

preserve.

set mxloops 100.

compute newvar2 = 10.

loop if newvar2 < 100.

compute newvar2 = newvar2 + 1.

end loop.

restore.

list newvar2.

Documenting data: The MOST important task

•

SPSS has many commands that allow you to document your data.

•

USE THEM!!!

•

The commands in this section do not read the active dataset; rather,

they are stored pending execution with the next command that reads

the dataset.

•

The 

sysfile info

, 

codebook

 and 

display

 commands allow you to view

the documentation of the data.

•

We will close and reopen the 

hsbdemo

 dataset.

Documenting data: The sysfile info command

•

The 

sysfile info

 command does not read the active dataset or execute

pending transformations.

•

The dataset does not even need to be open.

•

The 

sysfile info

 command gets the information from the saved dataset, so

it does not reflect changes made after the data were last saved.

dataset display.

* should have only hsbdemo and modex open.

get file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo.sav".

dataset name hsbdemo.

sysfile info file = "D:\data\seminars\SPSS_syntax_2022\hsbdemo.sav".

Documenting data:  The codebook command

•

The 

codebook

 command reports the dictionary information for the

active dataset.

•

The 

codebook

 command is a little different in that the square

brackets are necessary, as are the equal signs.

•

Without options, the 

codebook

 command will give the dictionary

information for the variables in the active dataset but not the file

information.

codebook.

codebook

 /fileinfo name location label documents casecount.

Documenting data: The file label command

•

The 

file label

 command adds a label to the active dataset.

file label High School and Beyond.

Documenting data:  The document command

•

The 

document

 command attaches a document to the dataset.

•

The documents can be viewed with the 

display documents

command.

•

Notice that this command does not require quotes around the text.

document Use this command to attach information about the data file

    to the data file itself.

display documents.

Documenting data:

The add document command

•

The 

add document

 command does require quotes around the text.

•

The text CANNOT wrap from one line to the next.

•

If you need to use multiple lines, end the quotes at the end of the line and

start new quotes at the beginning of the next line.

•

Blank lines can be added with a space between quotes.

add document

    'Adding information as necessary'

    ' '

    'even more information'.

display documents.

Documenting data:

The drop documents command

•

The 

drop documents

 command drops all documents associated with

the active dataset.

•

There are no subcommands for this command.

drop documents.

display documents.

•

We will use the 

document

 command to replace the document.

document Use this command to attach information about the data file

    to the data file itself.

Documenting data:

The datafile attribute command

•

The 

datafile attribute

 command allows you to define and assign attribute

values to the active dataset.

•

These attributes are saved with the dictionary information.

•

The 

datafile attribute

 command immediately updates the dictionary but

does not require a data pass.

•

Attributes can be deleted by using the keyword 

delete

 followed by an

equals sign and a list of defined attribute names.

datafile attribute attribute = originalversion ('1')

                       creationdate('08/14/2022')

                       revisiondate('08/20/2022').

display attributes.

Documenting data:

The variable labels command

•

The 

variable labels

 command assigns descriptive labels to variables.

•

It is not necessary to label all variables in a dataset.

•

The variable label must be enclosed in quotes and may contain

special characters, including blanks.

•

Variable labels can be up to 256 characters.

variable labels id "This is the individual ID variable".

variable labels cid "This is the classroom ID variable".

Documenting data:

The value labels command

•

The 

value labels

 command adds descriptive labels to the values of numeric

variables.

•

The maximum length of a value label is 120 characters.

•

Value labels can be added to values of both numeric and string variables.

•

It is not necessary to add labels to all values in a variable.

•

The 

value labels

 command deletes all previously assigned value labels to

the specified variables.

•

Value labels can be applied to multiple variables at once by listing the

variables or using the SPSS keyword 

to

.

•

If the value label is longer than the format of the variable, SPSS may not be

able to read the full value, and the value label may not be properly

assigned.

Documenting data:

The value labels command

•

Quickly creating variables to which to add value labels.

compute ses1 = (ses = 1).

compute ses2 = (ses = 2).

compute ses3 = (ses = 3).

value labels ses1 0 "not in ses level 1" 1 "in ses level 1"

    /ses2 0 "not in ses level 2" 1 "in ses level 2"

    /ses3 0 "not in ses level 3" 1 "in ses level 3".

codebook ses1 ses2 ses3.

Documenting data:

The add value labels command

•

The 

add value labels

 command is used to add value labels to

variables to which value labels have already been assigned.

if $casenum lt 20 ses3 = 4.

add value labels ses3 4 'this is an error'.

freq var = ses3.

codebook ses1 ses2 ses3

/varinfo valuelabels.

Documenting data:

The variable attribute command

•

The 

variable attribute

 command allows you to define and assign

attribute values to the active dataset.

•

These attributes are saved with the dictionary information.

•

The 

variable attribute

 command immediately updates the dictionary

but does not require a data pass.

•

Attributes can be deleted by using the keyword 

delete

 followed by an

equals sign and a list of defined attribute names.

•

After running the command, look at the Variable View (right side).

Documenting data:

The variable attribute command

variable attribute variables = female ses

    attribute = demographics ('multiple choice')

    /variables = read write math science socst

    attribute = tests('not multiple choice').

display attributes.

•

The values in square brackets indicate how the variable should be treated

when calculating the summary statistics.

codebook ses [o] write [s] science [s] socst [s]

 /varinfo position label type format measure valuelabels missing

 /fileinfo name location label documents casecount

 /statistics  percent mean stddev.

Variable display:

The variable alignment command

•

The 

variable alignment

 command specifies the alignment of variables

in the Data Editor.

•

It has no effect on the format of the variables or the display of the

variables or values in other windows or printed results.

variable alignment ses female (left)

    /id prog (right).

Variable display: The variable level command

•

The 

variable level

 command specifies the level of measurement of

variables in the active dataset.

•

There are three possible levels: nominal, ordinal and scale.

variable level ses (scale).

variable level cid id (nominal).

Variable display: The variable role command

•

Some dialogs support predefined roles that can be used to pre-select

variables for analysis.

•

Possible roles include: input, target, both, none, partition, split.

variable role

    /input ses female prog

    /target read write

    /both math science

    /none cid id.

Variable display: The variable width command

•

The 

variable width

 command specifies the column width for the

display of variables in the Data Editor.

•

It has no effect on the format of the variable or the display of the

variable or values in other windows or printed results.

variable width cid id (5).

Variable display: The formats command

•

The 

formats

 command changes the format of numeric and string

variables.

•

It does not affect the value of the variable; numeric variables may

have more decimal values than displayed.

•

The default formats for both numeric and string variables can be

modified with the 

set format

 and 

set decimal

 commands.

•

f, w, n, and e are followed by a number indicating the total width,

decimal, and another number indicating the number of decimal

places (see page 49).

Variable display: The formats command

•

There are dollar, comma, dot and pct formats.

•

dollar format:  $1,234.00.

•

comma format:  1,234.00.

•

dot format: 1.234,00.

•

percent format:  1234.00%.

•

There are other formats that were commonly used in the past (e.g.,

used with COBOL and Fortran), but we won't cover those here

because they are so rarely used.

Variable display: The formats command

formats write (f2.0).

list var = write

/cases = from 1 to 5.

format write (f4.2).

list var = write

/cases = from 1 to 5.

format write (f6.4).

list var = write

/cases = from 1 to 5.

Missing values

•

There are two types of missing data for numeric variables:  system

missing and user-defined missing.

•

System missing values are the lowest possible number in SPSS.

•

Many values can be defined as user-defined missing.

•

SPSS keywords such as 

lo

 (or 

lowest

), 

thru

 and 

hi

 (or 

highest

) may be

used to specify user-defined missing values.

•

To remove missing values, leave the parentheses empty.

•

The SPSS keyword 

all

 may be used to set user-defined missing values

for all variables in the active dataset.

Missing values: The missing values command

•

The 

missing values

 command assigns user-defined missing values to

variables in the active dataset.

•

The 

missing values

 command can be used with both numeric and

string variables.

•

Up to three discrete values, or two ranges and one discrete value may

be assigned to a variable.

•

To remove user-defined missing values from a variable, issue the

missing values

 command with the parentheses empty.

Missing values: The missing values command

dataset activate modex.

missing values v1num (0) v2num v3num (-99) v2 ("fg").

display dictionary.

•

Removing the user-defined missing value from 

v1num

 and changing

 the user-defined missing  values for 

v2num

 and 

v3num

.

missing values v1num () v2num v3num (lo thru 2).

display dictionary.

Missing values:  Missing values functions

•

The 

missing

 function returns a 1 if the value is either user-defined or

system missing.

•

The 

sysmis

 function returns a 1 if the value is system missing.

•

The 

nmiss

 function returns a count of the number of arguments that

have either user-defined or system missing values.  The argument(s)

should be one or more variables in the active dataset.

•

The 

nvalid

 function returns a count of the number of arguments that

have valid, non-missing values.  The argument(s) should be one or

more variables in the active dataset.

Missing values:  Missing values functions

•

Create a little example dataset.

data list list

/id (f2.0) v1 (f2.0) v2 (f2.0) stringvar1 (a5).

begin data.

12 63  80     abcde

16 98  .

22 .      55      fdsaq

55 .      .      poiuy

79 33    12     nhytg

end data.

list.

Missing values:  Missing values functions

* returns a 1 if the value is either user-defined or system missing.

compute v1miss = missing(v1).

* returns a 1 if the value is system missing.

compute v1sysmis = sysmis(v1).

* returns a count of the number of arguments that have either user-defined

or system missing values.

compute v1mniss = nmiss(v1).

* returns a count of the number of arguments that have valid, non-missing

values.

compute v1nvld = nvalid(v1).

list.

Missing values:  Missing values functions

compute sv1miss = missing(stringvar1).

list.

•

Notice the difference between this output and the previous output.

missing values stringvar1 (" ").

compute sv1miss1 = missing(stringvar1).

list.

Missing values:

Missing values functions with “and” and “or”

•

When two relations are joined with the AND operator, the logical

expression can never be true if one of the relations is indeterminate.

The expression can, however, be false.

•

When two relations are joined with the OR operator, the logical

expression can never be false if one relation returns missing. The

expression, however, can be true.

if missing(v1) and missing(v2) v1v2missand =1.

if missing(v1) or missing(v2) v1v2missor = 2.

list.

Saving data: The cd command

•

The 

cd

 command changes the working directory.

cd "D:\data\seminars\SPSS_syntax_2022".

Saving data: The save command

•

The 

save

 command saves the active dataset to the specified location

as an SPSS dataset.

•

The 

save

 command reads the active dataset and causes any pending

transformations to be executed.

•

Files can be saved in compressed form and/or with a password.

•

Variables can be kept, deleted and/or renamed.

save outfile = "D:\data\seminars\SPSS_syntax_2022\newdata.sav".

Saving data:  The save translate command

•

The 

save translate

 command saves the active dataset to the location

specified in a format other than an SPSS data file.

•

Variables subsets can be saved using the 

keep

 or 

drop

 subcommands.

•

Variable names can be changed using the 

rename

 subcommand.

•

Value labels can be saved to Excel or tab-delimited files rather than

numeric values using the 

cells

 subcommand.

Saving data:  The save translate command

•

Saving the file as a SAS file.

•

Usually do not want to use short extensions; use the full sas7bdat.

dataset activate hsbdemo.

save translate outfile =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo1.sas7bdat"

    /type = sas

    /version = 9

    /platform = windows

    /replace.

Saving data:  The save translate command

•

Saving the file as a Stata data file.

save translate outfile =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo1.dta"

    /type = Stata

    /edition = intercooled

    /replace.

Saving data:  The save translate command

•

Saving the file as an Excel file.

save translate outfile =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo1.xlsx"

    /type = xls

    /version = 12

    /replace.

Saving data:  The save translate command

•

Saving the file as a CSV file.

save translate outfile =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo1_csv.dat"

    /type = csv

    /replace.

Saving data:  The save translate command

•

Saving the file as a tab-delimited file.

save translate outfile =

"D:\data\seminars\SPSS_syntax_2022\hsbdemo1_tab.dat"

    /type = tab

    /replace.

Saving data:  The erase command

•

The 

erase

 command deletes files from the location specified.

erase file = "my/new/directory/trash.sav".

Output to data: The OMS commands

•

OMS stands for Output Management System, and it is a method of

creating a dataset from the output.  It routes the output and can be

used to suppress Viewer output.  Output formats include, Word,

Excel, PDF, and SPSS data files (sav files), viewer file format (spv), xml,

html and text.

•

When reading the SPSS command syntax reference for OMS entries,

the square brackets are necessary and do not indicate options.  All

equals signs shown in the syntax are required.

Output to data: The OMS commands

•

The following are the oms commands:.

•

oms

:  Begins an oms session and remains in effect until the oms end

command is encountered.

•

omsend

:  ends the oms session.

•

omsinfo

: displays a table of all active oms commands

•

omslog

: creates a log file for subsequent OMS commands during a

session.

Output to data: The OMS commands

•

Opening an oms log file (as a text file).

dataset activate hsbdemo.

omslog file = "D:\data\seminars\SPSS_syntax_2022\omslog.txt"

    /append = yes

    /format = text.

Output to data: The OMS commands

•

Capturing output from the 

crosstabs

 command and saving the output to an

SPSS data file.

oms  select tables

 /destination format = sav outfile =

"D:\data\seminars\SPSS_syntax_2022\results.sav"

 /if commands = ['crosstabs'] subtypes = ['Crosstabulation'].

crosstabs tables = female by prog.

omsend.

get file "D:\data\seminars\SPSS_syntax_2022\results.sav".

list.

Output to data: The OMS commands

•

Capturing output from two different commands, 

regression

 and 

correlations

, and saving

them to two different files, 

results1.sav

 and 

results2.sav

, respectively.

dataset activate hsbdemo.

oms select tables

 /destination format = sav numbered = "Table_Number" outfile =

"D:\data\seminars\SPSS_syntax_2022\results1.sav"

 /if commands = ['regression'] subtypes = ['Coefficients']

 /tag = "reg".

oms select tables

 /destination format = sav outfile = "D:\data\seminars\SPSS_syntax_2022\results2.sav"

 /if commands = ['Correlations'] subtypes = ['Correlations']

 /tag = "cor".

Output to data: The OMS commands

regression

 dependent = write

 /method = enter female read.

correlations

 /variables = write read math female.

omsend tag = ["cor"].

regression

 dependent = write

 /method = enter female math.

Output to data: The OMS commands

regression

 dependent = write

 /method = enter female read math.

omsinfo.

omsend tag = ["reg"].

get file "D:\data\seminars\SPSS_syntax_2022\results1.sav".

list.

get file "D:\data\seminars\SPSS_syntax_2022\results2.sav".

list.

Finishing up: The show command

•

The 

show

 command displays the current settings for running options.

•

Most of these can be changed using the 

set

 command.

show license.

show seed.

show directory version locale format.

Finishing up: Extension commands

•

Extension commands are community-contributed commands that

extend the functionality of SPSS.

•

Click on Extensions -> Extension Hub

•

The extension commands use either Python or R

•

The version of Python needed to run the extension commands installs

with SPSS.

•

To use extensions that use R, you need to install the version of R

corresponding to your installation of SPSS and then install the R

bridge.

Great resources

•

Raynald Levesque

•

Marija J. Norusis

•

Andy Field.

THANK YOU!!!

•

Thank you for attending this workshop!

•

Questions???