Preparation of Quantitative Data

Recoding Variables (SPSS or PSPP)

The recoding of variables is one of the most important and most common steps in data preparation. You can also do this via the user interface, but it is much more practical and easier with SPSS command syntax. When you open the syntax editor (File - New - Syntax), you can enter your commands there. You can also generate SPSS syntax by clicking on "Insert" from the context menu instead of "OK". This does not execute the respective command, but writes it to a syntax file, which you can then execute directly or edit first.

Each SPSS command must be terminated with a full stop, otherwise you will receive an error message. After recoding or variable calculation, the execution command "EXECUTE." is also required. To execute a command or command block, you only have to select the corresponding commands (including the terminating EXECUTE.) and click on the green arrow at the top of the syntax editor (or Ctrl+R). Traditionally (as here for illustrative purposes) the commands are written in upper case and the variable names in lower case, but unlike e.g. R, upper/lower case does not matter in SPSS syntax. You can also set line breaks (no blank lines) as you wish.

The two most frequent applications of the RECODE command are the repoling of questionnaire items and the merging of value ranges into categories. Important practical tip: When doing this, always create a new variable and leave the original variable untouched, otherwise you can, in the worst case, throw away your data set or re-enter/read everything.

If, for example, you want to repolarise several items scaled from 1 to 5 (here called "question1 question2 question3" as an example) for the later calculation of a scale mean, for example because they were formulated negatively and/or belong to be recoded according to the template, the command would look like this:

RECODE

question1 question2 question3

(1=5) (2=4) (3=COPY) (4=2) (5=1) INTO

question1r question2r question3r.

EXECUTE.

SPSS then adds three new variables to your dataset at the end called question1r, question2r and question3r, which are the inverted variants of the original items according to the assignment above: 1 becomes 5, 2 becomes 4 etc.. COPY means that the old value is simply taken over; you could thus also write "(3=3)", amounts to the same thing. If the scaling went from 1 to 6, the recoding rule would clearly be (1=6) (2=5) (3=4) (4=3) (5=2) (6=1) etc.. Instead of "r" (for recoded) you can of course also take another letter ("r" for recoded or whatever).

A second important use of the RECODE command is to group numerical variables retrospectively. In the survey (cf: Do you have questions about the sample or survey methods?, you should aim for the highest possible level of measurement (i.e. do not ask from the outset for e.g. age groups instead of age in years), but if you want to form age groups afterwards, you can do this analogously to the following template. The original variable here is simply called "age", the new variable "altgrup", again a) by all means define a new variable, b) you can call it whatever you like.

Age will mostly be asked in whole years, but in the following example the following age class already starts at the hundredths place, mainly to point out that in SPSS syntax the decimal point is a dot and not a comma (English notation). "Lowest" and "Highest" are placeholders for the smallest and largest value, respectively, contained in the data set for this variable; thus, you do not need to look up beforehand where minimum and maximum lie (or define absurdly small/large minimum and maximum values). Also included are the commands for labelling the values (VALUE LABELS); you should always do this with categorical variables.

RECODE age

(Lowest thru 30 = 1) (30.01 thru 40 = 2) (40.01 thru 50 = 3) (50.01 thru

Highest = 4)

INTO altgrup.

VALUE LABELS altgrup

1 "up to 30 years" 2 "31 to 40 years" 3 "41 to 50 years" 4 "over 50 years".

years".

EXECUTE.

NOTE: Find answers to Frequently Asked Questions here: FAQ