Doing More With Your Variables: Creating New Variables and Organizing Cases

Doing More With Your Variables: Creating New Variables and Organizing Cases

Creating New Variables:

You may decide that some variables in your data file need modifications before you do certain analyses. For example, you may want to: (1) compute a new variable from other variables in the data set (e.g., create a sum); and/or (2) change the way that the variable was coded (e.g., make age groups, collapse certain categories).

To do these analyses, make sure you are in the data sheet window (e.g., module1.sav).

To Create New Variables From Old Variables:

Compute. SPSS provides over 100 statistical functions to transform variables. Most commonly, you may want to use a SUM function to make a composite of a number of variables in your data set. You can then use this newly computed variables in subsequent analyses.

To use the compute command, click Transform à Compute. à Compute.

Once you are in the Compute window, it is up to you: (1) to state how your new variable will be named (Target Variable; remember that computer variable names need to be eight characters or less); (2) to choose the variables that will be used; and (3) to select one of the functions such as SUM for your Numeric Expression.

Count: A second way to create a composite of other variables in the data set is to click Transform à Count. à Count. The count command will evaluate a set of variables that you choose and give one point for every instance of a value or a set of values (that you specify).
First provide a computer name (eight characters or less) and a longer name (Target Label) for the new variable that you want to create.
Next, choose your variables from the list and use the black arrow key in middle to move the variables to the Numeric Variables box.
Finally, click Define Values to specify the values you want, Continue (to bring you back to the Count menu), and then OK to tell SPSS that it is time to create the new variable.
To Change The Coding Of Some Original Variables:
Recode: You can recode string variables (e.g., Y/N) as easily as numeric variables.
To use the recode command, click Transform à Recode à Into Different Variables. à Recode à Into Different Variables.
Choose the variable(s) you would like to recode and the new name(s) and label(s). Click Change to register changes.
Click Old and New Values to start specifying how you want the new variable to be coded.
For each value of the existing variable ("Old Value") you will need to state how you want the new variable to be coded ("New Value". For each value specified, click Add to register the proposed change. When you are through, click Continue (which will bring you back to the Recode menu), and OK (to process the command).

Go to the far right of the data sheet and see that your new variable is there.

Organizing Cases:

You also are likely to find yourself wanting to rearrange your data somewhat before you do analyses. For example, you may want to (1) sort your data on a particular variable; and (2) only analyze a subset of the cases in your data file.
To do these analyses, make sure you are in the data sheet window (e.g., module1.sav).
Sort. There are many reasons why you might want to sort on a variable in your data set including for data checking.
To do this, click Data à Sort à Sort Cases and choose the variable(s) on which you would like to sort (put in the Sort By box). You can choose ascending or descending order for the sort.

When you say OK, SPSS will process your command and you can look at your data sheet to see the results of the sort.

Selecting Cases: You may not want to do all analyses on all cases at all times, and this procedure allows you to choose subsets of cases.

There are a number of steps to selecting cases:

Click Data à Select Cases à If condition is satisfied à If. à Select Cases à If condition is satisfied à If.
Choose the variable(s) from the list and specify (as a formula) the condition of selection. For example, if you would like to analyze only women, you would type sex = 2, assuming that women are coded as 2 for the sex variable.
Click Continue, and OK in the select cases window.
Do whatever analyses you want (they will be conducted on only the women in the sample).
When you want to use the whole sample again, click Data à Select Cases à All Cases à OK. à Select Cases à All Cases à OK.

Back: Using Your Module Data: An Intermediate-Advanced Training Session Using SPSS For Windows

Related Information:

The Measurement Group Evaluation & Research Tools
The Measurement Group Data Mining and Statistical Modeling