Using SPSS to Understand Research and Data Analysis.

It is often desirable to transform continuous variables (variables whose scores are on a quantitative continuum, such as from 1 to 7) into dichotomous variables (variables whose scores represent two different categories). Researchers sometimes do this in order to be able to treat a continuous person variable (e.g., degree of masculinity, as measured in masctot) as a quasi-experimental independent variable consisting of two groups (e.g., low vs. high masculinity, as we have done with masc).

Thus, we transformed the scores on the continuous variable, masctot, (which range from 7 to 35) into a new dichotomous variable, masc, consisting of just two categories and two score values (1: low masculinity; 2: high masculinity). Once this is done, we can treat this new masc variable as an independent variable to examine differences in leader performance between employees in the high-masculinity category compared to those in the low-masculinity group. Although this may sound complicated, the process is relatively simple using the Recode procedure.

• 5.2b The Recode Procedure

To create this new variable, we used the following recoding rules:

• If an employee's masctot score is 17 or lower, then that score will be transformed to a value of 1 on the new masc variable (i.e., s/he will be placed in the low masculinity category)
• If an employee's masctot score is 18 or higher, then that score will be transformed a value of 2 on the new masc variable (i.e., s/he will be placed in the high masculinity category).

To produce this transformation, we selected Transform, Recode, Into Different Variables from the menu in the Data Editor (Figure 5.7).

This caused a Recode into Different Variables dialog window to appear (Figure 5.8). We scrolled down the left pane and highlighted masctot, then clicked the right arrow button between panes to move this variable to the Input Variable - > Output Variable: pane.

SPSS now expects the user to name the new output variable, as can be seen by the masctot - > ? question that appears in the Numeric Variable - > Output Variable: pane (Figure 5.9). We named the new output variable by typing masc into the Output Variable box on the right, then clicked the Change button to complete the naming.

The new variable name (masc) was now displayed in the Numeric Variable - > Output Variable: pane. Next, to specify the recode rules to be used in creating the new output variable, we clicked the Old and New Values button (Figure 5.10).

This caused yet another dialog window to appear (Recode into Different Variables: Old and New Values). As shown in Figure 5.11, we specified Range Lowest through 17 in the Old Value pane, and specified the New Value as 1 on the right. This instructs SPSS to recode any employee's masctot score that is lower than 18 into the new value of 1 on the new masc variable (the low-masculinity category). Last, we clicked the Add button to complete this variable transformation.

As shown in Figure 5.12, we then specified Range: 18 through highest in the Old Value pane, and specified the New Value as 2 on the right. This instructs SPSS to recode any employee's masctot score that is higher than 18 into the new value of 2 on the new masc variable (the high-masculinity category). Again we clicked the Add button to complete this variable transformation.

The rules we just specified are now summarized in the Old -- > New: pane on the lower right (Figure 5.13). Next, we clicked the Continue button at the bottom of this dialog window.

This caused the Old and New Values dialog window to close. For the last step, we clicked the OK button at the bottom of the Recode into Different Variables: dialog window (Figure 5.14).

SPSS then performed this transformation and added the new recoded masc variable at the end of the data file (recall that we then cut and pasted this new variable so that it now appears as the second variable in the ezdata.sav file you downloaded). If you scroll down your Data Editor window, you will see that all the values of this new variable are either 1 (low-masculinity) or 2 (high masculinity).

To view a video of this procedure, click Play below.

We used exactly the same Recode procedure to create the new fem variable, using the same recoding rules. That is:

• If the femtot score was 17 or less, then the fem score = 1 (Low-Femininity)
• If the femtot score was 18 or higher, then the fem score = 2 (High-Femininity)

Last, we used the Recode procedure to create the new categorical variables, soc and task, by recoding the original soc1 and task1 variables. Since soc1 and task1 are variables with score ranges from 1 to 9, we used a different set of recoding rules:

• If the soc1 score was 4 or less, then the soc score = 1 (Low Social Skills)
• If the soc1 score was 5 or higher, then the soc score = 2 (High Social Skills)
• If the task1 score was 4 or less, then the task score = 1 (Low Task Skills)
• If the task1 score was 5 or higher, then the task score = 2 (High Task Skills)

You're probably glad that we performed all of these transformations ourselves in the ezdata.sav file that you downloaded! It might also seem like a lot of variables to keep straight, but you will become more familiar with them once we begin doing data analyses involving these variables.

Further, although the Compute and Recode procedures we have described can seem confusing at first (and tedious to perform), we will give you a chance do some simpler transformations at the end of this chapter. Once you have tried them, you should see that the procedures themselves are fairly straightforward.

Before we do this, in the last section of this chapter we describe one more housekeeping procedure that makes the data file (and outputs from analyses) much easier to understand: providing more descriptive names for variables and adding labels to the levels of categorical variables.