Chapter 1: Recoding Variables
This exercise illustrates recoding of reverse-scored variables, and correlation. It uses the PERS dataset, consisting of 90 cases and 968 variables. The variables represent measures of traits and relevant behaviors for the dimensions of extraversion (outgoingness) and conscientiousness, reported each week for three weeks by a group of undergraduate psychology students.
In the codebook for the PERS dataset, examine the "Recoding suggestions for reverse-scored items." Note that some weekly conscientiousness behavior items actually indicate a lack of conscientiousness (are reverse-scored). Look at the actual questionnaire items for these weekly conscientiousness items (you can jump to this from the place in the codebook that lists these weekly conscientiousness behaviors), and see how this is the case. While most of the items are worded so that a higher score indicates greater conscientiousness, these reverse-scored items are clearly worded so that a higher score indicates less conscientiousness. The score for all of the weekly behavior items is 0 to 7, indicating the number of days in the past week that the behavior applied.
Get correlations between the first five of these variables:
For variables, select WBC1, WBC2, WBC3, WBC4, WBC5. Make sure you are getting Pearson correlations.
Notice that WBC1 (neglected to prepare for upcoming class discussion) correlates positively (r=.434) with WBC2 (took a night off from my studies), since both of these measures indicate a lack of conscientiousness. However, WBC1 correlates negatively with WBC3 (studied until I completed all of my work), since these two variables are scored oppositely in what they indicate about conscientiousness. To examine the relationships between several variables, it is usually easier (but sometimes totally necessary) to have the variables all scored in the same direction, for example, such that a high score indicates higher conscientiousness.
Following the suggestions in the codebook for "Recoding reverse-scored items," try recoding WBC1, WBC2, and WBC5 using the SPSS Compute function:
For the Target Variable, type rwbc1
For the Numeric Expression, type 7-wbc1
Click on OK; a new variable should be created, rwbc1, which is the reverse of wbc1.
Repeat the above steps for wbc2 (name it rwbc2) and wbc5 (name it rwbc5)
Now rerun the correlations, using the recoded variables. The variables to use for the correlations are rwbc1, rwbc2, wbc3, wbc4, rwbc5. You should get the same magnitude correlations as before, but all positives in sign. For example, the correlation between rwbc1 and wbc3 is now .375.
other correlations involving reverse-scored items, then recode them according
to the recode suggestions in the codebook, and compute
the correlations again.