What does the profile plot tell us about changes over time?.
In this exercise, we will test racial/ethnic differences in depression over time, using CES-D scores from the two waves of interviews with a subsample of these women. You will need to begin by excluding women in the white/other group because there were too few of them in this small subsample to permit their inclusion. Go to Data ➜ Select Cases and click Select “If condition is satisfied” in the opening dialog box. Click the If pushbutton, and then type the following into the box to exclude whites/others, who are coded 3: “racethn NE 3.” Click Continue, then OK to restrict the analysis to African-American and Hispanic women. Now run the main analysis, using Analyze ➜ General Linear Model ➜ Repeated. You will first be asked to give the within-subjects factor a name. We used “Wave” to designate Wave 1 or Wave 2 measurement of depression. The number of levels to enter, in the next box, is 2 (i.e., two waves). Click Add then go to the bottom, where you can name the dependent variable in a slot labeled Measure Name. Enter Depression, then click Add. Now click the Define pushbutton, which brings up a dialog box for defining variables. Select cesdwav1 and click the right arrow to move this variable into the list as the first Within-Subjects Variable. Then select cesd (Wave 2 scores) and move it into the list as the second Within-Subjects Variable. Next, move racethn into the slot for the Between-Subjects Factor. Now click the pushbutton Plots and in the dialog box that appears move wave into the Horizontal Axis box and racethn into the Separate Lines box. Click Add, then Continue. Back on the original dialog box, click Options and at the bottom of the next dialog box select the following Display options: Descriptives, Estimates of effect size, and Homogeneity tests. After clicking Continue and OK to run the analysis, answer the following questions: (a) What are the null hypotheses in this analysis? (b) How many women were in this analysis, by race? (c) Was the assumption of homogeneity of the variance–covariance matrix upheld? What are the implications for this analysis? (d) Was the assumption of sphericity upheld? What are the implications for this analysis? (e) Was the assumption of homogeneity of variances upheld? What are the implications for this analysis? (f) What are the results for the null hypotheses being tested in this RM-ANOVA, as identified in question a? (g) Should post hoc tests be run for this analysis? Why or why not? (h) What does the profile plot tell us about changes over time?