The research reported in Chapters 9 and 10 aimed at demonstrating the relationship between the cognitive constructs in the S/RT model and depressive symptoms and problem behaviors (Chapter 9), and the cognitive constructs and suicide and risk-taking behaviors (Chapter 10). The research used between-group comparisons to demonstrate the relationships. The data analyses reported in this chapter directly test whether the nine cognitive variables mediate between depressive symptomatology and problem behaviors on the one hand and suicide and risk-taking behavior on the other. It uses path analysis to do this.

An advantage of the group comparison approach is that, by looking at the mean scores of the groups across the nine cognitive variables and the screening variables, we can make a cross-study comparison. There were three groups in the first set of analyses (Chapter 9) and four groups in the second set (Chapter 10). There was a comparison group in each study. The results from these two comparison groups can be pooled since a discriminant analysis found they were not significantly different from each other.

Path analysis (see Figure 11.1) is an extension of multiple regression that allows one to simultaneously assess the relative importance of two or more independent variables while testing for the influence of a mediating variable.

All the paths, direct and indirect, can be tested for statistical significance. Structural equation model programs such as EQS (Chou & Bentler, 1995) perform this analysis and test the paths for significance.