Mediation, Moderation, and Canonical Correlation

This chapter introduces some complex associational methods that are not covered in earlier chapters: mediation, moderation, and canonical correlation. These analyses use correlation and regression to assess specific types of relationships. They build on correlation and multiple regression, but they enable more complex inferences about the relationships among multiple variables. Previous chapters taught readers to perform and interpret three types of complex associational statistics (multiple regression, principal components analysis, and factor analysis). This chapter uses multiple regression again, but it helps readers learn how to test for statistical mediation and moderation and to interpret the results of those analyses. In addition, the chapter teaches them how to conduct and interpret canonical correlation, which enables better understanding and simplification of the relations between two sets of variables. Canonical correlation creates components that maximize the relation between the sets of variables and correlate those components, providing results similar to correlations between two sets of components from a principal components analysis.