Multiple Correspondence Analysis

This chapter describes some common approaches that can be used for simultaneously visualising the association between multiple categorical variables by focusing on the analysis of only three variables. It confines the application of multiple correspondence analysis to the visual summary of the association between three categorical variables, although it can be applied to a much larger sized table without loss of generality. The chapter also describes three coding methods and how they can be used to perform multiple correspondence analysis. The coding methods covered are crisp coding, Burt matrix, and stacking. Multiple correspondence analysis does not truly describe the underlying multivariate association structure between the variables of a multi-way contingency table. Instead, since it involves some form of bivariate transformation of the original contingency table, they are at best a way of visualising the various bivariate association structures that exist.