ABSTRACT
In Chapters 16 and 17 the analysis of concatenated tables was considered, either when two categorical variables were combined interactively (e.g. country and gender in Exhibit 16.6), or several variables were stacked (e.g. Exhibit 17.1). It often occurs that we want to compare two tables of the same size, with the same row and column entities, in order to understand their similarities and differences. The gender comparison is a classic example, where we split a cross-tabulation into two tables, one for males and another for females. But we could be comparing other groups such as Western versus Eastern Europe, urban versus rural, treatment versus control, or first time period versus second time period. Two such binary factors could also classify the data, giving four tables, e.g. males versus females in Western Europe, to be compared with males versus females in Eastern Europe. Correspondence analysis (CA) can be used to display variation common to these tables, and variation accounting for their differences.
