Coding categorical data for simple correspondence analysis

Simple correspondence analysis is a very popular exploratory tool used to graphically identify the association between two categorical variables. It is generally performed by applying singular value decomposition on the standardised cell values of the contingency table. Analysing the counts summarised in the contingency table is the most common way to perform correspondence analysis, although categorical data can be coded in different ways. By applying simple correspondence analysis to this coded data we can also obtain a graphical description of the relationship between the variables. The benefit of such coding procedures is that they provide a natural extension for the analysis of multivariate categorical data. This paper will explore several ways that simple correspondence analysis can be performed by coding categorical data. Such a review supplements an extensive discussion recently made by Beh (2004) on a variety of issues concerned with the correspondence analysis of two categorical variables