ABSTRACT
Up to now we have analysed the association between two categorical variables or between two sets of categorical variables where the row variables have a different substantive “status” compared to the column variables, e.g. demographics versus survey questions. In this and the following two chapters we turn our attention to the association within one set of variables of similar status, where interest is in how strongly and in which way these variables are interrelated. In the present chapter we will concentrate on the two classic ways to approach this problem, called multiple correspondence analysis (MCA). One way is to think of MCA as the analysis of the whole data set coded in the form of dummy variables, called the indicator matrix , while the other way is to think of it as analysing all two-way cross-tabulations amongst the variables. These two ways are very closely connected, but suffer from some deficiencies that will be resolved in Chapter 19, where some improved versions of MCA are presented.
