ABSTRACT
This chapter shows how the principles of correspondence analysis (CA) can easily be generalized to studies of multivariate data sets, and how some statistical properties change from the analysis of a contingency table to the analysis of a binary indicator matrix. The fundamental principles in an Multiple correspondence analysis (MCA) are the same as in a CA, and even though some properties change, or must be interpreted differently, it is a straightforward procedure to expand the analysis from 2 to N variables. In an MCA, one is no longer confined to analyzing a contingency table, or to do a CA, a stacked table. The two most commonly used matrices in an MCA are the Burt-matrix and the binary indicator matrix (BIM). One of the major advantages of doing an MCA on a binary indicator matrix is that the BIM gives direct access to the cloud of individuals.
