ABSTRACT
In all the chapters up to now we have dealt exclusively with categorical data and frequency tables, either a single table or in sets. In this chapter we will look at other types of data and how they can be recoded, or transformed, in such a way that correspondence analysis (CA) can still be applied as a method of visualization. This strategy is particularly well developed in Benze´cri’s approach to data analysis, where CA is the central algorithm and different data types are preprocessed before being analysed. The types of data treated here are ratings, preferences, paired comparisons and data on continuous scales. In all of these cases the original CA paradigm should be remembered: CA analyses count data, so if we can transform other types of data to counts of some kind, then it is likely that CA will be appropriate. A standard checklist to perform on the recoded data will be to see if the basic concepts of profile, mass and χ2-distance make sense in the context of the data.
