I am interested in exploring how different characteristics of national pension systems are related to each other. I have used MCA for a dataset in which the rows are countries and the columns are different features of pension systems. However, I am not sure how to interpret the distances between points in the SPSS Joint Plot of Category Points. Using a symmetrical normalization, do the distances between points representing categories of different variables say something about how these categories are associated? Does a shorter distance mean a higher level of association?
-
$\begingroup$ If you have only rows and columns (a 2-way table), then it is called simple correspondence analysis, not multiple $\endgroup$– ttnphnsSep 7, 2012 at 16:33
1 Answer
The standard visualisation is the biplot. The interpretation depends on the details of the technique applied but will usually lean on some notion of inner product. But since I don't know what SPSS does when you ask for MCA then I hesitate to offer more concrete advice. Nevertheless you'll surely find all you need to interpret them in the (free) book Biplots in Practice, specifically chapters 9-10.
However, if you're wondering how to interpret its output then you might profitably first revise your theory of correspondence analysis. Greenacre's CA in Practice is a good applied text. Ch. 9 covers biplots and ch. 16-20 revise the multi-way extensions of simple correspondence analysis (they are short chapters). That should provide enough background to see what SPSS is offering you.
As @ttnphns points out, a two way table implies simple rather than multiple correspondence analysis. Then things are indeed easier (but still see references above).