· Correspondence analysis decomposes the association in a contingency table into several orthogonal dimensions (chi-square distance): row points and column points that lie close on the same plane are strongly associated. Total inertia = χ²/N; the explained share is the cumulative proportion of per-dimension inertia. Row and column scales are intrinsically comparable; interpret mainly by proximity.

· For cross-tabs of categorical variables (e.g. preference × brand, complaints × machine model).