ABSTRACT
The total inertia of a cross-tabulation is a measure of how much variation there is in the table. We have seen how this inertia is decomposed along principal axes and also how it is decomposed amongst the rows or amongst the columns. The inertia can be further broken down into row and column components along individual principal axes. The investigation of these parts of inertia, analogous to an analysis of variance, plays an important supporting role in the interpretation of correspondence analysis (CA). They provide diagnostics which allow the user to identify which points are the major contributors to a principal axis and to assess how well individual points are displayed.
