11.4. Mosaic plots

Up to this point, we’ve been exploring methods of visualizing relationships among quantitative/continuous variables. But what if your variables are categorical? When you’re looking at a single categorical variable, you can use a bar or pie chart. If there are two categorical variables, you can look at a 3D bar chart (which, by the way, is not so easy to do in R). But what do you do if there are more than two categorical variables?

One approach is to use mosaic plots. In a mosaic plot, the frequencies in a multidimensional contingency table are represented by nested rectangular regions that are proportional to their cell frequency. Color and or shading can be used to represent residuals from a fitted model. For details, see Meyer, Zeileis and Hornick (2006), or Michael Friendly’s Statistical Graphics page (http://datavis.ca). Steve Simon has created a good conceptual tutorial on how mosaic plots are created, available at http://www.childrensmercy.org/stats/definitions/mosaic.htm.