Categorical Data Analysis with Graphics

Statistical methods for categorical data, such as log-linear models and logistic regression, represent discrete analogs of the analysis of variance and regression methods for continuous response variables.

While graphical display techniques are common adjuncts to analysis of variance and regression, methods for plotting contingency table data are not as widely used. Moreover, while statistical methods can be used to determine which variables are related, the numerical summaries and parameter estimates do not provide easy ways to show how those variables are related.

This workshop provides a brief introduction to statistical methods for analysing discrete data and frequency data, together with some of the graphical methods which are useful for understanding the pattern of association among categorical variables. These methods can be helpful for both data exploration and for communicating results to others. Some of the methods described include:

methods for discrete frequency distributions,
association plots for two-way tables,
mosaic displays,
effects plots for log-linear models and logistic regression, and
correspondence analysis.
models for repeated measures

These graphical techniques are illustrated with real data in the form of two-way and multi-way frequency tables. Course notes will provide many detailed examples.

This course is an updated version of Graphical Methods for Categorical Data, with somewhat greater introductory material on logistic regression and loglinear models. These techniques are all described and illustrated in my book, Visualizing Categorical Data, which will be available in the York Bookstore. Another useful reference is Categorical Data Analysis Using the SAS(R) System by Stokes, Davis, and Koch (SAS Institute, 1995, ISBN 1-55544-219-6) which will also be available in the York bookstore.