This is a course on categorical data analysis. It presents the most important traditional methods for analyzing categorical data, such as tests of goodness-of-fit, regression models for count data, contingency table analysis, logistic regression, and loglinear models. It will also include some modern methods such as classification and regression trees. The course is designed for graduate and advanced undergraduate statistics majors as well as graduate students in other disciplines who need to analyze categorical data in their work. Students are expected to have some background in regression and analysis of variance.

Distributions for categorical variables and goodness of fit

• Binomial, multinomial, Poisson, and negative binomial distributions
• Overdispersion and lack of fit

Regression models for count data

• The generalized linear model
• Poisson regression
• Negative binomial regression
• Zero-inflated count regression
• Zero-truncated Poisson regression

Two-way contingency tables

• Tests and comparisons of proportions
• Tests of independence
• Odds ratio and relative risk
• Log-linear models for two-way tables
• Conditional and exact tests
• Structural zeroes and quasi-independence
• Ordered categories

Multi-dimensional contingency tables

• Simpson's paradox
• Conditional association
• Hierarchical log-linear models
• Models for ordered categories
• Conditional analyses

Regression models for binary data

• Logistic regression
• Case-control studies
• Categorical predictors
• Other link functions
• Over-dispersion
• Smoothing binomial data

Regression models for multiple category response data

• Multinomial logistic regression
• Proportional odds model
• Adjacent-categories logit model
• Continuation ratio models

Introduction to classification and regression trees

• QUEST and CRUISE for classification
• GUIDE for Poisson regression
• LOTUS for logistic regression

• R and SAS on personal computers or departmental LINUX workstations
• Free QUEST, CRUISE, GUIDE and LOTUS computer programs on personal computers or departmental LINUX workstations