Skip to Content

All University of South Carolina system institutions will be closed through the end of the spring semester. Columbia campus virtual instruction will continue through the end of the Summer II semester (Aug 1). Details can be found on the coronavirus landing page.

Department of Statistics

STAT 770

770—Categorical Data Analysis. (3) (Prereq: STAT 704 and consent of instructor, or BIOS 759) Advanced methods for analysis of discrete data. Higher-order contingency tables, log-linear and other generalized linear models. Multivariate methods for matched pairs and longitudinal data.

Sample Course Homepage: Recent Semester

Usually Offered: Fall Semesters

Purpose: To develop expertise in description and statistical inference for contingency tables. To extend expertise in constructing and interpreting models for binary response data. To develop expertise in constructing and interpreting log-linear and other generalized linear models for categorical data.

Current Textbook: Categorical Data Analysis, 2nd ed. by A. Agresti. Wiley, 2002.


Topics Covered Chapters Time        
Description and inference for two-dimensional contingency tables:
Categorical response data, sampling schemes and distributions, summary measures of association, 
large sample inference, exact test for small samples
1-3 2 weeks
Models for binary response variables and generalized linear models:
Logistic regression, logit models, probit models, model diagnostics
4-6 3 weeks
Ordinal and Polytomous Response data: Logit models for ordinal variables, RC models, Generalized logit models, 
cumulative and baseline category logit models
7 2 weeks
Log-linear models:Log-linear models for two dimensions, log-linear 
models for three or more dimensions, testing goodness of fit, estimation model parameters, 
iterative MLEs, hierarchical model fitting, diagnostics
8 2 weeks
Dependent samples: Symmetry models, marginal homogeneity, measuring 
agreement, case-control models
10 2 weeks
Repeated measures:Marginal homogeneity, modeling a repeated categorical 
response, modeling a repeated ordinal response, generalized linear models and quasi-likelihood
11 2 weeks

Notes: Projects to reinforce the concepts will be typical.

Contact Faculty: John Grego

Challenge the conventional. Create the exceptional. No Limits.