STAT 358
Introduction to Categorical Data Analysis
Last Offered Spring 2022
Division III Quantitative/Formal Reasoning
This course is not offered in the current catalog

This course focuses on methods for analyzing categorical response data. In contrast to continuous data, categorical data consist of observations classified into two or more categories. Traditional tools of statistical data analysis (such as linear regression) are not designed to handle such data and pose inappropriate assumptions. We will develop methods specifically designed for modeling categorical data, with applications in the social and biological sciences as well as in medical research, engineering and economics. This course has two parts. The first part will discuss statistical inference for parameters of categorical distributions (Bernoulli, Binomial, Multinomial, Poisson) and for measures of association arising in contingency tables (difference and ratio of proportions and odds ratios). Inferential methods covered include Wald, score and likelihood ratio tests and confidence intervals, as well as the bootstrap. The longer second part will focus on statistical modeling of categorical response data via generalized linear models, with a heavy focus on logistic regression models with both quantitative and categorical predictors and their interactions. Model fitting and inference will be based on maximum likelihood and carried out via R.