Logit Models

John Fox

The last 30 years have witnessed a proliferation of statistical methods for the analysis of qualitative/categorical data, prominent among which are linear logit (or logistic regression) models. Unlike the more familiar linear models for regression analysis and analysis of variance, linear logit models are appropriate for analyzing qualitative/categorical dependent variables. Like linear models, logit models are capable of handling one or several independent variables, which may be both qualitative and quantitative.

This course will introduce logit and related models for dichotomous (two-category) and polytomous (several-category) dependent variables, including ordered categories. We shall also consider the application of logit models to the analysis of multidimensional contingency tables. A basic understanding of linear least-squares regression analysis, including dummy-variable regression and analysis of variance, is assumed.