Tutorial on modeling ordered categorical response data.

In the past decade there has been great progress in the development of methodology for analyzing ordered categorical data. Logit and log linear model-building techniques for nominal data have been generalized for use with ordinal data. There are many advantages to using these procedures instead of the Pearson chi-square test of independence to analyze ordered categorical data. These advantages include (a) more complete description of the nature of associations and (b) greater power for detecting population associations. This article introduces logit models for categorical data and shows two ways of adapting them to model ordered categorical data. The models are used to analyze a cross-classification table relating mental impairment and parents' socioeconomic status.