An analysis of multivariate ordinal polytomous data using the log‐linear probability model

Abstract

This paper proposes a statistical method to deal with multivariate ordinal polytomous data. The gamma coefficient developed by Goodman and Kruskal is extended to a partial measure of association in the framework of the log‐linear model probability model. Then, this method is applied to Sewell and Armer's data on educational aspirations of high school students in order to reexamine the validity of their conclusions.