Measures of Nominal-Ordinal Association

Measures are formulated for summarizing the strength of association between a nominal variable and an ordered categorical variable. The measures are differences or ratios of probabilities of events concerning two types of pairs of observations. They can be used to describe the degree of difference between two or more groups on an ordinal response variable. The measures summarize and complement the results of fitting models to nominal-ordinal cross-classification tables, especially when a single structural model form cannot be found that adequately describes an entire table or set of tables.