Contingency table testing for categorical data: SAS implementation

Nonparametric and distribution-free tests of categorical data provide an evaluation of statistical significance between groups of subjects differing in their assignment to a set of categories. This paper describes an implementation in the SAS programming language of three tests to evaluate categorical data. One of these tests, the Contingency Table Test for Ordered Categories evaluates data assessed on at least an ordinal scale where the categories are in ascending or descending rank order. The remaining two tests, Fisher’s Fourfold-Table Test for Variables with Two Categories and Fisher’s Contingency Table Test for Variables with More than Two Categories, evaluate data assessed on either a nominal or an ordinal scale. The program described completes analysis of a 2°C categorical contingency table as would be obtained from the application of a multiple-level rating scale to the behavior of a treatment and a control group.