Abstract
A program has been developed to test the significance of an m × n table by using an extension of the Fisher exact probability test. Theoretical corrections are made regarding the previous work by Bedeian and Armenakis (1977). Various methods of discovering the source of association at a cellwise level are given.