[en] Frequently data are described by 2 x 2 contingency tables. For example, each 2 x 2 table arises from two dichotomous classifications such as control/treated and respond/did not respond. Multiple 2 x 2 tables result from stratifying the observational units on the basis of other characteristics. For example, stratifying by sex produces separate 2 x 2 tables for males and females. From each table a measure of difference between the response rates for the control and the treated groups is computed. The researcher usually wants to know if the response-rate difference is zero for each table. If the tables are homogeneous, the researcher can generalize from a statement concerning an average to a statement concerning each table. If tables are not homogeneous, homogeneous subsets of the tables should be described separately. This paper presents tests for homogeneity and illustrates their use. 11 refs., 6 tabs