ABSTRACT
When the relationship between a pair of dichotomous random variables is being investigated, it is sometimes examined in two or more populations. As a result, the data to be analyzed consist of a number of 2 × 2 contingency tables rather than just one. In some cases, these tables originate from different studies; more often, they are the results of a single investigation that have been subclassified, or stratified, by some factor that is believed to influence the outcome. In either event, it is possible to make inference about the relationship between the two variables by examining the association in each table separately. In many cases, however, it is more useful to be able to combine the information across tables to make a single, unifying statement.
