Testing the homogeneity of the relative risk under multiple matching

Abstract

A new procedure is developed for testing the homogeneity of the relative risk in a multiply-matched case-control design, both with constant and variable matching. The test does not require explicit stratification on the matching covariate.