Abstract
We consider the problem of testing the equality of propor-tions in several multinomial populations. The standard likelihood ratio procedure is modified to construct a repeated significance test. This test is as powerful as the likelihood ratio test, but it requires less number of populations for sampling purposes.