Abstract
Given k independent Poisson populations, suppose one is interested in testing the null hypothesis that all the means are equal against an ordered alternative. Three approaches to this problem are considered: the maximin statistic, the Armitage statistic, and a modified version of the Jorgenson statistics. A simulation study of powers for moderate E(Yl) with k = 4 shows that all three approaches are comparable in power. Considering its simplicity, efficiency and general applicability, the maximin statistic is the recommended test for monotone trend in k independent Poisson means.

This publication has 5 references indexed in Scilit: