A Wald test on the problem of homogeneity of variances

Abstract

In this article, we propose a Wald test for homogeneity of variance in several normal distributions. The Wald test is an asymptotic test, and it is known that asymptotic tests perform well in large sample sizes, but not in small sample sizes. For this reason, we modified this test using a parametric bootstrap method called the computational approach test. To evaluate the efficiency of the proposed test against the existing tests in the literature, we compared this test in terms of the estimated size and power of tests. The results of our simulation study show that the proposed test performs well in especially small and different sample sizes under normal distributions. Besides, we illustrate the proposed test using a numerical example.