Tests of Variance Equality When Distributions Differ in Form and Location

Publisher Website
Google Scholar
Abstract
Type I error rates and power were estimated for ten tests of variance equality under various combinations of the following factors: similar and dissimilar distributional forms, equal and unequal means, and equal and unequal sample sizes. The results indicated that four tests had appropriate Type I error rates for most of the distribution combinations. Of these tests, O'Brien's test tended to be most powerful with equal-sized samples and with an indirect relationship between sample sizes and population variances. Using the Welch-James statistic with O'Brien's r(.5) statistic as the dependent variable tended to lead to the most powerful test when the sample sizes and population variances had a direct relationship.

This publication has 11 references indexed in Scilit: