Tests for equality of variances between two samples which contain both paired observations and independent observations

Abstract

Tests for equality of variances between two samples which contain both paired observations and independent observations are explored using simulation. New solutions which make use of all of the available data are put forward. These new approaches are compared against standard approaches that discard either the paired observations or the independent observations. The approaches are assessed under equal variances and unequal variances, for two samples taken from the same distribution. The results show that the newly proposed solutions offer Type I error robust alternatives for the comparison of variances, when both samples are taken from the same distribution.