A Computational Approach Test for the Equality of Two Multivariate Normal Mean Vectors under Heterogeneity of Covariance Matrices

Abstract

In this paper, a computational approach test (CAT) was proposed to test the equality of two multivariate normal mean vectors under heterogeneity of covariance matrices. The proposed test was compared with the other popular tests as well as their CAT versions in terms of estimated type I error rate and power. Simulation study shows that the proposed test and CAT versions of tests can be used as a good alternative test to test the equality of two multivariate normal mean vectors under heterogeneity of covariance matrices.

Keywords

• Computational Approach Test
• Parametric Bootstrap Approach
• Simulation Study