IJPAM: Volume 40, No. 1 (2007)

A DISCRIMINANT CONDITION FOR THE TEST OF
GREATEST POWER IN THE MANOVA MODEL WHEN
THE DIMENSION IS LARGE COMPARED
TO THE SAMPLE SIZE

Tetsuto Himeno
Graduate School of Mathematics
Kyushu University
6-10-1, Hakozaki, Higashiku, Fukuoka-City, 812-8581, JAPAN
e-mail: himeno@math.kyushu-u.ac.jp


Abstract.The asymptotic non-null distributions of the likelihood ratio, Lawley-Hotelling, and Bartlett-Nanda-Pillai test statistics for the MANOVA procedure are obtained when both the sample size and the dimension tend to infinity. These tests are of equal power in the limit. Using the asymptotic distributions of the three test statistics, we compare their asymptotic power. We derive a simple method for selecting the test of greatest power.

Received: August 9, 2007

AMS Subject Classification: 62F05, 62E20

Key Words and Phrases: MANOVA, likelihood ratio, Lawley-Hotelling, Bartlett-Nanda-Pillai, power comparison, asymptotic distribution

Source: International Journal of Pure and Applied Mathematics
ISSN: 1311-8080
Year: 2007
Volume: 40
Issue: 1