A More Robust Procedure for Testing the Null Hypothesis in MANOVA

by Chris N. Kladopoulos and Philip H. Ramsey.

Abstract: Earlier research on Pillai's V statistic and Wilks' generalized correlation ratio statistic, U, suggest that a combination of these two procedures can be used to form a more robust test in MANOVA. The present investigation considers two methods of forming such a composite procedure. One of these procedures is shown to provide a robust alternative to existing methods for testing the full-null hypothesis for small samples while the other is shown to be superior for medium to large samples.

Key Words: Significance testing, robustness, multivariate analysis

Authors:
Chris N. Kladopoulos, Kladcuny@aol.co
Philip H. Ramsey, phramsey@aol.com

Editor: Dayanand N. Naik,dnaik@odu.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (80674 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 2562 times since July 24, 2006.


Return to the InterStat Home Page.