|
Activity Number:
|
552
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #305750 |
|
Title:
|
Nonparametric Methods in Multivariate Factorial Designs
|
|
Author(s):
|
Arne Bathke*+ and Solomon W. Harrar
|
|
Companies:
|
University of Kentucky and South Dakota State University
|
|
Address:
|
875 Patterson Office Tower, Lexington, KY, 40506-0027,
|
|
Keywords:
|
nonparametric statistics ; multivariate ; rank statistics ; factorial designs ; asymptotics ; MANOVA
|
|
Abstract:
|
We present three different multivariate nonparametric tests: ANOVA-Dempster-Type, Lawley-Hotelling, and Bartlett-Nanda-Pillai. Asymptotic theory has been derived for two different asymptotic frameworks: the situation where the number of replications is limited, whereas the number of treatments goes to infinity; and the regular setup with large sample sizes. The nonparametric tests are based on separate rankings for the different variables. For these tests, we also discuss different approximation procedures. The finite performance of the tests is investigated through simulations. It turns out that the proposed nonparametric tests perform very well as compared to their parametric competitors, especially in the presence of outliers. Application using SAS standard procedures is demonstrated with examples corresponding to both asymptotic frameworks.
|
