NONPARAMETRIC MULTIPLE COMPARISONS WITH MORE THAN ONE CONTROL USING NORMAL SCORES AND SAVAGE STATISTICS

by ELEANNE SOLORZANO.

Abstract: Distribution-free methods are considered for comparing multiple treatments to multiple controls in the one-way layout. Using normal scores statistics, one-sided tests are developed for multiple treatment versus multiple control comparisons of medians. Using Savage statistics, analogous results are developed for comparing scale parameters of the treatments to the controls with non-negative data. Upper bounds for the Type I error probabilities are established for a general number of treatments and controls using the concept of partitions and a union-intersection algorithm. Asymptotic relative efficiencies of these tests relative to their competitors are identical to their two-sample counterparts.

Key Words: Distribution-free, linear-rank statistic, many-one procedure, treatment versus control, score function

Author:
Eleanne Solorzano, eleannes@cisunix.unh.edu

Editor: Munsup Seoh,munsup.seoh@wright.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (161119 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 2269 times since July 24, 2006.


Return to the InterStat Home Page.