Date of Award
Spring 1980
Document Type
Thesis - Restricted
Degree Name
Master of Science (MS)
Department
Mathematics
First Advisor
Lochner, Robert H.
Second Advisor
Pearson, Larry M.
Third Advisor
Davie, Anne S.
Abstract
Several authors have compared multiple comparison procedures under the assumption of normality and when sampling from skewed populations. In this study we shall compare four multiple comparison procedures under the assumption that the parent populations are Weibull, when the contrasts of interest are the pairwise comparisons among the means of K samples. The methods studied are: Scheffe's S-method, , Ounn-Sidak's OS-method, Spjotvoll and Stoline's T -method and Hochberg's GT2-procedure. Emphasis will be placed on comparing the average confidence interval lengths for each method. It is shown that the GT2-method generally gives , the shortest average confidence interval lengths. The T - method gives the shortest average confidence interval lengths when the sample sizes are "nearly" equal.
Recommended Citation
Mata, Luis E., "Robustness of Four Multiple Comparison Methods When the Parent Populations Are Weibull" (1980). Master's Theses (1922-2009) Access restricted to Marquette Campus. 2158.
https://epublications.marquette.edu/theses/2158