Comparisons of populations in life testing

Abstract

Previous works on comparisons of populations are largely concerned with cases where the lifetime has an exponential or a Weibull distribution with complete or Type II censored data. However, relatively little work has been done for the gamma distribution. Comparisons of gamma populations have been limited to the case of two populations with complete data. Furthermore, no multiple comparison procedure has been developed. The purpose of this thesis is to develop statistical methods for comparing gamma distributions with complete, Type I or Type II censored data. That is, when there exist m gamma populations, m-sample test and pairwise multiple comparison procedures are developed, using the likelihood ratio method or the concept of least significance difference. All procedures are computerized.