Comparison of Failure Rates Using Rank Tests

Abstract

Abstract This article deals with tests of the hypothesis that the failure rates of two products are equal, against the alternative hypothesis that one failure rate dominates the other. The hypotheses are tested using the locally most powerful rank tests based on samples from the two products. Practical implications of such tests, with regard to used products, are discussed. Some numerical aspects of the use of the asymptotic results on small samples are also presented.