Two-Way Analysis of Variance for Weibull Populations.

Abstract
Techniques and software are developed for the analysis of experiments in which the response variable follows the two-parameter and three-parameter Weibull distribution. For the two-parameter case the analogue of the analysis of variance is considered in which the hypothesis that the scale parameter is affected by the levels of an external factor is tested by a shape parameter ratio test. The analogue of Weibull regression is also treated for the case where the scale parameter varies as a power function of a quantitative external factor. A new method is found for drawing inferences on the location parameter of the three-parameter Weibull model, based on maximum likelihood shape parameter estimates. The sample sizes and censoring numbers for which empirical distribution have been determined are summarized. A summary is given of the software developed. Several applications of the techniques are also summarized. (Author)

This publication has 0 references indexed in Scilit: