Accession Number : ADA023937

Title :   Inferential Techniques for Weibull Populations. II.

Descriptive Note : Final rept. 15 Apr 74-15 Jun 75,

Corporate Author : SKF INDUSTRIES INC KING OF PRUSSIA PA

Personal Author(s) : McCool,John I.

Report Date : JUN 1975

Pagination or Media Count : 76

Abstract : Two test statistics are examined and shown to have the required properties to serve as a test of the null hypothesis that k 2-parameter Weibull populations, having a common but unknown shape parameter, have a common scale parameter. The test statistics are the ratios of the maximum likelihood (ML) shape-parameter estimate derived on the assumption that the scale parameters differ to (1) the ML shape-parameter estimate calculated using the earliest failure in each sample and to (2) the ML shape-parameter estimate calculated after combining the k data samples. The theory for drawing inferences on Weibull parameters in the presence of a single competing Weibull failure mode having the same (but unknown) shape parameter as the primary mode is generalized to the case of k competing modes. An illustration is given of the construction of a joint confidence region for the proportion of failures due to mode j and a percentile of the distribution of failure due to mode j when only that mode operates.

Descriptors :   *Weibull density functions, *Statistical samples, *Statistical tests, Estimates, Analysis of variance, Monte Carlo method, Population(Mathematics), Random variables

Subject Categories : Statistics and Probability

Distribution Statement : APPROVED FOR PUBLIC RELEASE