Two-Way Analysis of Variance for Weibull Populations.
Abstract
Expressions are given for the maximum likelihood estimators of the Weibull shape parameter using data from a two-way classification experiment under five hypothesized relationships for the scale parameter in each cell, viz, the scale parameter (1) is constant, (2) has a multiplicative row effect, (3) has a multiplicative column effect, (4) has a multiplicative row and column effect, and (5) has multiplicative row and column and interaction effects. Procedures for analyzing a two-way layout without interaction (a randomized block design) are discussed and illustrated. Techniques are given for estimation of, and inference on, the location parameter of the three-parameter Weibull distribution and a military application of the technique is given. Software improvements are cited in the Monte Carlo simulation computer program for the computation of the distributions of pivotal functions required for analyzing Weibull regression experiments. Methodology for the analysis of a single classification experiment with Weibull distributed response is described and the contents of an applications oriented paper is described. Other applications of the methodology are cited. Software is described for simulation of the distribution of the maximum likelihood estimates of the Weibull parameters that result when the data are grouped. Results of varying interval width and sample size are given.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1980
- Accession Number
- ADA092751
Entities
People
- John I. Mccool