المستخلص: |
This article treats the statistical inference for factorial experiments under an inverse Gaussian distribution for the failure times. A reciprocal-linear model for the factor effects is motivated from the context of the underlying Wiener process. Simple and explicit form for the solution of the likelihood equations is given. Likelihood ratio tests for the main and interaction effects are derived. The test statistics for the main effects have an exact F distribution, while the test statistic for the interaction effects has an approximate F distribution. An analysis of reciprocals analogue of the usual normal theory analysis of variance, is investigated. An application of the procedures is illustrated with a data set of strength measurements of an insulating material. Fries and Bhattacharyya (1983) consider similar model, but they provide no explicit solution to the likelihood equations but via inverting some random matrix. The test statistics for main effects and interaction effect which provided by Fries and Bhattacharyya (1983), all have approximate F distributions. Also, a correction for their analysis of reciprocals is corrected in the present work.
|