Article ID Journal Published Year Pages File Type
1149028 Journal of Statistical Planning and Inference 2012 8 Pages PDF
Abstract
We consider optimal testing procedures for specific models of early and instantaneous failures in reliability studies. These models are typically used to accommodate lifetime data that have a higher concentration of failures near time zero. We show that it is possible to derive uniformly most powerful tests, for testing the mixing parameter in the instantaneous failure model, for general lifetime distributions. A novel procedure to test for early failures, which uses an invariance property of the maximum likelihood estimate of the nuisance parameter, is also presented.
Related Topics
Physical Sciences and Engineering Mathematics Applied Mathematics
Authors
, ,