Bayesian statistical inference for start-up demonstration tests with rejection of units upon observing d failures

Abstract

This paper is concerned with Bayesian estimation and prediction in the context of start-up demonstration tests in which rejection of a unit is possible when a pre-specified number of failures is observed prior to obtaining the number of consecutive successes required for acceptance of the unit. A method for implementing Bayesian inference on the probability of success is developed for use when the test result of each start-up is not reported or even recorded, and only the number of trials until termination of the testing is available. Some errors in the related literature on the Bayesian analysis of start-up demonstration tests are corrected. The method developed in this paper is a Markov chain Monte Carlo (MCMC) method incorporating data augmentation, and it additionally enables Bayesian posterior inference on the number of failures given the number of start-up trials until termination to be made, along with Bayesian predictive inferences on the number of start-up trials and the number of failures until termination for any future run of the start-up demonstration test. An illustrative example is also included.

Keywords:

• start-up demonstration test
• Bayesian estimation
• MCMC
• data augmentation

Acknowledgements

The author thanks the Editor of this journal and the two referees of this paper for their helpful remarks and suggestions. This research was supported by a Discovery Grant from the Natural Sciences and Engineering Research Council of Canada (NSERC).