North America

Blue Room - Session 8

1:00 to 2:00 p.m. Thursday, June 15, 2017

Shifting to a Bayesian Binomial Test Paradigm — A Case Study

Reliability is growing in importance, but in an era of shrinking defense budgets, one of the first items decision makers wish to reduce is system testing. This is a case study for adjusting the requirement for an occasional use system from a mean time between failure requirement to a binomial probability of success, and the development of a Bayesian test program to dramatically reduce the associated testing necessary. The presentation addresses how this approach leveraged additional data sources that management would not accept under a frequentist approach, additional considerations relative to extended use in a marine environment, and the challenges in communicating the impact of the requirement change to engineering management with limited statistical expertise.

Key Words: Reliability, Confidence, Bayes, Bayesian Statistics, Binomial, Weighting Prior

Matthew Moseley, Michael Coon and Scott Bindel

Naval Surface Warfare Center – Carderock Division

West Bethesda, Maryland