Bayesian Analysis of Change - Point Hazard Rate

Change-point hazard rate models arise, for example, in applying \burn-in" techniques to screen defective items and in studying times until undesirable side eeects occur in clinical trials. The classical approach develops estimates of model parameters, with particular interest in the threshold or change-point parameter, but exclusively in terms of asymptotic properties. Such asymptotics can be poor for the small to moderate sample sizes often encountered in practice. We propose a Bayesian approach, avoiding asymptotics, to provide more reliable inference conditional only upon the data actually observed. The Bayesian models can be tted using simulation methods. We develop a very general formulation of the model but also look at special cases which ooer particularly simple tting. We illustrate with an application involving failure times of electrical insulation.