Discussion of “Specifying prior distributions in reliability applications”

Abstract

AbstractSpecifying prior distributions in the Bayesian method is a fundamental but complex problem, especially when conjugate prior does not exist. Tian et al. (Appl Stoch Models Bus Ind; 2023) have captured the spirit of specifying prior distributions in Bayesian analysis for reliability data and presented different approaches coherently. In this discussion, I will focus on specifying prior distributions based on the knowledge of the aging behavior or hazard function and study the effect of misspecifying the informative priors. A Monte Carlo simulation study based on Type 2 censored data from Weibull distribution is used to illustrate the performance of the estimation procedures based on informative and non‐informative priors.