Bayesian modeling of expert-to-expert variability and dependence in estimating rare event frequencies

Abstract This paper presents a Bayesian model for combining expert opinions when there is an inherent variability among the estimates. The proposed model accounts for the uncertainty due to such variability as well as that arising from potential error made by each expert. Dependence between experts is also modelled explicitly. Several examples are provided to test the behavior of the model and its application for estimating rare event frequencies.