Bayesian Robustness of the Posterior Predictive p-Value

Abstract

In this paper, the Bayesian robustness of the posterior predictive p-value is studied. First of all, it is proved that Lavine's linearization technique can be extended for analyzing this problem. Then, the result is applied to the ϵ-contamination class of prior distributions.

Keywords:

• Bayesian robustness
• ϵ-contamination class
• Linearization technique
• Posterior odds
• Posterior predictive p-value

Acknowledgments

This research was partially supported by DGESIC (Spain) under grant BFM2001-0169 (the first author) and under grant BF2002-03769 (the second author). We thank the referee for some useful suggestions.