Bayesian model comparison: Log scores and DICDIC

We investigate Bayesian approaches to answering the frequently-arising model-specification question (Q:) Is model MjMj better than Mj′Mj′? We contrast two Bayesian model-comparison methods–log scores and DICDIC–on their ability to correctly discriminate between fixed-effects Poisson (FEP) and random-effects Poisson (REP) sampling models.