MethodologyforBayesianModel.PDFVIP

Methodology for Bayesian Model Averaging: An Update Jennifer A. Hoeting? Colorado State University Abstract The standard practice of selecting a single model from some class of models and then making inferences based on this model ignores model uncertainty. Ignoring model uncertainty can im- pair predictive performance and lead to overstatement of the strength of evidence via p-values that are too small. Bayesian model averaging provides a coherent approach for accounting for model uncertainty. A variety of methods for implementing Bayesian model averaging have been developed. A brief overview of Bayesian model averaging is provided and recently developed methodology to perform Bayesian model averaging in speci?c model classes is described. Liter- ature references as well as software descriptions and relevant webpage addresses are provided. 1 Introduction Chronic wasting disease of the deer family has recently garnered daily headlines in newspapers in the state of Colorado. The disease is a member of a group of infectious diseases known as prion diseases which a?ect animals and humans. “Mad-cow” disease is one of the most widely known prion diseases and scientists are hopeful that the study of chronic wasting disease will increase understanding of this and other prion diseases. Scientists would like to determine the prevalence of the disease in Colorado and are considering predictors of disease prevalence such as population density, habitat and other potentially related covariates. One concern about the study of prevalence is that the currently available data are generally from hunter returns where hunters bring in samples of harvested deer for testing. While this allows scientists to map disease prevalence around the state, the data are clearly subject to biases related to this non-random sampling. Class