Bayesian model averaging: A tutorial

Standard statistical practice ignores model uncertainty. Data analysts typi cally select a model from some class of models and then proceed as if the s elected model had generated the data. This approach ignores the uncertainty in model selection, leading to over-confident inferences and decisions tha t are more risky than one thinks they are. Bayesian model averaging (BMA) p rovides a coherent mechanism for accounting for this model uncertainty. Sev eral methods for implementing BMA have recently emerged. We discuss these m ethods and present a number of examples. In these examples, BMA provides im proved out-of-sample predictive performance. We also provide a catalogue of currently available BMA software.