Weighted distributions viewed in the context of model selection: A Bayesian perspective