To P or Not to P: Backing Bayesian Statistics