Blog: Sabermetric Research: Research conclusions *have* to be bayesian