>>13192494The analysis in this picture has nothing to do with Bayesian analysis at all. This is not a Bayesian comparison between M1 and M2. A Bayesian analysis would compare an continuous spectrum of models and conclude that q = 116/202 (by Laplace's rule of succession) is the model most favored by this experimental result and the most likely given a uniform prior; the Bayesian would not compare M1 and M2 directly, because no priors between those models are stated.
This is a discrepancy between frequentist analysis and something else that is definitely not Bayesian. Bayesian analysis is not involved in this shootout at all.
