One more comment: statistics has a lot of problems even in theory. Using statistics to interpret how well people compare to an allegedly ideal statistical model, when you're studying complex systems like brains, and when methodology is really hard, is also just a host of degrees of freedom along which to err or fuge the results.