The alpha and beta errors in randomized trials.