Tests of Significance in Theory and Practice

摘要: The best (most widely) received theory for tests of significance is that due largely to Fisher. Embellished with Neyman's mathematics, Fisher's very well received. But logic not consistent Bayes' theorem. And theorem beyond reproach. Thus, deficient. However, in practice, there often some redress. Indeed, sometimes level P coincides mathematically the posterior probability null hypothesis, i.e. P=p(hOIE), where E sample event (evidence). More generally, a good Fisherian tends intuitively (although certainly inevitably) toward inference he would make if employed explicit subjective priors. In effect, almost Bayesian. 1