Bayesian Evaluation of a Specific Hypothesis
开发了一种贝叶斯方法，用于计算参数估计值接近特定值的后验概率，从而直接衡量支持某个假设的证据。该方法应用于评估有效市场假说，发现支持证据很少。
Abstract A bayesian procedure is developed to compute the posterior probability that a set of parameter estimates is arbitrarily close to a specified set of values. This procedure allows researchers to directly measure the evidence favoring a specific hypothesis. Because of the current inability to measure this evidence directly, some researchers are confusing the inability to reject a null hypothesis with the justification for accepting it and are thus committing a null hypothesis error. To illustrate the applicability of this new procedure, it is used to evaluate the evidence supporting one version of the efficient market hypothesis. Contrary to the results of an earlier study, little support was found for the efficient market hypothesis.