Toward Evidence-Based Medical Statistics. 2 The Bayes Factor.pdfVIP

Toward Evidence-Based Medical Statistics. 2: The Bayes Factor Steven N. Goodman, MD, PhD Bayesian inference is usually presented as a method for peared in response to an article proposing a Bayesian determining how scientific belief should be modified by analysis of the GUSTO (Global Utilization of Strep- data. Although Bayesian methodology has been one of tokinase and tPA for Occluded Coronary Arteries) the most active areas of statistical development in the past trial (4), are typical. 20 years, medical researchers have been reluctant to em- brace what they perceive as a subjective approach to data When modern Bayesians include a “prior probability distribution for the belief in the truth of a hypothesis,” analysis. It is little understood that Bayesian methods have they are actually creating a metaphysical model of a data-based core, which can be used as a calculus of attitude change . . . The result . . . cannot be field-tested evidence. This core is the Bayes factor, which in its simplest for its validity, other than that it “feels” reasonable to form is also called a likelihood ratio. The minimum Bayes the consumer. . . . factor is objective and can be used in lieu of the P value as The real problem is that neither classical nor Bayesian a measure of the evidential strength. Unlike P values, methods are able to provide the kind of answers cli- Bayes factors have a sound theoretical foundation and an nicians want. That classical methods are flawed is un- interpretation th