To the Editor.
—Drs Brophy and Joseph1 propose the use of Bayesian methods as an alternative to classical methods for statistically analyzing clinical trial data. They indicate (correctly) that researchers want to say, "There's a 98% probability of t-PA mortality being less than SK mortality." Bayesian methods seem to provide this capability, using the same mathematics that allow us to use screening tests to conclude that there is a 92% chance that a patient with a positive test result has a particular disease. However, the mathematics that allow the latter declaration are based on frequencies of occurrence, and even classical statisticians agree with that. When modern Bayesians include a "prior probability distribution for the belief in the truth of a hypothesis," they are actually creating a metaphysical model of attitude change that borrows from the familiar world of frequency-based probabilities. The result (eg, probability oft-PA mortality being lower than
Browne RH. Bayesian Analysis and the GUSTO Trial. JAMA. 1995;274(11):873. doi:10.1001/jama.1995.03530110035028
Customize your JAMA Network experience by selecting one or more topics from the list below.