Full Program »
Confirmationist and falsificationist paradigms in statistical practiceClassical hypothesis testing is generally taken to follow a falsificationist, Popperian philosophy: research hypotheses are put to the test and rejected when data do not accord with predictions. Bayesian inference is generally taken to follow a confirmationist philosophy: data are used to update the probabilities of hypotheses. We reject this conventional Bayesian-frequentist divide. We argue that (1) classical significance testing is actually used in a confirmationist sense and does not do what it purports to do; and (2) Bayesian inference cannot in general supply reasonable probabilities of models being true.