Jan Sprenger (University of Turin) (webinar)

“Bayesian Philosophy of Science: A Guide for Hypothesis Testing in Social Science”

Abstract: Hypothesis tests are among the most widespread statistical tools in social science. In particular, their use (and frequent misuse) in psychology stands at the center of a longstanding debate, with eminent methodologists such as Paul Meehl and Jacob Cohen among their fiercest critics. More recently, the prevalent hypothesis testing methodology focused on the concept of statistical significance has also been identified as one of the causes of the replication crisis. In spite of all these criticisms, scientists find it hard to settle on an alternative. This talk presents Bayesian reasoning—i.e., a school of scientific inference where agents update their rational degrees of belief in the light of the observed evidence—as a superior alternative to frequentist hypothesis testing. Specifically, I will highlight how key concepts of frequentist inference such as severe hypothesis testing, pre-experimental error control and asymmetric test design can be implemented into (suitably modified) subjective Bayesian inference.