Talk: Eric-Jan Wagenmakers (Amsterdam)

Title:

APPROXIMATE OBJECTIVE BAYES FACTORS FROM P-VALUES AND SAMPLE SIZE: THE 3p√n RULE

Abstract:

This presentation consists of a main course and a dessert. The main course concerns the approximate objective Bayes factor (henceforth JAB) proposed by Sir Harold Jeffreys in 1936. We revisit JAB and introduce a simple piecewise transformation that establishes a direct connection to the frequentist two-sided p-value. This provides p-value practitioners with a straightforward opportunity to obtain Bayesian benefits such as the ability to monitor evidence as data accumulate without reaching a foregone conclusion. For dessert I demonstrate that the much-bemoaned sensitivity of the Bayes factor to the prior distribution is in fact a condition sine qua non for coherent (i.e., non-farcical) inference.