On Bayes Factor Functions

Bayes Factor Functions (BFFs) represent Bayes factors as functions of prior hyperparameters (Johnson et al., 2023; Datta et al., 2025a). While earlier work has primarily focused on normal-moment and gamma priors, we build on recent results by Wagenmakers and Grasman (2025) to explore broader classes of priors for constructing BFFs. For tests based on common test statistics, our findings show that inverse-moment and inverse-gamma priors can yield higher expected posterior probabilities in favor of both true null and true alternative hypotheses than other default priors. Using test statistics from replicated experiments, we also demonstrate how method-of-moments estimators can inform prior parameter selection within these families, thereby enhancing inference in replicated studies and meta-analyses.