The Sceptical Bayes Factor for the Assessment of Replication Success

Replication studies are increasingly conducted but there is no established statistical criterion for replication success. We propose a novel approach combining reverse-Bayes analysis with Bayesian hypothesis testing: sceptical prior determined the effect size such that original finding longer convincing in terms of Bayes factor. This then contrasted to an advocacy (the reference posterior based on study), and success declared if data favour over at higher level than favoured null hypothesis. The factor highest where can be declared. A comparison existing methods reveals combines several notions replicability: it ensures both show sufficient evidence against penalises incompatibility their estimates. Analysis asymptotic properties error rates, as well case from Social Sciences Project advantages method assessment replicability.