Special QME Seminar: A Weakly-Informative Group-Specific Prior Distribution for Meta-analysis
While Bayesian meta-analysis has flourished both in methodological and substantive work, group-specific Bayesian modeling remains scarce. Common practice for choosing prior distributions entails using typical non-informative priors. Currently, there is a push to use more informative prior distributions. In this paper I propose a weakly-informative group-specific prior distribution. The proposed prior distribution uses a frequentist estimate of conditional between-studies heterogeneity as the noncentrality parameter in a folded noncentral t distribution. This prior distribution is then modeled individually for groups based on some categorical factor. I discuss previous prior distribution choices for heterogeneity parameters and examples of existing group-specific Bayesian meta-analysis work. This is followed by a presentation of the new approach for using a group-specific weakly-informative prior distribution. Select results from a larger simulation study are given. I conclude with a discussion, including overall findings, limitations, and avenues for potential future work.