ABSTRACT
While meta-analysis is typically conducted using frequentist methods, it also possible to conduct Bayesian meta-analyses. Bayesian meta-analysis is based on the Bayesian hierarchical model. In its essence, this model is identical to the “conventional” random-effects model. The main difference, however, is that prior distributions are assumed for the true overall effect size and between-study heterogeneity variance.
This chapter describes how the Bayesian hierarchical model can be formulated. It also covers the concept of prior distributions, and how they can be specified for a random-effects meta-analysis. Lastly, a hands-on example is used to illustrate how Bayesian meta-analysis models can be fitted in R, and how one can generate a special type of forest plot for the results.
