Bayes factor model comparison for psychological science

In this dissertation it is argued that rival scientific models should be compared by treating them as competing forecasters and assessing their relative predictive adequacy using the Bayes factor. The first part of the dissertation is concerned with bridge sampling, a computational procedure for estimating the marginal likelihood -- the key quantity for computing Bayes factors. The second part of the dissertation is concerned with Bayesian methods for meta-analyzing a set of studies. One central concept of this part is the idea to combine several forecasters using Bayesian model averaging (BMA). The third part of the dissertation introduces Bayesian approaches to a number of standard statistical tests. A central idea of this part is the incorporation of prior knowledge into the analyses to make the models' forecasts more precise.