Bayesian Multilevel Modeling and Its Application in Comparative Journalism Studies

Abstract

Comparative approaches are frequently used in communication research, especially journalism studies. The purpose of this article is to argue that Bayesian multilevel regression is the most justifiable option for analyzing comparative data. We argue that it is the only approach that can simultaneously account for the non-atomicity (nested nature) and non-stochasticity (nonrandom sampling) of comparative data. Using the openly available Worlds of Journalism Study and useNews data sets, we demonstrate how to apply the Bayesian approach for the analysis of comparative data. We address the common challenges when using the Bayesian approach and highlight the advantages of posterior predictive checks for modeling checking.