Abstract:

Compared to the traditional statistical methods, Bayesian linear mixed-effects modeling (BLMM) has a great number of advantages in dealing with the hierarchical structures underlying datasets and providing more intuitive statistical results. These advantages together popularize BLMM in psychological and other field research. However, there is still a lack of tutorials on the practical applications of BLMM in psychology studies in Chinese. Therefore, we first briefly introduced the basic concepts and rationales of BLMM. Then we employed a simulated dataset to demonstrate how to understand fixed effects and random effects, and how to use the popular brms R package to specify models for BLMM based on the experimental design. We additionally covered the procedure of pre-specifying priors with prior predictive checks, and the steps of performing hypothesis testing using the Bayes Factor. BLMM, with its extensions such as Generalized BLMM, has great flexibility and capability, they can and should be applied in various psychology research.

Key words: Bayesian, linear mixed-effects modeling, hierarchical models, Bayes Factor, brms