Bayesian parameter estimation and hypothesis testing offer a useful alternative to the classic frequentist paradigm within psychological science. This class will cover the foundations of Bayesian inference and hypothesis testing with the primary emphasis on fitting multiple regression and multi-level models common within psychology. A variety of response distributions will be discussed: Gaussian, binary and count, ordinal, survival, probability, and zero-inflated models, among others. Topics include: model calibration, regularization, prior and posterior predictions, Bayes factors, missing data, Bayesian power, cross-validation, Bayesian meta-analysis, distributional models, and multivariate response models. Models will be fit using the R package brms, which relies on the more general Stan language.
PREREQ: PSYCH 5068 or a proficiency in multilevel modeling