Applied Bayesian Analytics for Latent Variable Modeling

3 credits · Graduate School

This course is intended as both a theoretical and practical introduction to Bayesian latent variable modeling. Bayesian statistical modeling and inference are an attractive alternative to traditional (frequentist) approaches in that a Bayesian perspective offers a coherent approach to statistical modeling, including building and fitting models with complex features, making inferences, and representing uncertainty. An understanding of Bayesian statistical modeling will be developed by relating it to students’ existing knowledge of traditional (frequentist) approaches and latent variable modeling (e.g., multilevel modeling, item response modeling, and mixture modeling). Course topics include Bayesian inference, maximum likelihood (ML) estimation, Markov chain Monte Carlo (MCMC) sampling, estimation method comparisons (ML vs. shrinkage estimators), prediction and prediction error for latent variables, MCMC for multilevel models, MCMC for item response models, MCMC for mixture models, Bayesian model comparisons, and Bayesian model-data fit [3].