Bayesian Hierarchical Modeling
Jim Albert, Jingchen Hu · 2019
In essence, hierarchical modeling takes into account information from multiple levels, acknowledging differences and similarities among groups. In the posterior analysis, one learns simultaneously about each group and learns about the population of groups by pooling information across groups. This chapter describes hierarchical modeling in two situations that extend the Bayesian models for one proportion and one normal mean. It introduces hierarchical normal modeling using a sample of ratings of animation movies released in 2010. The chapter describes hierarchical beta-binomial modeling with an example of deaths after heart attack. It also motivates the consideration of hierarchical models, outlines the model structure, and implements model inference through Markov chain Monte Carlo simulation.