Chapter 4 - Bayesian hierarchical modeling for the linkages between air pollution and population health

Abstract

Bayesian hierarchical modeling provides a flexible and robust framework to develop appropriate statistical models to better understand the complex relationships between air pollution and population health. Bayesian hierarchical modeling benefits from two most significant advances in modern applied statistics: generalized linear mixed models (GLMMs) and Bayesian estimation via Markov chain Monte Carlo simulation. This chapter starts with a brief introduction to GLMMs and Bayesian inference; describes the three stages of Bayesian hierarchical modeling: data model, process model, and parameter model; and demonstrates how to construct and implement Bayesian hierarchical spatial modeling for the linkages between air pollution and population health using a case study.