Multilevel models for zero-inflated count data in environmental health and health disparities research

Zero inflation has become very common in biostatistics and its application areas, such as environmental statistics and health disparities research. However, how one handles the excessive zeros is quite a challenge. There exists several approaches to modeling in the presence of zero inflation, but these methods have a few disadvantages. This thesis aims to not only introduce a new framework that accommodates zero inflation, but also applies this framework in a variety of application areas. Chapter 1 introduces a novel statistical framework for handling zero inflation in longitudinal data. This chapter applies this new framework to studying the effect of air pollution on arrhythmias. Chapter 2 extends this framework to a Bayesian setting and also accommodates spatially correlated data. Health disparities is the focus of this chapter. The final chapter, Chapter 3, is dedicated to studying the impact of model misspecification on two modeling frameworks that accommodates zero inflation.