ABSTRACT
This chapter describes why simple linear regression is not ideal for Poisson data. It focuses on a Poisson regression model and analyses the assumptions for inference. The chapter describes the likelihood for a Poisson regression and describes how it could be used to estimate coefficients for a model. It explains the estimated coefficients from a Poisson regression and construct confidence intervals for them. The chapter explores the use of deviances for Poisson regression models to compare and assess models and an offset to account for varying effort in data collection. It considers a zero-inflated Poisson model. The chapter presents three case studies: Household size in the Philippines, campus crime and weekend drinking. These three case studies provide context for some of the familiar concepts related to modeling such as exploratory data analysis, estimation, and residual plots.
