Count Data Models

This text presents statistical methods for the analysis of events. The primary focus is on single equation cross-section models. It addresses both the methodology and the practice of the subject and provides both a synthesis of a diverse body of literature and a contribution to the progress of the methodology, establishing several new results and introducing new models. Starting from the standard Poisson regression model as a benchmark, the causes, symptoms and consequences of misspecification are worked out. Both parametric and semi-parametric alternatives are discussed. While semi-parametric models allow for robust interference, parametric models can identify features of the underlying data generation process.