Regression Analysis of Count Data

Rating:
Summary: authoritative and current treatment of count data
Review: This is a very thorough and authoritative treatment of regression methods for analyzing count data. It is very current and covers many topics not commonly found in books on point processes. Economic applications are emphasized but the broader applicability of the methods is eluded to.

The authors intent is for this book to be read by researchers, graduate students and practitioners in the many fields that make use of count data. Chapter 1 introduces count data, the Poisson distribution and the Poisson process and also shows how the Poisson process can be derived based on the assumption of independent and identically distributed exponential waiting times. It concludes with specification of regression models for counts and a number of practical examples where modeling count data would naturally arise. The importance of the integers is emphasized with the quote from Kronecker at the beginning of the chapter, "God made the integers, all the rest is the work of man."

Chapter 2 provides an extensive treatment of model specification and estimation methods. The authors cover many approaches and provide excellent references to the literature. Generalized linear models provide one common approach in the statistics literature and these methods are well described in this chapter.

Poisson regression is the main topic of Chapter 3 but the chapter goes on to discuss negative binomial models that handle overdispersion. An example of data on doctor's visits is used to illustrate the techniques. Statistical tests for overdispersion are also presented. A variety of other modeling techniques are also provided.

More general models including mixture models are considered in Chapter 4. Chapter 5 looks at ways of evaluating potential models. Chapter 6 provides some real economic data from health services to illustrate the methods of the earlier chapters.

Chapter 7 covers time series analysis for integer data. The authors provide information and literature that is not standard in a text on time series analysis but is applicable to count data.

Subsequent chapters deal with more complexity including multivariate data,longitudinal data analysis and measurement error models. Important recent developments in bootstrap methods and Bayesian statistics are covered in the context of problems for which these methods have demonstrated their value.

This is a great reference book for statisticians and econometricians interested in problems involving counting processes. It could also be used for a graduate school text on point process regression.

Rating:
Summary: It is true!
Review: Well most prople are used in telling stories about the unapliability of economics in real life. That math and logic can not apply to most economic books! Most of the people say that economics are only theory.I will tell you that this book is one of a kind, reading it you will be able to have a better understanding of Econometrics.If you have enough money to pay the rent and some left buy this book ! its better than food!