Researchers in fields ranging from biology and medicine to the social sciences, law, and economics regularly encounter variables that are discrete or categorical in nature. While there is no dearth of books on the analysis and interpretation of such data, these generally focus on large sample methods. When sample sizes are not large or the data are otherwise sparse, exact methods--methods not based on asymptotic theory--are more accurate and therefore preferable.
This book introduces the statistical theory, analysis methods, and computation techniques for exact analysis of discrete data. After reviewing the relevant discrete distributions, the author develops the exact methods from the ground up in a conceptually integrated manner. The topics covered range from univariate discrete data analysis, a single and several 2 x 2 tables, a single and several 2 x K tables, incidence density and inverse sampling designs, unmatched and matched case -control studies, paired binary and trinomial response models, and Markov chain data. While most chapters focus on statistical theory and applications, three chapters deal exclusively with computational issues. Detailed worked examples appear throughout the book, and each chapter includes an extensive problem set.
Written at an elementary to intermediate level, Exact Analysis of Discrete Data is accessible to anyone having taken a basic course in statistics or biostatistics, bringing to them valuable material previously buried in specialized journals.
Despite the growing appeal of exact statistical methods, beginner-level material on the subject is hard to come by. This book introduces the theory, techniques, and applications of exact statistical inference for categorical data. Using elementary concepts and actual data applications, it brings information formally buried in specialists' journals to the broad audience that needs it. Topics range from the one-binomial case to stratified 2x2 tables and dose-response data. The tools of statistical inference used throughout the book are mid-p values, confidence interval curves, and exact power. To comprehend the material, readers require only a first course in statistics.
"The book's infrastructure makes it a good candidate as a teaching resource. The chapters offer detailed worked example. Furthermore, each chapter includes exercises . . . some practitioners in discrete data analysis also will find this book useful."
- In Technometrics, August 2008, Vol. 50, No. 3