Overview

Discrete or count data arise in experiments where the outcome variables are the numbers of individual classified into unique, non-overlapping categories. This revised edition describes the statistical models used in the analysis and summary of such data, and provides a sound introduction to the subject for graduate students and practitioners needing a review of the methodology. With many numerical examples throughout, it includes topics not covered in depth elsewhere, such as the negative multinomial distribution; the many forms of the hypergeometric distribution; and coordinate free models. A detailed treatment of sample size estimation and power are given in terms of both exact inference and asymptotic, non-central chi-squared methods. A new section covering Poisson regression has also been included. An important feature of this book, missing elsewhere, is the integration of the software into the text. Many more exercises are provided (including 84% more applied exercises) than in the previous edition, helping consolidate the reader's understanding of all subjects covered, and making the book highly suitable for use in a classroom setting. Several new datasets, mostly from the health and medical sector, are discussed, including previously unpublished data from a study of Tourette's Syndrome in children.