Models for Discrete Data

Discrete or count data arise in experiments where the outcomes are countable and classified into unique, non-overlapping categories. This book describes the statistical models for evaluating such data. It provides an introduction for graduate students and a concise review for practitioners. The book provides the first in-depth coverage of a number of topics, including the negative multinormal distribution, the many forms of hypergeometric distribution, and coordinate-free models. A detailed treatment of the issues of sample size and power are given in terms of exact inference and asymptotic, non-central chi-squared approximations. Throughout the text, the author interweaves standard statistical software, particularly SAS, with practical examples and current theory.