Powers D.A., Xie Y. Statistical Methods for Categorical Data Analysis

Emerald Group Publishing Limited, 1999. — 324 p. — ISBN: 0125637365, 978-0125637367.
At the end of the book there is a list of errors noted.Statistical Methods for Categorical Data Analysis is designed as an accessible reference work and textbook about categorical data (that is, data arising from counts instead of measurement. Examples include data about birth, death, marriage, and so forth). It integrates statistical and econometric approaches to the analysis of limited and categorical dependent variables, thereby offering a practical, mathematically uncomplicated approach to the topics of modern data analysis. The volume offers a comprehensive presentation of many different models in a one-volume format (with website).
Two features distinguish this book from other analyses of categorical data. First, the authors present both the transformational and latent variable approaches and so synthesize similar methods in statistical and econometric literatures. Second, the book has an applied orientation and features actual examples from social science research. The authors keep discussions of theory to a minimum. Key features include: exercises and examples utilize popular data already familiar to many social scientists; examples of the use of various popular software packages; non-standard applications of existing software for estimating models which cannot be handled directly using existing pre-programmed software.Review of Linear Regression Models.
Logit and Probit Models for Binary Data.
Loglinear Models for Contingency Tables.
Statistical Models for Rates.
Models for Ordinal Dependent Variables.
Models for Unordered Dependent Variables.
A. Matrix Approach to Regression.
B. Maximum Likelihood Estimation.Remark. The second edition of this book was published in 2008. It includes new material on multilevel models for categorical data. Several chapters have undergone extensive revisions and extensions to include new applications and examples. Highlights of the 2nd edition include a detailed discussion of classical and Bayesian estimation techniques for hierarchical/multilevel models, extensive coverage of discrete-time hazard models and Cox regression models, and methods for evaluating and accommodating departures from model assumptions. The accompanying website
webspace.utexas.edu/dpowers/www/
contains errata, programming scripts to replicate each example using various statistical packages, which has proven to be an invaluable resource for instructors, students, and researchers.