This book describes the principles and techniques needed to analyze data that form a multiway contingency table. Wickens discusses the description of association in such data using log-linear and log-multiplicative models and defines how the presence of association is tested using hypotheses of independence and quasi-independence. The application of the procedures to real data is then detailed.
This volume does not presuppose prior experience or knowledge of statistics beyond basic courses in fundamentals of probability and statistical inference. It serves as an ideal reference for professionals or as a textbook for graduate or advanced undergraduate students involved in statistics in the social sciences.
"…a good introduction to basic concepts and methods for readers with a minimal knowledge of mathematics and statistics….well written….not only an appropriate textbook for upper-level undergraduate and graduate courses taught in social science departments, but also one that we recommend."
—Journal of Mathematical Psychology
"…serves admirably as a text for a graduate-level course on the analysis of complex categorical data by log-linear and related methods."
"The best introduction to log-linear analysis currently available for social scientists who want to both understand the statistical underpinnings and use these powerful analyses effectively."
author of UNDERSTANDING SOCIAL SCIENCE STATISTICS: A Spreadsheet Approach
Contents: Introduction. Two-way Tables. Models for Three-Way Tables. The Statistical Basis of Sampling and Testing. Fitting and Testing Models. Testing Specific Hypotheses. Predictor-Outcome Models. Analyzing Unstructured Tables. Measures of Effect Size. Structurally Incomplete Tables. Descriptions of Association. Least-Squares Models. Ordered Categories. Appendices: Percentage Points of the Chi-Square Distribution. Bonferroni Chi-Square. Power Chart for Chi-Square Tests. Percentage Points of the Normal Distribution. Percentage Points of the Largest Root of a Wishart Matrix. The Greek Alphabet.