1st Edition

Understanding Log-linear Analysis With Ilog An Interactive Approach

By Roger Bakeman, Byron F. Robinson Copyright 1994

    Whenever data are categorical and their frequencies can be arrayed in multidimensional tables, log-linear analysis is appropriate. Like analysis of variance and multiple regression for quantitative data, log-linear analysis lets users ask which main effects and interactions affect an outcome of interest. Until recently, however, log-linear analysis seemed difficult -- accessible only to the statistically motivated and savvy.

    Designed for students and researchers who want to know more about this extension of the two-dimensional chi-square, this book introduces basic ideas in clear and straightforward prose and applies them to a core of example studies. ILOG -- a software program that runs on IBM compatible personal computers -- is included with this volume. This interactive program lets readers work through and explore examples provided throughout the book. Because ILOG is capable of serious log-linear analyses, readers gain not only understanding, but the means to put that understanding into practice as well.

    Contents: Preface. Analyzing Quantitative and Qualitative Scores: An Introduction. Basic Statistics for Two-Dimensional Frequency Tables. Models for Two-Dimensional Frequency Tables. Fitting Models for Multidimensional Tables. ILOG Basics: Specifying a Frequency Table. Analyzing Frequency Tables With ILOG. Tallies: Enough, Too Many, None at All. Explicating Results: Percentages, Residuals, and Two-Way Tables. Appendices: Installing ILOG. Critical Values of Chi Square.


    Roger Bakeman, Byron F. Robinson

    "The book and program do the field a great service. They are models of clarity. I have tried comparable runs with both SAS CATMOD and SPSSX HILOGLINEAR, and can report that ILOG far outpaces them for ease of use."
    George W. Howe
    Associate Research Professor of Psychology and of Psychiatry and Behavioral Scie