The present text is the result of teaching a third year statistical course to undergraduate social science students. Besides their previous statistics courses, these students have had an introductory course in computer programming (FORTRAN, Pascal, or C) and courses in calculus and linear algebra, so that they may not be typical students of sociology. This course on the analysis of contingency tables has been given with all students in front of computer terminals, and, more recently, micro computers, working interactively with GLIM. Given the importance of the analysis of categorical data using log linear models within the overall body of models known as general linear models (GLMs) treated by GLIM, this book should be of interest to anyone, in any field, concerned with such applications. It should be suitable as a manual for applied statistics courses covering this subject. I assume that the reader has already a reasonably strong foundation in statistics, and specifically in dealing with the log-linearllogistic models. I also assume that he or of GLIM itself. In she has access to the GLIM manual and to an operational version other words, this book does not pretend to present either a complete introduction to the use of GLIM or an exposition of the statistical properties of log-linearllogistic models. For the former, I would recommend Healy (1988) and Aitkin et al (1989). Por the latter, many books already exist, of which I would especially recommend that of Pingleton (1984) in the present context.

Klappentext

This book shows how to apply log linear and logistic models to categorical data using GLIM. Each model is illustrated by a numerical example. All of the necessary programs in the GLIM macro language are supplied, as well as all data for the examples. The material has been the contents of a course for social science students, but would also be useful for applied statistics courses in such varied fields as medicine, geography, economics, biology,... It should also be extremely useful for research workers in these and other fields where such models are applied, since it provides a step by step explanation of how to analyse such data using these models. Almost all of the GLIM macro programs are new and have not previously appeared in the literature. Nor have many of the logistic/log linear models been applied using GLIM before. TOC:Contents: One-Way Frequency Tables.- Time and Causality.- Metric Variables.- Ordinal Variables.- Zero Frequencies and Incomplete Tables.- Patterns.- Appendix I: GLIM Commands.- Appendix II: Data and GLIM Programs for the Examples.- Appendix III: GLIM Macros.- References.

Inhalt

1. One-Way Frequency Tables.- 1. A Time Trend Model.- 2. Further GLIM Instructions.- 3. A Symmetry Model.- 4. Periodicity Models.- 5. Local Effects.- 2. Time and Causality.- 1. Retrospective Studies I.- 2. Retrospective Studies II.- 3. Panel Studies.- 4. First Order Markov Chains.- 5. Second Order Markov Chains.- 3. Metric Variables.- 1. Time Trends.- 2. Model Simplification.- 4. Ordinal Variables.- 1. The Log-Multiplicative Model I.- 2. The Log-Multiplicative Model II.- 3. The Proportional Odds Model.- 4. The Continuation Ratio Model.- 5. Zero Frequencies and Incomplete Tables.- 1. Sampling Zeroes.- 2. Incomplete Tables and Quasi-independence.- 3. Population Estimation.- 4. Social Mobility.- 5. The Bradley-Terry Model.- 6. Guttman Scales.- 6. Patterns.- 1. Extremity Models.- 2. Symmetry Models.- 3. Diagonal Models.- 4. Distance and Loyalty Models.- Appendix I GLIM Commands.- Appendix II Data and GLIM Programs for the Examples.- Appendix III GLIM Macros.- References.