An Introduction to Generalized Linear Models (Chapman & Hall Statistics Text Series) - Softcover

Dobson, Annette J.

Softcover

ISBN 10: 0412311003 ISBN 13: 9780412311000

Publisher: Springer, 1990

Synopsis

This updated edition provides a unifying framework for many commonly used multivariate statistical methods including multiple regression and analysis of variance or covariance for continuous response data as well as logistic regression for binary responses and log-linear models for counted responses. The theory for these models is developed using the exponential, family of distributions, maximum likelihood estimation and likelihood ration tests. This is followed by information on each of the main types of generalized linear models. The statistical computing program GLIM which was developed to fit these models to data is used extensively and other programs, especially MINITAB, are used to illustrate particular issues. The reader is assumed to have a working knowledge of basic statistical concepts and methods (at the level of most introductory statistics courses) and some acquaintance with calculus and matrix algebra. The main changes from the first edition are that many sections have been extensively rewritten to provide more detailed explanations, GLIM and other programs are explicitly used, and many more numerical examples and exercises have been added. Outline of solutions for selected exercises are given. The methods described in this book are widely applicable for analysing data from the fields of medicine, agriculture, biology, engineering, industrial experimentation, and the social sciences.

About this title

Review

..." a unique and useful text for intermediate undergraduate teaching." -Times Higher Education Supplement "I liked Dobson's basic and relatively brief presentationThanks go to the publisher for the softcover edition and attendant modest price, another of the book's virtues besides its brevity. These attributes make this book a recommended purchase for those who need a book on logistic regression. It is a good place to start." -Technometrics, November 2002

Synopsis

This updated edition provides a unifying framework for many commonly used multivariate statistical methods including multiple regression and analysis of variance or covariance for continuous response data as well as logistic regression for binary responses and log-linear models for counted responses.

Edition details

• Publisher: Springer
• Publication date: 1990
• ISBN 10: 0412311003
• ISBN 13: 9780412311000
• Binding: Paperback
• Language: English
• Edition number: 2
• Number of pages: 188