From the reviews:
“This book is dedicated to describing the Bayesian and frequentist regression methods and to illustrating the use of these methods. … This book could be used for three separate graduate courses: regression methods for independent data; regression methods for dependent data; and nonparametric regression and classification. … the book would be a valuable asset for graduate students, researchers in the area of Bayesian and frequentist methods and an invaluable resource for libraries.” (B. M. Golam Kibria, Mathematical Reviews, January, 2014)
This book is a gem. It is a unique modern regression book, because it includes both Frequentist and Bayesian methods for many of the data types encountered in modern regression analysis, generally put one after the other, so that readers can learn about and compare the two approaches immediately. Topics go through and beyond nonlinear mixed models. All the methods are motivated by interesting data sets, and there are many other data sets available from the author’s web site. There is both R and WinBUGS code for everything. The writing is unusually clear: philosophically, about the practical problems, about the development of the methods and in the data analysis, and it also has a strong series of exercises. It serves especially well as a textbook, but can also be used as a methods reference book.
-Raymond J. Carroll, Distinguished Professor, Texas A&M University
Arguably the most important development of the statistical discipline over the last 40 years has been its progressive de-compartmentalisation. In the 1970s, generalized linear models unified a wide range of statistical methods for analysing independently replicated data. In the 1980s, generalized linear mixed models did much the same for dependent data of various kinds, including spatial, longitudinal and genetic settings. Jon Wakefield’s impressive new book takes this process a stage further by recognising that modern computational developments have made models outside the generalized linear class equally accessible, and by taking a refreshingly pragmatic view of different approaches to inference. The book delivers much more than its title suggests – it could very easily be used as the core-text for a year-long masters course in statistical modelling and inference.
-Peter J Diggle, Distinguished University Professor, Lancaster University
Estimating equations, sandwich estimators, Bayesian inference, MCMC and longitudinal models, all in one book. This text is truly unique both for its broad coverage and its pragmatic approach to inference. Wakefield rejects the usual Bayesian-Frequentist divide and instead shows how good data analysis embraces the best of both worlds. To quote the author: "Each of the frequentist and Bayesian approaches have their merits and can often be used in tandem ..." Thorough and clear, this book is a wonderful resource for students and researchers who want a complete and practical understanding of modern regression methods.
-Larry Wasserman, Professor, Department of Statistics and Machine Learning Department, Carnegie Mellon University
From the Back Cover
Bayesian and Frequentist Regression Methods
provides a modern account of both Bayesian and frequentist methods of regression analysis. Many texts cover one or the other of the approaches, but this is the most comprehensive combination of Bayesian and frequentist methods that exists in one place. The two philosophical approaches to regression methodology are featured here as complementary techniques, with theory and data analysis providing supplementary components of the discussion. In particular, methods are illustrated using a variety of data sets. The majority of the data sets are drawn from biostatistics but the techniques are generalizable to a wide range of other disciplines. While the philosophy behind each approach is discussed, the book is not ideological in nature and an emphasis is placed on practical application. It is shown that, in many situations, careful application of the respective approaches can lead to broadly similar conclusions. To use this text, the reader requires a basic understanding of calculus and linear algebra, and introductory courses in probability and statistical theory. The book is based on the author's experience teaching a graduate sequence in regression methods. The book website contains all of the code to reproduce all of the analyses and figures contained in the book.