The Subjectivity of Scientists and the Bayesian Approach

industry. It seems that Bayesian methodology has only recently been making inroads into the industrial realm of statistics and data analysis. Lately I have noticed an increase in articles using Bayesian methods appearing in journals like Technometrics and The Journal of Quality Technology. Most practitioners of statistics that I know personally have very little experience with these methods. Nonetheless, many problems that commonly arise in industry, such as dealing with measurement error, missing data, design of experiments involving prior knowledge, and uncertainty in regression coefŽ cients, can be better approached from a Bayesian perspective. Because applying Bayesian methods can be difŽ cult in many real world situations, having software available to perform the computations is a real beneŽ t. A major step toward fulŽ lling this need is the BUGS software, and this book is an invaluable guide to its usage. Therefore, I would recommend this book to any industrial statistician as a good starting point for learning about Bayesian methodology and also to those already familiar with Bayesian techniques as a helpful guide to developing proŽ ciency in using BUGS. To obtain the maximum beneŽ t from this book, it is necessary to obtain install, and learn to use the BUGS software. This will enable the reader to run the approximately 200 examples provided by the author. All of the examples in the book are available via ftp at ftp://www.wiley.co.uk/pub/books/congdon. Installing and running the example programs should not be particularly difŽ cult for anyone with even moderate computer skills. What may be more “challenging,” however, is creating a new model that cannot be patterned after those provided. Assistance is available for those who attempt to do so in the form of an Internet discussion group devoted to the BUGS software. The BUGS website, maintained jointly by the MRC Biostatistics Unit in Cambridge and Imperial College School of Medicine at St. Mary’s, London, contains information on these matters. I would highly recommended that the reader visit this site. Some minor cosmetic problems were evident in the copy of the book that I reviewed. In Chapter 3 there is a gap in the numbering of the sections; that is, Section 3.3 is missing, and Subsection 3.3.3 follows immediately after 3.2.2. Also, in the reference section many of the citations are entered more than once; for example, Carlin and Louis’s Bayes and Empirical Bayes Methods for Data Analysis appears three times.