Practical Geostatistics 2000

comparing groups in a multivariate analysis of variance. It is interesting to note that, in both of these particular cases, the authors suggest, after introducing the various criteria for each problem, that if all the various criteria agree, it simply does not matter which you choose. They go on to conclude that, if the criteria do not all agree, one should look for those criteria that converge on similar conclusions. Such practical advice, although I’m sure is routinely followed by experienced practitioners, is rarely found in print. Many authors would rather spend time elaborating on the details of each of the criteria. In this context the introduction of the criteria, along with associated references, seems to suffice quite well. The first chapter is an introduction to the book’s approach, including a justification of their choice of SAS as the statistical package. It would have been useful to supply a general motivational example as an introduction to multivariate analysis in this chapter. The chapter basically parroted the information presented in the Preface. The second chapter is intended as an introduction to basic matrix algebra, but, as these chapters usually are, it is very simplistic and not a substitute for taking a decent course in linear algebra. However, in this book, unlike many others, there seems to be little lost by not having had a course in linear algebra because the reader here is spared most of the mathematical details anyway. Each of the remaining six chapters covers a specific multivariate topic. At the end of each chapter there are problems on which the student can try out the techniques. Each chapter can be read independently of the others. The chapters are (3) “Multivariate Normal Distribution and Tests of Significance,” (4) “Factorial Analysis of Variance,” (5) “Discriminant Analysis,” (6) “Canonical Correlation,” (7) “Principal Components and Factor Analysis,” and (8) “Confirmatory Factor Analysis and Structural Equation Modeling.” The chapters average about 40 pages, including the example. SAS output is explained in some detail with each example. It would have been useful to annotate the actual output; the reader is left bouncing between the tables and the descriptions in the text. The book may seem too heavily dependent on the use of SAS as the package of choice by non-SAS users. In choosing SAS the authors felt that “at times the output from these packages is so similar that only seasoned users are able to distinguish between them.” If the reader has even limited experience with the SAS package, they will be able to follow the syntax used in the examples. However, if the reader is totally unfamiliar with SAS, he/she might better gain some preliminary experience before working on the examples because many of the more operational details of SAS are not covered. Readers using a package other than SAS may even find whole topics unavailable in their package. For instance, canonical correlation analysis and structural equation modeling are not available in MINITAB (Minitab 1997). This book is a good introduction to multivariate statistical methods for someone who wants to understand the basic underpinnings of the aforementioned techniques and try them out on some data. The authors do not claim, or aim, to be a reference for the theory behind the applications. They do, however, provide a great deal of guidance for those people who, with the advent of readily available and very sophisticated statistical software, will be applying these methods to their own data.