Reviews of Books and Teaching Materials

Since a large fraction of statisticians use their trade by interacting with nonstatisticians working on applied problems, this book should have broad appeal. The authors have developed a diverse and readable book that tackles many aspects of consulting in good proportion. It describes the consulting process and common statistical methodologies, strategies of effective communication skills, and provides a broad range of case studies as a means of gaining consultingrelated experience. It illustrates what information a consultant needs to obtain from the client, and what motivates those needs. Although there are other books that effectively tackle the individual aspects described above, this book seems to be the one most ideally suited to teaching a well-rounded statistics course at the undergraduate or graduate level. Derr (2000) did a more thorough job of dealing with all aspects of oral and written communication, since that is the primary focus of her book. Boen and Zahn (1982) discussed the business aspects of being a consultant more comprehensively. Spurrier (2000) and Chatfield (1988) gave a more complete overview of statistical methods and the problem-solving aspects of determining an appropriate method for a particular situation. Other sources for case studies, ranging from straightforward to quite advanced, are Holcomb (2002), Peters and Gray (1994), Spurrier (2000), and Andrews and Herzberg (1985). Statistical Consulting gives informative and self-contained discussions for the many aspects of consulting in balanced proportions that would make using the book for a textbook delightfully straightforward. The collection of case studies is diverse in disciplines considered and level of difficulty, and seems to focus on interesting problems that students will find highly motivating. The book is divided into two major parts: The first part, called the Methodology of Statistical Consulting, discusses the communication and statistical skills that a statistical consultant needs to be successful. The second part consists of three collections of increasingly complex case studies. Details of some of the contents of individual chapters are given in the following. Chapter 1 gives an overview of the scientific method and addresses where a statistician should actively seek to contribute. As well, it discusses broad categories of problems commonly found in five major areas where statisticians are typically employed. This chapter gives a context for the discussions in the remainder of the book and does a superior job of giving students an overview of their possible roles in a number of careers. Chapter 2 addresses oral and written communication skills commonly used by consultants. Chapter 3 discusses a wide array of statistical issues and frequently occurring methodologies, with terminology that a moderately sophisticated client would be able to grasp. Statistical methods described range from the elementary to the quite sophisticated. In addition, a good list of textbook references are given to provide a starting point for learning some of the methods presented. Key terminology and ideas of each topic are presented. Given the richness of the list of topics presented, it is clearly beyond the scope of the book to give complete details about any individual topic. The style of presentation of topics is ideal for readers with some general statistical expertise but with no previous exposure to the particular methods. A nice comparison of the statistical software packages SAS and S-Plus is given, outlining some of their relative strengths and weaknesses. The final chapter in part I of the book provides details of the entire consulting interaction between consultants and clients, from initial contact to final written report. Excellent information is given about the first meeting, administrative details, the nature of the correspondence between client and consultant, as well as the thought process of the consultants at various stages of the project. In our graduate-level course in statistical consulting, we frequently teach consulting methods to students who are waiting for their first real-world experience as a consultant, which occurs partway through the course. This chapter will be particularly beneficial to help students to develop more realistic expectations about the consulting process and their roles in it. Part II of the book contains case studies grouped into categories by their level of complexity. Twelve studies with partial analyses are presented with four from each of three categories, including“simple” (contingency tables, surveys, analysis of variance, and observational study assessment), “more-complicated” (regression, logistic regression, response surface methods, and time series analysis) and “research-oriented” (mixed models and some multivariate methods). Sufficient details about the contexts of the problems are given so that the reader has a realistic sense of the importance of each project to the client, the complications faced in analyzing the data, and the requirements for summarizing the results. Unlike so many textbooks, the data in these case studies has the feel of “real data,” with inconsistencies, missing values, and multiple responses. Careful tables at the start of each case study make it easy to get a quick sense of the statistical methods suggested for the analysis, and the nature of the data collected. In addition, there are eight additional partial case studies that present only the setting of the problem. The appendixes contain helpful information grouped into several categories. The resources section has links to Web pages where one can obtain datasets for all of the detailed case studies and some of the partial case studies. A second section has basic help for both SAS and S-Plus. The authors have designed this section to get a beginning user started, but it would need to be supplemented with other readily available resources. Overall, this book is a valuable resource for statistical consultants, both beginning and established. As an instructor of a statistical consulting class, this is a prime candidate for use as a stand-alone textbook in my course, since it contains a desirable balance of materials with statistical methodology, oral and written communication skills, and rich case studies. A sample course outline is given in the book to give suggestions about how to organize a single-semester 15-week project-oriented course around the book. It will make a solid long-term reference for students. Also, for instructors of more traditional senior undergraduate and junior graduate courses, it provides useful case studies to illustrate standard methods in realistic settings that can easily be implemented.