Statistics and Data Analysis From Elementary to Intermediate

(1926) landmark paper on DOE as the basis for discussion. This sets the tone for the entire book, which is applications oriented, with an emphasis on classical DOE principles. Readers are assumed to be knowledgeable about basic statistical methods such as conŽ dence intervals, t testing, analysis of variance, and regression. The author does a commendable job of presenting examples within each chapter and in follow-up problems (selected answers provided in the appendix). The majority of the examples come from agriculture and the life sciences, with the remainder being industrial in nature. However, industrial statisticians and experimenters who need a reference on DOE should also consider the newest edition of Montgomery (2000) and the now-classic Box, Hunter, and Hunter (1978). After setting the stage by discussing research design principles, the book gets readers started with completely randomized designs. The middle part of the text covers factorial designs, including material on random effects and mixed models. Then the author provides in-depth coverage of blocked designs. The material on incomplete block design is especially strong. Appendixes to these chapters provide plans for cyclic designs and other options. Of particular interest to industrial statisticians and experimenters will be two chapters (12–13) that cover fractional factorials and response surface designs for process and mixture. [For an in-depth treatment of these topics, see Myers and Montgomery (1995).] A chapter (14) on split-plot design is aimed more at the agriculture area, but it also could be very useful for industrial experiments with hard-to-change factors. Another chapter (16) in the book covers crossover designs for researchers in life sciences who must contend with animal subjects or human factors. This audience will also appreciate an in-depth discussion of repeated-measures designs (Chap. 15). The Ž nal chapter (17) provides useful details on how to remove the in uence of covariates on treatment comparisons in completely randomized and complete block designs. In nearly every chapter, the book provides very useful determinations of design efŽ ciency and/or needs for replication. New material includes coverage of resolvable block designs (Chap. 10), simultaneous conŽ dence intervals (Chap. 3), the generalized linear model (Chap. 4), and Taguchi methods (Chap. 12). The new edition offers a generic ANOVA format and removal of calculator computations, which give the text a cleaner appearance. The author chooses to leave choice of software up to the readers, but in the more difŽ cult exercises he says to use one that is “appropriate.” In the answers to the exercises, one Ž nds a few references to Minitab and SAS. This book deserves consideration for its stated purpose—to present the principles of statistical design and analysis of experiments to graduate students in applied statistics and experimental sciences, especially those who will Ž nd careers in agriculture and life sciences. [A good alternative for this purpose is Oehlert (2000).] Kuehl provides excellent discussions on choosing the proper design for speciŽ c applications. The text maintains a clear distinction between design and analysis. However, the book may not be “user-friendly” for students with majors in the sciences: They need a solid foundation of classes in basic statistics and regression. Industrial statisticians will Ž nd this book to be a useful addition to their libraries of DOE references because it offers details on many designs that are now coming back into vogue, such as incomplete block and split-plot. Industrial experimenters seeking an introductory-level book should look elsewhere, either to Montgomery (2000) or something even simpler.