Probability , Statistical Optics , and Data Testing : A Problem Solving Approach ( 3 rd ed . )

Chapters 7–10 cover the core of the basic statistical methods taught in an entry-level, one-semester course. Topics covered include point estimates, interval estimates, and hypothesis testing for continuous variables and proportions. Again, what distinguishes this book is the breadth and depth of the coverage. Examples of this include the effective explanation, with examples, of the use of maximum likelihood and method of moments for parameter estimation. Besides discussing type I and II errors, the authors spend a good deal of time explaining statistical power, OC curves, how type II errors are calculated and, most importantly, how to make sample size calculations for a number of tests. Goodness of Ž t is covered not only for its use with contingency tables, but also for testing the validity of any purposed distribution function. This is all useful stuff. Chapters 11–16 deal with material usually taught (if at all) in the second semester of an introductory course or in a second-year course on more advanced topics. The subjects covered are simple and multiple linear regression (11 and 12), one-way ANOVA for Ž xed and random effects (13), design of experiments (14), nonparametric statistics (15), and SQC (16). Chapters 11–15 are a veritable tour de force through their respective subject matters. I think that these chapters border on too much too fast (more on this below). Chapter 16 slows down and ends with some perspectives on the use of statistical methods in corporate America. No reference is made to 6-Sigma or any other corporate-wide program. I think this is wise, because corporate initiatives come and go but basic, sound statistical methods have a much longer life time (I Ž rst learned about DOEs and control charts during the total quality heyday of the mid 1980s). Answers to most of the odd-numbered problems are included in an appendix. Other appendixes contain a good bibliography for further study and a glossary of terms. The book comes with a CD that contains a complete e-copy of the text plus more. The “more” includes complete datasets for some of the examples and problems in the text, completely worked out solutions to a number of the problems, and topics not covered in the paper version. The latter are usually more advanced topics like bootstrapping methods, a proof that the sample standard deviation is biased, lack-of-Ž t tests, ridge regression, and nonlinear regression models to name a few. I have only one non “nitpicky” issue with this book. Sometimes the authors get too technical for the intended reader. It is as if they switch and address a more academic, or at least more statistically savvy audience. For example, consider their deŽ nition of a random variable: “A random variable is a function that assigns a real number to each outcome in the sample space of a random experiment” (p. 54). True, but unless the student is a math major, more explanation than the authors provided is needed to make the intent of this deŽ nition clear. In the Ž rst 10 chapters, this is an episodic problem. In Chapters 11–15 the question of whom they are writing to becomes more of an issue. The authors wrote their book under the assumption that this may be the only course in statistics that the reader will have. Consequently, they Ž lled it with most of what a practicing engineer will need to know. There is indeed a rich array of tools presented in these chapters. For example, the two chapters on linear regression cover variance in ation factors, Mallows’s C O p, the Cook distance, the PRESS statistic, regression with dummy variables, and a host of model-building methods, all very useful and most of them typically taught in a subsequent course devoted to linear regression models. The same comments can be made about the other chapters in this section. The downside is that the background needed to understand the material is presented at a level and a pace that I think will be beyond most of the intended students. The variance estimators, some of the matrix algebra, the partitioning of sums of squares, and degrees of freedom come pretty fast and furious. The presentation is concise and precise to the point at which much of the pedagogical hand-holding appropriate to an introductory text is missing. It is as if the primary intent in these chapters is to write a highly annotated reference book. For a general practitioner of statistics (like this reviewer), the material in these chapters is excellent. For engineers taking their Ž rst course in statistics, I fear this too dense. In short, I wish they would slow down in these chapters and spend some more words on explaining “physically, what does this mean and why is it important?” Having said this, I would also add that in the hands of a good teacher, the necessary embellishments could be provided in the classroom that will bridge the “split audience” issue. This book is as applied and thorough a preparation for the real engineering world as can be expected in a one-year course. The only important material missing is a chapter on reliability modeling. (Reviewers of the previous two editions also made this recommendation.) I think this book will be a long reach for students taking their Ž rst course in statistics; but if they can stay the course, they will be far up on the learning curve for the application of statistics to problem solving. This book will also serve as an excellent reference for the practicing industrial statistician.