Statistics and Data Analysis: Book Resources
Review from Technometrics
by Melinda Miller Holt of Texas Woman’s University
This textbook is designed to be a one-year, calculus-based coverage of statistics and data analysis for advanced undergraduates or graduate students who have completed a semester of probability and have a familiarity with matrix algebra. Tamhane and Dunlop state that they have attempted to model their work after the following notable textbooks: Statistics, by Freedman, Pisani, and Purves (1998); Introduction to the Practice of Statistics, by Moore, and McCabe (1999); and Mathematical Statistics and Data Analysis, by Rice (1988). I believe that they did, in fact, rise to the high standard that they set for themselves. Statistics and Data Analysis is a readable, well-organized, and interesting textbook. It is incredibly thorough and written at an appropriate level for the target audience. The authors included a wealth of illuminating and motivating examples and exercises.
Of the three textbooks just mentioned, only the one by Rice (1988) is designed for an audience fairly similar to that of Statistics and Data Analysis, so I will briefly compare and contrast the two. Rabinowitz (1989) noted four unique features of the first edition of Rice’s textbook that Ziegel (1995) reiterated in his review of the second edition (Rice 1994). They are worthy of mention here because Tamhane and Dunlop’s work preserves each of them. Paraphrasing Ziegel, they are as follows:
The use of practical applications from many different fieldsI should point out that, although Tamhane and Dunlop do emphasize statistical computing, they have elected not to include any software instruction within the text and few example printouts are included. The authors state in the book’s preface that packages are constantly evolving and that any software instruction would quickly become obsolete. They were right. According to the book’s preface and Web site, one may currently elect to adopt this text bundled with SPSS Student Version 9. Anyone considering it for adoption, however, should be aware that SPSS introduced Version 10 at the end of 1999 and that representatives of the publisher with whom I have spoken are completely unaware of this bundling arrangement.
The generous use of statistical computing in examples and exercises
The introduction to survey sampling before statistical inference
The reduction of detail within some of the proofs (it is often included in optional sections)
Chapter 1: IntroductionMy only complaint is that, in their attempt to write an all-encompassing textbook, Tamhane and Dunlop all but omit Bayesian inference. The brevity of the section, the lack of interesting practical examples, and the near absence of exercises from that section are less than satisfying considering the ever-increasing role of Bayesian methods within statistics. Despite that fact, the authors have done a marvelous job of presenting an enormous amount of material and should be applauded for doing so. They state that one of their goals was to develop both a textbook and a thorough reference book. Tamhane and Dunlop have certainly accomplished that. As the authors note, however, most curricula do not allow for a two-semester course after a prerequisite of probability. As long as that remains true, Statistics and Data Analysis is not likely to be widely adopted. I do, however, hope that I am wrong.
Chapter 2: Review of Probability
Chapter 3: Collecting Data
Chapter 4: Summarizing and Exploring Data
Chapter 5: Sampling Distributions of Statistics
Chapter 6: Basic Concepts of Inference
Chapter 7: Inferences of Single Samples
Chapter 8: Inferences for Two Samples
Chapter 9: Inferences for Proportions and Count Data
Chapter 10: Simple Linear Regression and Correlation
Chapter 11: Multiple Linear Regression
Chapter 12: Analysis of Single Factor Experiments
Chapter 13: Analysis of Multifactor Experiments
Chapter 14: Nonparametric Statistical Methods
Chapter 15: Likelihood, Bayesian, and Decision Theory Methods
Appendix A: Tables
Appendix B: Abbreviated Answers to Selected Odd-Numbered Exercises
Review from the Journal of Quality Technology by Marvin M. Kilgo, III, Carum Technology
Introductory statistics texts vary greatly depending on their intended audience and authors’ pedagogical principles. The new book by Tamhane and Dunlop is primarily a text for upper level undergraduates, and it is intended to provide an introduction to statistical concepts and methodology as well as an overview of applications of the methods. The authors meet these objectives quite well, and they provide a good "broad brush" introduction to statistical practice. The pace is brisk, and a solid calculus background is assumed. The emphasis, however, is not on mathematical rigor but on the application of techniques. Extensive exercises are presented at the end of each chapter, and a disk containing data sets is included with the book. Both examples and exercises are drawn from a wide variety of fields, and many are designed to develop a student’s facility with the use of statistical software packages.
Particularly noteworthy are the extensive discussions of data collection concerns in Chapter 3 and exploratory data analysis and data summarization in Chapter 4. These chapters precede consideration of statistical methodology, and they emphasize the need for careful planning of experimental and analytical studies. Subsequent chapters cover material common to introductory texts, including point estimation, hypothesis testing, regression, and analysis of designed experiments, as well as discussion of essential elements of mathematical statistics which are often omitted from such books. The breadth of topics considered necessarily limits the depth with which many of these issues are explored. For example, "method of moments" techniques are covered in less than two pages; resampling techniques (including permutation tests, bootstrap, and jack-knife methods) are covered in nine pages; and Bayesian methods are covered in four pages. Application of these tools to nontrivial problems would be difficult based only on this material, and, as a result, the exercises take on particular importance. Although brief, these discussions at least introduce the topics and provide references for further investigation.
The book has an attractive layout, and it has a useful companion website containing errata, data sets, and suggested syllabi for use with the text. The graphics in the text are generally well designed and useful. In particular, the discussion of single sample inference in Chapter 7 is enhanced by excellent sketches of the sampling distributions and power curves for tests.
This book would be a good text for an upper division introduction to statistical methods for engineering and science students. The pace is more rapid than in many introductory texts, and thus a broader range of topics can be introduced. Furthermore, the early emphases on data collection and EDA provide a good introduction to statistical practice. While the depth is not as great as that seen, for example, in Box, Hunter, and Hunter (1978), the breadth of coverage and the basic introduction to mathematical statistics will make this text attractive to many readers of JQT.
Review from The American Statistician by Deborah J. Rumsey, The Ohio State University
This book provides more than enough information for a one-year, calculus-based course in mathematical statistics. Although it was written to prepare students to undertake and interpret most standard types of statistical analyses, [it] is also intended to serve as a reference book. A wide range of applications is highlighted, including biomedical sciences, business, education, engineering, physical and chemical sciences, and the social sciences, using real data throughout. A wide selection of exercises is provided at the end of each chapter. Supplementary materials include an instructor’s solutions manual, and a floppy disk containing data files for many of the exercises (available in ASCII, MINITAB, Excel). The corresponding textbook Web site includes errata, supplementary text material, technology resources, compressed data files from many of the exercises (available as ASCII, MINITAB, Excel, SAS, or SPSS) and links to other data resources. (Note: the Web site address for this textbook has changed from what is listed in the preface. The new Web site address as of this writing is: http://cw.prenhall.com/bookbind/pubbooks/tamhane/).
The authors’ intended audience is junior and senior level undergraduates who have had calculus (and preferably a previous course in probability). I think however, that the prerequisite should be a bit stronger: moderate to high levels of mathematical ability and knowledge of probability and statistics are required to follow this text. (In other words I do not think that the title of this book is quite appropriate.) While the authors say that the chapter on probability could be used as a condensed introduction to probability in a combined probability-statistics course, I have to disagree. The probability chapter is written as a review reference only, and would require the instructor to add a great deal of background and motivational material to get students who have never had a course in probability up to speed. The best audience for this book would be junior or senior undergraduate or graduate students who have a strong calculus background and a previous course in probability, such as engineers, biomedical or medical students, MBA candidates, or mathematics/statistics majors. Chapters 1–9 focus on sampling distributions; basic concepts of inference (confidence intervals are presented very briefly); single-sample and two-sample inference; and inference for proportions and count data. The second half of the book covers linear and multiple linear regression, single and multifactor experiments, nonparametric methods, and decision theory methods (Bayesian inference, MLE, etc.). The breadth of statistical content of this book is wide, presenting a thorough list of statistical concepts as well as data exploration and analysis techniques, including some of the more modern, computer-intensive procedures, such as bootstrapping and resampling. Two features that stand out are the early concentration on data collection techniques (Chapter 3) and the thorough treatment of one-and two-variable data analysis before the discussion of sampling distributions and statistical inference.
An interesting sprinkling of interpretational thoughtfulness and advice can be found throughout this text. For example, in the section on summarizing bivariate data, Simpson’s Paradox is discussed, but a means for actually providing a fair comparison of the data is also presented. This is a strong and not often seen quality of a data analysis textbook, and I would like to see it emphasized more.
While I have no doubt that this book will serve students well as a reference book, I have concerns about the presentation of the material within a teaching and learning context. My primary concern lies in the overall presentation and development of the statistical ideas throughout the text; the definitions and theoretical ideas typically come first (often with a long narrative) followed by examples. Sometimes there are no nonmathematical examples at all. (Chapter 6, "Basic Concepts of Inference," is particularly terse.) Although the examples used are generally interesting, they are clearly secondary to the theoretical and mathematical nature of the statistical ideas. This lessens the impact that statistics can have as a discipline, because it fails to focus on the scientific questions that lie at the heart of the issue for our students. Although statisticians are not experts in the fields of application, we can and should develop statistical ideas within a relevant context, not develop relevant contexts to go along with statistical ideas. Perhaps then we would not need the authors to tell students in Chapter 1: "Statistics is an interesting and valuable subject." Students would be able to discover that for themselves.
I think that this text does do a good job of explaining a very wide range of statistical ideas and techniques from a mathematical and analytical standpoint. The authors also try to help develop the students’ data analysis and interpretation skills. I think it will serve as a good reference book, and as a reasonable text for those who already understand why statistics is important, and want to focus on building a strong toolbox for data analysis and interpretation. However, it misses the opportunity to provide this information within a context of inquiry and intuition. Instructors might want to add some intuitive, motivational prefaces to the sections (e.g., bringing the examples forward) and be ready to fill in some of the gaps in students’ backgrounds regarding mathematics and probability where needed.
© 1995-2000 by Prentice-Hall, Inc.|
A Pearson Company