PREFACE

 Introduction to the Practice of Statistics (IPS) is an introductory text that focuses on data and on statistical reasoning. It is elementary in mathematical level, but conceptually rich in statistical ideas and serious in its aim to help students think about data and use statistical methods with understanding. Although the first edition of IPS was a somewhat radical departure from the then-standard course, which emphasized probability and inference, this third edition now represents the current standard, in which data analysis, design of data production, and the demands of statistical practice join probability-based inference as foci for study.

Statisticians have, in fact, reached a general consensus on the nature of first courses for general college audiences. As Richard Scheaffer says in discussing a survey paper by one of us, "With regard to the content of an introductory statistics course, statisticians are in closer agreement today than at any previous time in my career."1 IPS is one expression of that consensus, the principles of which have been summarized by a joint committee of the American Statistical Association and the Mathematical Association of America as follows:

In this preface we first briefly describe our philosophy and then discuss new features of the third edition. The title of the book expresses our intent to introduce readers to statistics as it is used in practice. Statistics in practice is concerned with gaining understanding from data; it focuses on problem solving rather than on methods that may be useful in specific settings. A text cannot fully imitate practice, because it must teach specific methods in a logical order and must use data that are not the reader's own. Nonetheless, our interest and experience in applying statistics have influenced the nature of IPS in several ways.

Statistical Thinking Statistics is interesting and useful because it provides strategies and tools for using data to gain insight into real problems. As the continuing revolution in computing automates most of the tiresome details, an emphasis on statistical concepts and on insight from data becomes both more practical for students and teachers and more important for users who must supply what is not automated. No student should complete a first statistics course, for example, without a firm grasp of the distinction between observational studies and experiments and of why randomized comparative experiments are the gold standard for evidence of causation. We have seen many statistical mistakes, but few that simply involved getting a calculation wrong. We therefore ask students to learn to explore data, always starting with plots, to think about the context of the data and the design of the study that produced the data, the possible influence of wild observations on conclusions, and the reasoning that lies behind standard methods of inference. Users of statistics who form these habits from the beginning are well prepared to learn and use more advanced methods.

Data Data are numbers with a context. The number 10.3 alone is meaningless. Hearing that a friend's new baby weighed 10.3 pounds at birth, however, engages our background knowledge and brings immediate meaning. Note that the weight could not plausibly be 10.3 ounces or 10.3 kilograms. Because context makes numbers meaningful, our examples and exercises use real data with real contexts that we briefly describe. Calculating the mean of five numbers is arithmetic, not statistics. We hope that the presence of background information, even in exercises intended for routine drill, will encourage students to always consider the meaning of their calculations as well as the calculations themselves. Note in this connection that a calculation or a graph is rarely "the answer" to a statistical problem. We strongly encourage requiring students to always state a brief conclusion in the context of the problem. This helps build data sense as well as the communications skills that employers value.

Mathematics Although statistics is a mathematical science, it is not a field of mathematics and should not be taught as if it were. A fruitful mathematical theory (based on probability, which is a field of mathematics) underlies some parts of basic statistics, but by no means all. The distinction between observation and experiment, for example, is a core statistical idea that is ignored by the theory.3 Mathematically-trained teachers, rightly resisting a formula-based approach, sometimes identify conceptual understanding with mathematical understanding. When teaching statistics, we must emphasize statistical ideas and recognize that mathematics is not the only vehicle for conceptual understanding. Introduction to the Practice of Statistics requires only the ability to read and use equations without having each step parsed. We require no algebraic derivations, let alone calculus. Because this is a statistics text, it is richer in ideas and requires more thought than the low mathematical level suggests.

Calculators and Computers Statistical calculations and graphics are, in practice, automated by software. We encourage instructors to use software of their choice or a graphing calculator that includes functions for both data analysis and basic inference. IPS includes some topics that reflect the dominance of software in practice, such as normal quantile plots and the version of the two-sample t procedures that does not require equal variances. All students should have at least a "two-variable statistics" calculator with functions for correlation and the least-squares regression line as well as for the mean and standard deviation. Although not all exercises are feasible for students with such a calculator, we have taken care to make the book usable by students without access to computing.

Judgment Statistics in practice requires judgment. It is easy to list the mathematical assumptions that justify use of a particular procedure, but not so easy to decide when the procedure can be safely used in practice. Because judgment develops through experience, an introductory course should present clear guidelines and not make unreasonable demands on the judgment of students. We have given guidelines---for example, on using the t procedures for comparing two means but avoiding the F procedures for comparing two variances---that we follow ourselves. Similarly, many exercises require students to use some judgment and (equally important) to justify their choices in words. Many students would prefer to stick to calculating, and many statistics texts allow them to. Requiring more will do them much good in the longer run.

Teaching Experiences We have successfully used IPS in courses taught to quite diverse student audiences. For general undergraduates from mixed disciplines, we cover Chapters 1 to 8 and one of Chapters 9, 10, and 12, omitting all optional material. For sophomores planning to major in actuarial science or statistics, we add Chapters 10 and 11 to the core material in Chapters 1 to 8 and include most optional content. We deemphasize Chapter 4 (probability) because these students will take a probability course later in their program, and we make very intensive use of software. The third group we teach contains beginning graduate students in such fields as education, family studies, and retailing. These mature, but sometimes quantitatively unprepared, students read the entire text (Chapters 11 and 13 lightly) with little emphasis on Chapter 4 and some parts of Chapter 5. In all cases, beginning with data analysis and data production (Part I) helps students overcome their fear of statistics and builds a sound base for studying inference.

The Third Edition

In revising Introduction to the Practice of Statistics after two successful editions, we have concentrated on helping students read the book and on maintaining close ties to statistical practice.

Organization

The table of contents for the latter part of the text has been revised. The new contents makes it clearer that instructors can choose from among the material after Chapter 8 by making each natural unit into a separate, shorter, chapter. This also allows a three-part structure that makes the major divisions of the book more apparent.

There are some changes in scope and sequence. For example, correlation now precedes regression in Chapter 2. Correlation is a somewhat more general topic, because it does not require the distinction between explanatory and response variables. More important, the new order allows the least-squares regression line to be described in terms of the means and standard deviations of the two variables and the correlation between them. A completely new treatment of conditional probability (Chapter 4) and of inference for two-way tables (Chapter 9) are other examples of major changes.

Accessibility

We have carefully rewritten the text to improve readability without losing the conceptual richness and real-problem settings that characterize IPS. In particular, we have provided more structure for readers via more frequent subsection headings and by bringing more material out of the flow of the text into boxes and other displays.

Technology

We assume better technology---at least a "two-variable statistics" calculator. Calculation recipes intended for basic calculators have disappeared. Our principle is that if a formula is "just a rule," it should be automated. For example, equations for the slope of the least-squares regression line based on sums of squares provide no insight to students and are just rules. No such formula appears in this edition: they are replaced by a button on a calculator or a menu item in software. On the other hand, the equation b = r sy/sx for the regression slope does provide insight. We discuss it carefully and expect students to know the equation and its implications. Some exercises in Chapters 4 and 5 now suggest simulations that help students understand probability.

Statistical Output

We recognize the diversity of technology and prevalence of graphical interfaces. There are no longer any Minitab commands in the text, because these are now menu items in the software. We include examples of output from a wider range of software, adding the TI-83 graphing calculator and the Excel spreadsheet to statistical software such as Data Desk, Minitab, and SAS. The purpose of displaying output is primarily to convince students that they can read and use the information provided by any of these platforms once they have mastered the ideas and terminology in the text.

"Beyond the basics" Sections

We have added short subsections entitled "Beyond the basics" to most chapters. These briefly introduce somewhat more advanced material that is now becoming standard in statistical practice. Although this material is considered "enrichment" rather than required learning, instructors using software that implements these methods may wish to ask students to use them regularly. It is often easy, for example, to ask for a density estimate rather than a histogram or to employ a scatterplot smoother to summarize the overall pattern of a plot.

New Exercises and Examples

Many new exercises and examples based on new data sets have been added. Weaker or outdated exercises have been replaced and the overall exercise count has modestly increased. More exercises assume automated calculations and graphics, so that the separate "Computer Exercises" of previous editions are no longer needed. There are more exercises and examples from business and finance, so that the content reflects the full range of statistical applications.

Data Sets

The data from exercises and tables have been greatly improved and are available to instructors on a dual-platform CD-ROM included with each copy of IPS.

Additional Chapters on CD-ROM

In writing IPS we have concentrated on the material we think most valuable in a first course. Our criteria were direct usefulness in practice and the extent to which studying this material prepares students to go on to more advanced statistical methods. We think that encyclopedic introductory texts intimidate students and that very brief coverage of many methods reinforces students' perception that statistics consists of recipes. The new CD-ROM that accompanies each book allows us to offer additional (optional) material without violating these principles. The CD-ROM contains two new chapters, complete with exercises. Chapter 14, Nonparametric Tests, presents the most common rank tests: Wilcoxon-Mann-Whitney, Wilcoxon signed rank, and Kruskal-Wallace, along with comments on their use in practice. Chapter 15, Logistic Regression, introduces a topic that is very prominent in statistical practice and that we would like to see become more common in beginning instruction.

Supplements

A full range of supplements is available to help teachers and students better teach and learn from IPS. Several complimentary supplements are available for adopters of the third edition of IPS, including:

And for Students...

Acknowledgments

We are pleased that the first two editions of Introduction to the Practice of Statistics have helped move the teaching of introductory statistics in a direction supported by most statisticians. We are grateful to the many colleagues and students who have provided helpful comments, and we hope that they will find this new edition another step forward. In particular, we would like to thank the following colleagues who offered specific comments on the new edition:

John Barnard, Harvard University
Tim Brown, University of Melbourne
Patricia Buchanan, Pennsylvania State University
Easaw Chacko, University of Canterbury
Smiley Cheng, University of Manitoba
Brian Corbett, The Open Polytechnic at New Zealand
Constance Eshbach Cutchins, University of Central Florida
Roy V. Erickson, Michigan State University
Kenneth Fairbanks, Murray State University
Lloyd Gavin, California State, Sacramento
Mark E. Glickman, Boston University
Joel B. Greenhouse, Carnegie Melon University
Miriam Schapiro Grosof, Stern College
Robert Hannum, University of Denver
W. L. Harkness, Pennsylvania State University
I. Hibschweiler, Daemen College
Elizabeth A. Houseworth, University of Oregon
John Z. Imbrie, University of Virginia
T. Henry Jablonski Jr., East Tennessee State University
Dave Jobson, University of Alberta
Gerald Kaminski, University of St. Thomas
Douglas Kelker, University of Alberta
Eugene Klimko, Binghamton University
Marlene J. Kovaly, Florida Community College at Jacksonville
Don O. Loftsgaarden, University of Montana
George Marrah, Clemson University
Brendan McCabe, University of British Columbia
Kip Murray, University of Houston
Tom O'Bryan, University of Wisconsin, Milwaukee
Richard Pulskamp, Xavier University
Robert L. Raymond, University of Minnesota
Raman Patel, California State University, Chico
Larry J. Ringer, Texas A & M University
Neal Rogness, Grand Valley State University
Kenneth A. Ross, University of Oregon
Ronald M. Schrader, University of New Mexico
Al Schainblatt, San Francisco State University
Patty Solomon, University of Adelaide
Bert Steece, University of Southern California
Robert K. Smidt, California Polytechnic University
Paul Stephenson, Grand Valley State University
Linn M. Stranak, Union University
Thaddeus Tarpey, Wright State University
David van Dyke, Harvard University
Chamont Wang, The College of New Jersey
Peter Westfall, Texas Tech University
William H. Woodall, University of Alabama
Alan Zaslavsky, Harvard University

Most of all, we are grateful to the many people in varied disciplines and occupations with whom we have worked to gain understanding from data. They have provided both material for this book and the experience that enabled us to write it. What the eminent statistician John Tukey called "the real-problems experience and the real-data experience" has shaped our view of statistics. It has convinced us of the need for beginning instruction to focus on data and concepts, building intellectual skills that transfer to more elaborate settings and remain essential when all details are automated. We hope that users and potential users of statistical techniques will find this emphasis helpful.

Notes

  1. D. S. Moore and discussants, "New pedagogy and new content: the case of statistics," International Statistical Review, 65 (1997), pp.123--165. Richard Scheaffer's comment appears on page 156.

  2. These are main heads in a brief version of the committee's report endorsed by the Board of Directors of the American Statistical Association. The full text appears on page 127 of the article cited in Note 1.

  3. A detailed discussion appears in G. Cobb and D. S. Moore, "Mathematics, statistics, and teaching," American Mathematical Monthly, 104 (1997), pp. 801-823.

[TOP]
Moore & McCabe, Introduction to Practice of Statistics, 3rd Edition, 1998
W. H. Freeman & Co. and Sumanas, Inc.