Seeing Through Statistics (with CD-ROM and InfoTrac®) [Paperback]

Hi...I used this book in an upper division math course at the University of New Mexico and loved it. It is a good book for people who need to understand and use statistics but who don't have a mathematics/statistics degree. The book emphasizes data collection techniques as well as methods of statistical calculation and ample examples are given for every topic in order to make concepts more concrete. I currently use this book as a quick reference guide and, as long as I'm not needing a rigorous development of a concept, it is perfect!

This book consists of 26 chapters divided into four parts. Part One, Finding Data in Life, provides most of the necessary tools for the reader to become a skeptical consumer of statistics. In it, chapter six, Getting the Big Picture, provides the reader with a seven step method for evaluating statistics-laden news articles is sufficient for the development of a healthy skepticism of numbers in the news. Part Two, Finding Life in Data, provides some of the simpler, more common tools that the reader can use, albeit under limited conditions, in order to appropriately interpret statistics quoted by the media. Unfortunately, I noticed that the author used none of these important tools in the examples she provides in Part Four of the text. Part Three, Understanding Uncertainty in Life, imparts the fundamentals of probability theory, and addresses most of the more popular and common uses and abuses of probability. Part Four, Making Judgments from Surveys and Experiments, attempts to develop within the reader a functional understanding of confidence intervals, hypothesis testing and statistical significance, with less than acceptable results.

One very good point of the text is its discussion of the difficulties and disasters associated with defining and measuring characteristics of interest, obtaining good survey and experimental samples, conducting observational and experimental studies, and establishing cause and effect relationships. Another good point about the text is its presentation of confidence intervals for sample means and proportions, and its discussion of the limitations of statistical inference (moving from samples to conclusions about populations). One minor flaw of the text is its use of approximate, or back-of-the-envelope statistical calculations which are only valid with bell-shaped samples each consisting of at least one hundred observations. However, more exact calculations would require a statistics text complete with t-tables, and is beyond the scope of a text directed at promoting a more general statistical literacy.

Two main problems exist with the text. The first is that it does not stress the importance of going to the source of a quoted statistic. Newspaper articles and TV news items often make compelling cases, but by themselves do not provide conclusive evidence for or against anything of interest. The statistics making the news may have been incorrectly quoted, or more commonly, taken out of their proper context. Only by going to the source can you determine the trustworthiness of the statistic, as this will tell you conclusively whether or not the study adhered to good statistical practice, and most important, whether or not the conclusions made by (or in many cases, implied by) the study are indeed justified. Folks often resort to the use of statistics to prove a point, support a position, or more typically, to get or otherwise encourage the reader to do something that may or may not be in his or her best interests. We often see statistics used to support claims or positions in the health and medical science reporting, as well as in economic and financial reporting. More often than not, what makes it into print for popular public consumption is the sensational, the eye-catching- in other words, only the juiciest numerical morsels make it to print. One small, fast read that explores this particular aspect of numbers in the news is Joel Best's Damned Lies and Statistics.

The second problem is the author's inability to bring together all of the concepts presented in the book in a meaningful way, which can best be demonstrated by an example taken from Part Four of the text. On pages 409-410, Utts argues that an important, but not statistically significant, gender-based salary difference exists among faculty at the school where she teaches, even when seniority levels and department and/or college affiliation is taken into account. Although she disclosed the p-value (0.08) and provided the average difference and sample sizes for one college at her university, any reader who paid close attention to the concepts and lessons presented in the text would notice several problems with this particular example. Utts tells the reader that a regression equation was used to predict female salaries based on knowledge of male salaries, which are then compared to actual female salaries; however, Utts provided neither the scatterplot of the data nor the r-value, both of which would have told the reader the strength of the association between seniority level and pay (as well as the goodness of fit, as there may well be one or two outliers; after all, this is salary data). Nor did she provide the standard deviations so that t-statistics could be calculated and thus allow the reader to make the judgment of `importance' for his or herself. Essentially, Utts gives the reader just enough information to support her position (she seeks to explicitly imply that gender-based pay differences exist at her university), but not enough to confirm or refute her position. Thus, the author behaves in the same way as those axe-grinding individuals that she encourages the reader to be skeptical of.

Closer inspection of this example also reveals a more glaring problem that only a detailed knowledge of the subject matter would impart. For potential users of this book, I provide a couple of interesting insights into this example. First, Utts did not properly define what she meant by salary (was it simply and solely what the faculty received in pay directly from the university, or did it include other sources of income, as professors tend to get other income, such as honorariums- a kind of fee for reviewing grant proposals or serving on editorial boards of journals or even as advisors on government boards and committees, consulting fees and in the case of the author of this text, royalties on all textbooks sold), and second, Utts failed to provide information on the way that salary decisions are made on the campus. This latter point is most interesting, as faculty pay is based on a formula having many factors (seniority being one of them), with each factor carrying different weights. Moreover, formulas differ based on function served or role played, such as teaching and research as opposed to administrative tasks. By way of example, faculty involved in research could earn more than those involved in administration (especially at her college, as it is part of a research-oriented university system), and faculty that emphasize teaching and writing of textbooks (like the author, Utts) would find themselves at a disadvantage to others who emphasized scholarly publications because journal articles get more weight than textbook authorship. In essence, there could well be many reasons for the difference in pay besides gender (which may well be a contributing factor), and I found it not only odd that Utts did not investigate male and female faculty engaged in similar types of work (for example, both teaching faculty or both involved in administration) and then look for differences in pay, but also that the lessons learned in Parts One and Two of this text were not invoked. This, and a few other glaring instances of axe-grinding throughout the text (apparently, the author is a researcher of paranormal phenomena; thus many examples in support of claims of ESP and UFOs find their way into the text) make the text unsuitable as a teaching tool.

In sum, although the book is adequate for courses devoted to satisfying general education requirements, better books pitched at the target audience exist. Two of the more accessible texts on the same subject matter are A J Jaffee's Misused Statistics and David S. Moore's Statistics: Concepts and Controversies. A third text, Statistics by A. Adhikari and R. Pisani also covers the same ground but in a more modular fashion (smaller, bite-sized chapters) and in greater depth, with slightly more mathematical computation and manipulation than most general education requirement students would probably like. All three texts emphasize the importance of taking into account information outside of that presented to the reader and admonish the reader not to jump to conclusions based on `naked' statistics (those divorced from their original context) or otherwise incomplete statistical information. Those looking to obtain satisfactory statistical literacy in a (mathematically) painless fashion would do well to pick up any edition of Moore's Statistics: Concepts and Controversies, and skip this flawed and biased text.

The "search inside this book" feature was not available when this review was posted. Hope it helps.

BOOK CONTENTS

Part I: FINDING DATA IN LIFE.

01. The Benefits and Risks of Using Statistics.

02. Reading the News.

03. Measurements, Mistakes and Misunderstandings.

04. How to Get a Good Sample.

05. Experiments and Observational Studies.

06. Getting the Big Picture.

Part II: FINDING LIFE IN DATA.

07. Summarizing and Displaying Measurement Data.

08. Bell-Shaped Curves and Other Shapes.

09. Plots, Graphs and Pictures.

10. Relationships Between Measurement Variables.

11. Relationships Can be Deceiving.

12. Relationships Between Categorical Variables.

13. Reading the Economic News.

14. Understanding and Reporting Trends Over Time.

Part III: UNDERSTANDING UNCERTAINTY IN LIFE.

15. Understanding Probability and Long-Term Expectations.

16. Psychological Influences on Personal Probability.

17. When Intuition Differs from Relative Frequency.

Part IV: MAKING JUDGEMENTS FROM SURVEYS AND EXPERIMENTS.

18. The Diversity of Samples from the Same Population.

19. Estimating Proportions with Confidence.

20. The Role of Confidence Intervals in Research.

21. Rejecting Chance--Testing Hypothesis in Research.

22. Hypothesis Testing--Examples and Case Studies.

23. Significance, Importance and Undetected Differences.

24. Meta-Analysis: Resolving Inconsistencies Across Studies.

25. Putting What You Have Learned to the Test.

Case Studies.

References.

Solutions to Selected Exercises.

Index.

› Go to Amazon.com to see all 14 reviews  3.6 out of 5 stars