Morgan, Susan E., Reichert, Tom, & Harrison, Tyler R. (2002). From Numbers to Words: Reporting Statistical Results for the Social Sciences. Reviewed by Lise DeShea, University of Kentucky

Morgan, Susan E., Reichert, Tom, & Harrison, Tyler R. (2002). From Numbers to Words: Reporting Statistical Results for the Social Sciences. Boston: Allyn and Bacon. Pp. xiii + 125 $26.67       ISBN 0-8013-3280-X Reviewed by Lise DeShea University of Kentucky November 6, 2001 Statistics may be described as the language of research. Budding researchers often have trouble making the translation between this numeric language and the words needed to describe their analyses in a research report. Having experienced this difficulty first-hand in graduate school, the authors of From Numbers to Words: Reporting Statistical Results for the Social Sciences provide verbal templates for translating statistical results into text. Susan E. Morgan, Tom Reichert, and Tyler R. Harrison have drawn examples of well-written results sections from many disciplines within the social sciences – psychology, communication, journalism, political sciences, and so on. These examples demonstrate a variety of statistics, from descriptive statistics through multivariate tests. The authors put forth an important caveat in the preface: This is not a statistics book and cannot be used as a substitute for statistical references and training. Statistics books occasionally are named as references for specific topics (e.g., Siegel & Castellan's 1988 book on nonparametric statistics, and Tabachnik & Fidell's 1996 multivariate statistics text). From Numbers to Words will not help researchers choose the correct statistical analysis for a given research situation, but it should succeed in its goal of helping researchers write a complete results section well. This slender volume is easy to use, clearly written, and well-organized. Readers should take advantage of Chapter 1's summary of the contents and take to heart the statements about the examples from actual journal articles sometimes deviating from APA style and the book's own recommendations. Initially, the reviewer was concerned that these deviations would be a drawback, but in fact the authors do such a good job in pointing out the deviations that readers should have no trouble understanding how their own results should be modified. For example, several examples did not include effect size indicators, which the authors consistently recommend. Chapter 2, "Frequently Asked Questions about Reporting Statistics," opens with a careful explanation of what is a research report and what should be included in a results section. Upon studying the contents of the second chapter, readers should be able to jump directly to the chapter that discusses whatever statistic is of interest. Chapters 3 through 7 each begin with a paragraph describing the contents of the chapter, followed by a list of topics or statistical tests to be described in the chapter. These features contribute to the readability of the text, as well as its usefulness for readers thumbing through it to find a particular statistic or topic. Considering the sometimes broad categories used to organize Chapters 3 through 7 (Descriptive Information, Reliabilities, Correlation, Nonparametric Statistics, and Parametric Statistics), the book is remarkably easy to follow. Newcomers to research will appreciate the clarity of the descriptions in each chapter. Chapter 8 provides explanations of the best use of tables and figures, with reference to the work of Wainer (1997). This chapter could have benefitted from the inclusion of the advice of Tukey as well: "There is no excuse for failing to plot and look" (p. 43, 1977); otherwise, readers may draw the conclusion that tables are better than graphs, simply because they are described in greater detail. But entire books have been written on exploratory data analysis, and the authors' brief summary of how to create commonly used graphs is certainly adequate. An appendix provides a chart summarizing statistics, the details to report, abbreviations, and syntax. At times it seemed as if the authors were operating as guides through a series of caves, shining a light on the path ahead of the readers and offering the advice of a more seasoned explorer in the terrain of research. For example, when explaining how to handle the description of a nonsignificant result, the authors said, "Always remember that good science is transparent, not deceptive." This guiding principle generalizes beyond their immediate example. The authors also provide an excellent point-by-point description of the facts that should be included in a results section – a kind of "connect the dots" between the numbers and the words that will be written to explain the numbers. They remind less experienced researchers that theories are not supported by individual hypothesis tests. They even occasionally provide details about word processors, such as the way to insert Greek letters into the text when using Microsoft Word or WordPerfect. The authors attempt to look ahead and provide some advice on future trends in results sections, such as the importance of including effect size indicators. The reviewer disagrees with their emphasis on tests of assumptions and prefers to recommend robust statistics, but this is a minor quibble. One might expect a future edition to include references to the most recent APA Publication Manual; the authors refer to the fourth edition, yet this book is being released coincidentally with the Publication Manual's fifth edition (APA, 2001). In addition, references to the report of the APA Task Force on Statistical Inference (American Psychologist, 1999) would have been a nice addition to the discussion of trends. This book is not without problems. As a statistics teacher and quantitative researcher, the reviewer cringes at the use of word "accept" being used in conjunction with hypothesis testing (p. 10), as well as the reference (p. 59) to a bad statistic, the Newman-Keuls multiple comparison procedure (MCP) (see Toothaker, 1991). In fact, an example demonstrating the results of an MCP (p. 60) did not explain which MCP was conducted, one of the few cases where the reviewer judged the authors' example to be unacceptable. Also in the context of MCPs, the authors discuss the occasional use of a t statistic in place of "the usual F test;" they must be unaware of the Studentized range statistic q, its relation to the t statistic, and other statistics used in making mean comparisons. A couple of examples included the large-sample version of statistics (Mann-Whitney on pp. 41-42, Kruskal-Wallis on p. 43), so readers trying to use these statistics may be confused between the large- and small-sample statistics. The authors also uncritically presented a chi-square statistic with a large overall sample size and cell frequencies below 5 (p. 39), in addition to a z statistic being shown incorrectly with degrees of freedom (p. 52). In these instances, the reviewer hopes readers will trust a good statistics text in choosing the appropriate statistic and that they will not rely on this book for that purpose. In other cases, the reviewer recognizes her own tendency to be picky about the description of statistical procedures, such as the case on page 51 where the authors discuss the assumptions for parametric statistics. Nonparametric statistics have assumptions, just not about the shape of the distribution. On page 69, semipartial correlations are parenthetically referred to as "variance accounted for per variable." There is a relationship between semipartial correlations and explained variance, but they should not be equated (see Glass & Hopkins, 1996, p. 168, or Hays, 1988, p. 618). With the understanding that this book is a supplement to the APA Publication Manual and statistical training, many novice researchers will find From Numbers to Words to be an illuminating companion as they find their way through the new terrain of writing about statistical results. References American Psychological Association (2001). Publication manual of the American Psychological Association (5th ed.). Washington, D.C.: Author. Glass, G. V. & Hopkins, K. D. (1996). Statistical methods in education and psychology (3rd ed.). Boston: Allyn and Bacon. Hays, W. L. (1988). Statistics (4th ed.). Fort Worth, TX: Holt, Rinehart and Winston Inc. Siegel, S., & Castellan, N. J., Jr. (1988). Nonparametric statistics for the behavioral sciences (2nd ed.). New York: McGraw-Hill. Tabachnik, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). New York: HarperCollins College Publishers. Toothaker, L. E. (1991). Multiple comparisons for researchers. Newbury Park, CA: SAGE Publications. Tukey, J. W. (1977). Exploratory Data Analysis. Reading, MA: Addison-Wesley Publishing Co. Wainer, H. (1997). Improving tabular display with NAEP tables as examples and inspiration. Journal of Educational and Behavioral Statistics, 22, 1-30. Wilkinson, L., & the Task Force on Statistical Inference (1999). Statistical methods in psychology journals: Guidelines and explanations. American Psychologist, 34, 594-604. About the Reviewer Lise DeShea, an Assistant Professor in the Educational and Counseling Psychology Department at the University of Kentucky, is co-author of a statistics textbook and teaches graduate statistics courses. Her research interests include robustness of test statistics, bootstrap procedures, and individual differences in willingness to forgive.