Dooley, David. (2001). Social Research Methods. (Fourth Edition) Reviewed by W. Paul Vogt, Illinois State University

Dooley, David. (2001). Social Research Methods. (Fourth Edition) Upper Saddle River, NJ: Prentice-Hall Pp. xiii + 385. $79     ISBN 0-13-955428-9 Reviewed by W. Paul Vogt Illinois State University July 6, 2001 Dooley's is one of several popular texts introducing students to research methods in the social sciences. Such texts inevitably go over much of the same ground. The interesting differences among them, and the reasons for an instructor to choose one over another, lie in the emphasis on, and the depth of coverage of, the various elements of social research methods. Two broad categories capture most works published in the field of research methods texts: 1) books that treat methodology as a branch of statistics and 2) books that treat statistics as merely one of several groups of methods. Dooley's text is an interesting combination. At first glance it would appear to fit into the category that subordinates statistics to methodology, because statistics properly so called is dealt with in an appendix (pp. 315- 340). In addition, chapter 8 on “Inferential Statistics: Drawing Valid Conclusions from Samples” tries to give the reader the bare-bones minimum necessary set of concepts necessary to read quantitative research articles and to understand p values. This minimalist approach will be too casual for some instructors' tastes, but it has the undoubted advantage of covering a lot of ground in just a few pages. It will also popular with students who have “formula anxiety.” On the other hand, despite his treating statistics with a light hand, the research methods Dooley discusses are all informed by what we might think of as the statistical model of reasoning. However informally, he is thinking of probability distributions and comparing observations to those distributions. This is his mental model of inference, and he uses it very consistently and to good effect. Researchers who are somewhat skeptical of whether experiments are always the royal road to knowledge might cavil, but this is Dooley's position, to the point that he is occasionally guilty of overstating the case: “True experiments generally prove immune to all threats to internal validity” (p. 269, emphasis added). So Dooley's is a quantitatively oriented book, but one that does not pursue in much depth the statistical techniques that undergird most of the methods he discusses. Qualitative methods, by contrast, are dealt with in one chapter of 16 pages that mostly discusses ethnography and participant observation. And even this chapter investigates the validity of and possible advantages of qualitative research mostly using the categories of the statistical model of reasoning. When it is impossible to apply this statistical model, he essentially says, qualitative methods can be a fallback that is better than nothing. Dooley is strong on the logic of investigation and causal inference in the social sciences. His theme is “identifying and avoiding rival explanations” (p. 49) that threaten the validity of research conclusions, especially threats to internal validity. In the debate among teachers of research methods over whether applications or understanding should be emphasized, Dooley's book takes less of a “hands-on” and more of a “brains-on” approach. For example, he has a very sophisticated, though easy to read and brief, discussion of the problems of drawing inferences from cross-sectional and longitudinal data (pp. 242-243). His example discusses the seeming paradox of a direct correlation between literacy rate and birth rate among regions in India but an inverse correlation between literacy and birth rates among individuals living in those districts. An equally impressive brief account, given the complexity of the topic, is his discussion of the comparative value of meta-analysis and traditional literature reviews (pp. 276-277). The discussion is clear, balanced, but not really elementary; it goes beyond the basics without being too difficult for beginning students to understand. Another way to categorize research methods texts is according to the disciplinary background of their authors. As with many writers of methodology texts, Dooley's disciplinary base is psychology, but his book contains a good mix of examples from sociology and economics. His psychology orientation may, however, have led him to at least one questionable conclusion. He states, more than once, that regression toward the mean “derives from measurement error” (p. 210). In many circumstances, this is true; regression toward the mean can be thought of as a psychometric problem of test-retest measurement error, especially when the construct is very difficult to measure. And this may be how regression to the mean is most often encountered in psychology. But this is not its most important meaning in other disciplines such as economics and sociology. Ever since the original studies of height across generations by Galton, regression to the mean has indicated partial causation, not error. The reason Galton's very tall fathers tended to have somewhat shorter sons was not that people were clumsy when they measured height in the 19th century. The reason is that fathers' height is not a perfect determinant of son's height. Unless two variables are perfectly correlated, there will be regression toward the mean, as when parents' education level (measured in standardized units) predicts, but imperfectly, children's education level. In other words, regression toward the mean is imperfect correlation. This is why we symbolize correlation “r” (for regression) not c. On the whole, this is a small point. Such disciplinary differences in statistical method are not usually overwhelming. For the most part, statistics is statistics is statistics. The principles are the same regardless of what is being studied and in what discipline—for the most part. But different disciplines focus on different aspects of the statistical repertoire. Dooley's chapter on program evaluation, which he entitles “applied research,” is very solid. His background in economics and economic psychology shows to good effect here. As with many other sections of the book, Dooley's discussion of evaluation research is easy to read, but not really elementary. He is especially interesting on the use of evaluation research. Much evaluation work seems wasted because not used sufficiently by decision- and policy-makers. If that is the case, then “why continue to conduct” evaluations? Dooley concludes that “we need systematic research on how best to give away evaluation results” (pp. 295- 296). An important positive feature of the book is that each chapter ends with suggestions of very useful web sites students can visit, such as one containing student software for hierarchical linear modeling or another linking to a web-based sample size estimator. Also quite useful in this regard is Dooley's Appendix A on “Social Research and the Internet” (pp. 301-314). While being very up-to-date concerning the use of the Internet, some of the examples of research Dooley uses to illustrate his points are quite dated. I have no problem with this, but I find (impressionistic research only) that my students react quite differently to older studies, even older research examples nicely exemplify a concept. Given students' sometimes negative reaction to “historical” research, were I writing a new edition of a textbook, I would try harder to find more recent illustrations—even when some of my older examples were very apt. Many of Dooley's examples come from research published in the 1960s and 1970s, such as his use of the debate over “white flight” to illustrate problems of interpreting data from time series studies (pp. 217-220). It is a good illustration, but I find that my students often think there is something odd about reading a study published before they were born, even when I try to assure them that it is a “classic.” In sum, Dooley's Social Research Methods covers a great deal of subject matter in a book that is fairly short: 340 pages of text including two appendices, 300 pages without appendices (but the appendices are much too important to skip; they should probably be chapters). Because of its brevity, the book would make a good text to use for instructors who assign supplemental readings and activities. Many of the discussions are quite advanced while being comparatively easy for students to read. Dooley accomplishes this by keeping applications and technical discussions to a minimum and focusing on more conceptual accounts of research methods. About the Reviewer Paul Vogt is Professor of Research Methods and Evaluation in the Department of Educational Administration and Foundations at Illinois State University. He is the author of Dictionary of Statistics and Methodology. Second Edition, 1999. SAGE Publications.