by Marty Sapp; Springfield, Illinois, Charles C Thomas, Publisher, Inc., 2006, 272 Pages, $62.95
Dr. Fletcher is an associate professor of psychiatry at the University of Massachusetts Medical School, Worcester.
Trudging through Basic Psychological Measurement, Research Designs, and Statistics Without Math, page by agonizing page, brought to mind Dorothy Parker's remark in her Constant Reader review of A. A. Milne's The House at Pooh Corner that at a certain point in the reading, "Tonstant Weader Fwowed up." Although I never quite reached that point, it is just as well that this text is not titled Statistics With No Pain because some readers may require oral surgery brought on by the incessant teeth gnashing that precedes the masticating and digesting of each grueling page.
The title suggests that Marty Sapp's laudable objective in writing this text was to make research design and statistics more accessible to the student. It is unfortunate that he failed on every possible level. Sapp's writing style can only be characterized as turgid obfuscation. He apparently believes that peppering the pages with a plentitude of the conjunctive "and" magically makes the meaning more accessible to the reader. Nothing could be further from the truth. At the very least, overly liberal use of "and" leads to what my high school grammar book called "stringy" sentences. Grammatical errors and disordered presentation of ideas also contribute to the reader's confusion.
The text might best be characterized as headline statistics. Sapp provides many names of statistical procedures but few detailed explanations of them. Indeed, he all too frequently provides no explanation at all. Evidently his attitude is that explanations can always be found elsewhere, either in the numerous other texts Sapp advises the reader to consult for more detail or perhaps from the hapless teacher who has chosen to teach from this text. In the few instances where Sapp attempts further explanation, the garbled grammar, typos, and generally mystifying presentation of information is bound to leave the reader hopelessly lost. Take this excerpt from the blurb on the back of the book as a warning: "Once exposed to measurement concepts, research design is the next topic. With this area, the topic of variables are covered—dependent, dependent, moderator, control, suppressor, and intervening." Huh?
The book makes much of how it covers a wide variety of topics, from common descriptive statistics and univariate statistics to the more advanced. These topics include multiple regression, log linear regression, multilevel regression, multivariate analysis of variance, discriminant analysis, step-down analysis, canonical correlation, factor analysis, structural equations analyses, and path analysis.
In actuality, multivariate and logistic regression are each covered in two pages, and multilevel and discriminant analyses in one page each. Exploratory factor analysis and structural equations modeling, which includes path analysis, are covered together in two pages. Canonical correlation is covered in two paragraphs in less than a page, and log linear analysis is covered in one paragraph in less than half a page.
Do not torture your students or yourself; use a different text to teach research design and statistics.