Book Review
Research Methods 
PDQ Statistics. Third edition.Geoffrey R Norman, David L Streiner. Hamilton, London (ON): BC Decker Inc; 2003. 218 p. CAN$24.95.
Reviewer rating*: Excellent
Review by: Paul Grof, MD, FRCPC
Ottawa, Ontario
This very useful book, appearing in a near pocket-size, is part of a PDQ (pretty darned quick) series offering condensed reviews of various topics. Testifying to its usefulness is the fact that the book is already in its third edition—the only volume in this series to achieve this merit. Another distinction is that one of its authors, Dr David L Streiner, is also the statistical advisor for The Canadian Journal of Psychiatry and is well-known to regular readers of the Journal through his frequent articles on important statistical problems that psychiatrists encounter in research and in the literature.
Unlike other statistical texts, this book is easy to follow because the authors use plain English as much as possible and minimize the use of algebra, calculus, and calculations. They intended to produce a book that helps readers interpret the “results” sections of research articles, and they have achieved their objective admirably. Their explanations help us to understand what researchers are doing and when they are doing it wrong; they provide a service that many clinicians badly need.
This book does not tell readers how to do statistics: its purpose is to help them become critical observers. It exposes readers to the principles of all the statistical methods they are likely to encounter when exploring medical literature. This book helps them to understand what is going on with a particular approach to analysis and will warn them when the author is applying a method incorrectly. After reading this book, a psychiatrist will certainly find the statistics sections of journal articles much less obscure and intimidating. Each chapter builds logically upon previous ones: the book first explains fundamental principles and then evolves gradually to greater complexities. Therefore, the book will be most helpful if one first reads through it completely. Afterwards, when readers encounter a particular statistical method in a journal article, it is easy to return to individual chapters.
The authors stress that the objective of all statistical analysis is to reveal underlying systematic variation in the data set emerging either from experimental manipulation or from the effect of other measured variables. They explain that the strategy of all statistical tests is to compare an observed effect or difference and the anticipated results of random variation.
One particularly valuable aspect of the authors’ sobering approach is their explanation of what statistics cannot do. For example, statistical analysis cannot make up for a defect in study design, significance will not indicate the actual importance of a discovered effect, and statistics will never establish with certainty that one variable causes another. Further, readers must judge for themselves the applicability of some research results to their own setting. Throughout the book, the authors highlight “CRAP” factors: particular areas wherein researchers frequently misuse or misinterpret statistical tests.
This book is a refreshing and sobering approach to learning about statistics. The text, divided in 4 natural parts (“Variables and Descriptive Statistics,” “Parametric Statistics,” “Nonparametric Statistics,” and “Multivariate Statistics”), is clearly written by authors with enormous experience and sensitivity to the problems clinicians encounter when dealing with statistics. Having struggled to understand statistics all my professional life and having taken several not-so-helpful courses, I wish I had discovered this text much earlier. My reading of many journal articles and my evaluation of the related statistical analyses would have been much easier. Throughout this book, one encounters important, simple principles that are apparently crucial for a correct application of statistical methods in psychiatric research but are not seen in other textbooks. Finally, the text is peppered with exquisite wit that makes reading a real delight. I consider this book a gem and a steal at the price.
*Reviewer
Rating Scale/ Échelle d’évaluation du réviseur
Excellent / Excellent
Very Good / Très bon
Good / Bon
Fair / Passable
Not recommended / Pas recommandé
|
