Statistics for Evidence-Based Practice and EvaluationPDF
ACP J Club. 2008 Jan-Feb;148:A10. doi:10.7326/ACPJC-2008-148-1-A10
The aim of Statistics for Evidence-Based Practice and Evaluation is “to introduce statistics to undergraduate and graduate students in the helping professions.” The statistical methods are presented from first principles, and the intended audience for the book is very broad.
From the start, Rubin gives a good argument for the use of statistics and why anyone in the helping professions requires some basic knowledge. All chapters have lengthy explanations of the reasons for using the different statistical tools as well as long descriptions of the tools themselves. Each chapter contains “InfoTrac” exercises, using information from resources on the Internet (such as primary research papers), which should encourage the reader to critically examine articles. The downside is that possibly not everyone will have an Internet connection.
The emphasis is on explaining statistical methods through text results in fewer graphs, formulas, and tables than are found in other statistical books. The quality of the information varies. For example, the discussion of the different uses of the mean, median, and mode as location parameters is helpful, while the discussion on normal distributions (Chapter 7) is incorrect, as the distributions referred to are bell-shaped and symmetrical (not all of them follow a normal distribution).
In general, there is an oversimplification in many areas of the book, with formulas not adequately described and calculations not clearly explained. To be fair, a book describing all the caveats for the use of the statistical tests mentioned would be at least twice the size of this.
The book comprises 3 parts. Part 1 gives a good description of how to prepare data for the analysis. This issue is often overlooked in textbooks and is a crucial part of all statistical analysis. Part 2 explains the basic statistical measures and the use of graphs for the presentation of data. Part 3 describes the usual statistical methods used in research, including hypothesis testing, ANOVA, and regression.
This book will be useful to people who know little about statistical methods and who prefer word explanations to graphs or figures. The examples are relatively limited, and the layout (2 columns per page) is clearly designed for text. Readers with a moderate understanding of the methods will probably find the descriptions too long and are unlikely to use it as a reference book.
Rafael Perera, DPhil
Centre for Evidence-Based Medicine