Becoming 95% Confident About Teaching Statistics
(Book Review)

A. Daniel Johnson
Wake Forest University
Winston-Salem, NC
johnsoad@wfu.edu

Review of Intuitive Biostatistics by Harvey Motulsky, MD

(ISBN 0–19–508607–4
PRICE: $34.95 (ppbk)

Students often perform statistical analyses, but what if they have no background, or a flawed view of how to use and interpret the results? How do we teach them without reiterating an entire course, or worse, if we do not really understand statistics ourselves? These are the exact problems addressed by Intuitive Biostatistics.

Dr. Motulsky is not a statistician, but a medical educator who limits theory in favor of teaching “consumers of statistics” practical skills. Concise chapters discuss standard error, survival curves, hypothesis testing using t–tests and ANOVA, Chi–square analysis, and regression. However, the similarity to other textbooks ends there. The book has a strong narrative voice that reads like a one–on–one chat with your favorite instructor. Wherever possible, excessive or confusing terminology has been removed as well. For example, the introductory chapter does not contain a single definition. Instead, it explains why we need and use statistics. Equally important to me as an educator, though, is that it gives equal coverage to what statistical tests cannot do, and how they are misused.

Dr. Motulsky develops his readers’ statistical literacy in a fundamentally different way than most textbooks. He begins with 95% confidence intervals, which are used to estimate the margin of error for a dataset. As he describes each statistical method, he explains how to interpret the outcome in these same terms. Although p–values are more commonly used (and misinterpreted), they do not appear until Chapter 10, and even then they are defined relative to 95% CI. This approach eliminates “p<0.05” as the ultimate indicator of success or failure for experiments. Instead, a reader learns to interpret p–values as shorthand for confidence or margin of error, a key step towards a more intuitive understanding of the limits of all statistical analyses.

The book has several other useful features. Footnotes indicate which advanced topics can be skipped without losing continuity, or are less relevant to basic scientists versus clinicians. Summary chapters explain how to minimize confounding variables, choose the best test, and avoid common statistical mistakes. Unlike many authors, Dr. Motulsky embraces the use of computers to perform routine calculations, and he lists common commands for MS Excel in a handy appendix.

I highly recommend Intuitive Biostatistics to anyone who needs a concise, practical introduction to statistics for their students. I used its organizational structure and several of the examples in a statistics primer I wrote for our departmental lab manuals. Students have come back repeatedly and told us how useful they found this primer for other courses. I routinely loan my copy to graduate teaching assistants who need a crash course or refresher before they enter a classroom. This book also would be my first choice for teaching statistics as part of a research methods course for undergraduates.