Article Text

Basic Skills in Statistics
Free
1. J N V Miles
1. Department of Health Sciences, University of York, York YO10 5DD, UK; jnvm1york.ac.uk

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

A Cook, G Netuveli, A Sheikh. London: Class Publishing, 2004. 85 pp. ISBN 1 8595 9101 9

What is the purpose of books on statistics written for non-statisticians? If one read many introductory statistics books, one might assume that the answer is “to raise the reader to a level of expertise sufficiently high that they are able to apply and understand a range of statistical techniques”—in other words, to do without a statistician (at least some of the time). However, to quote Will Rogers: “It isn’t what we don’t know that gives us trouble, it’s what we know that ain’t so”.

This book does (and claims to do) no such thing. The (self-stated) aim of the book is to help clinicians to understand enough about statistics to understand papers and to “facilitate the discussion of their ideas with a statistician”. As such, the book is light on formulae and equations—although not completely free of them—and does not contain instructions on how to analyse different statistical problems using software. It only tells you how to use a fairly limited range of statistical techniques—you won’t get much beyond a t test, χ2, or Cohen’s kappa.

The book does exactly what it says on the cover—it encompasses basic skills in statistics. Chapter 1 looks at issues such as measurement and probability and introduces the idea of a probability distribution; chapter 2 examines the univariate description of a single variable, covering measures of average, dispersion, and distributions, and chapter 3 discusses how to link two variables where you will find the formula for the phi correlation (for dichotomous variables) but not the Pearson correlation for continuous variables. If one were to read the book from cover to cover one might be confused by the use of confidence intervals and probability values in this text which are covered in chapter 4. While the first four chapters build on one another, the final three cover different material and do not build on one another in the same way. Chapter 5 looks at study design including sample size calculations, chapter 6 describes the principles of meta-analysis, and a brief chapter 7 looks at data management with some suggestions for software.

The book started life as a series of papers published in Primary Care Respiratory Journal and, as such, there are some minor problems of “flow” through the book. Given that few readers will sit and read a text on statistics from cover to cover, I do not think that this is an issue to be concerned about. Given the breadth and depth of material that there is to include in a text such as this, there will always be disagreements about what should and should not be included. All my quibbles would be minor and barely worth mentioning: I think the authors are a little too enthusiastic in their recommendation of Bonferroni correction and I would strongly advise against attempting to do statistical tests in Excel without a good idea about its shortcomings.

The book admirably succeeds in its aims. As someone who spends considerable time advising practitioners who are carrying out research, I would be very happy if everyone who came to see me had read this book.