Naked Statistics: Stripping the Dread from the Data

Charles Wheelan W.W. NORTON & COMPANY LTD 2013, 320 PAGES PRICE (HARDBACK) £18.99 ISBN 978-0-39307-195-5

Charles Wheelan is also the author of the bestselling Naked Economics – so I suspected I was definitely in for a good read.

I have taught A-level Statistics, and more recently Advanced Placement Statistics, so I am not really new to the subject. I confess, reading the earlier chapters, I was a little hungry for more mathematics and more detail. However, I gradually learned to suspend my own prior knowledge, and read this from the perspective of a reader approaching the subject (in any depth) for the first time - which is of course, the likely background of the intended audience.

I was immediately impressed with the wealth of accurate analogies Mr. Wheelan uses to explain the key broad concepts of the Central Limit Theorem, Significance and Hypothesis Testing. Soon, I was able to reflect upon how strong and powerful these analogies were, for there is virtually no mathematical formulae at all, in the chapters, and precious little in the appendices for each chapter.

The latter half of the book contains discussion of these considerations, from several very high-profile and significant trials and experiments, ranging from longitudinal studies on various aspects of health, to whether attendance at an Ivy League school improves your earning potential. The discussion of topics including confounding factors, too much data, correlation vs causation and sampling bias is absolutely riveting.

Does increasing the amount of mandatory schooling improve mortality? Does increasing the size of a police force lower the incidence of crime? How would you even attempt to collect relevant data?

This book will not turn you into a competent statistician, but it will do something far more important. It will make you aware of the wealth of potential that ‘good’ statistics can bring to an experiment, but more crucially, it will really make the reader aware of just how important good experimental design is, and how important good sampling is.

There are a lot of really well-written ‘mathematical’ statistics textbooks, which do a terrific job of explaining the math behind statistical inference. However, in my experience, they all fall a little short on stressing the importance of experimental design and sampling. This book fills that gap wonderfully. I am so very impressed with this book, that I shall cite Naked Statistics as a recommended text for my AP students.

Andrew S Jones CMath MIMA CSci