Doing statistics better

The Art of Statistics by David Spiegelhalter (Pelican Books), reviewed by Marcus Munafò.

It’s fair to say that most of us probably wouldn’t rank statistics as the most exciting part of our training. But statistics underpins much of the research that makes up psychological science – and weak statistics means weak research. We have seen the consequences of this in the recent and ongoing reproducibility debate, within psychology and beyond.

How can we do better? Not least by having a better understanding of what we can do with statistics, and why we might want to use them. This is the starting point for The Art of Statistics, using real-life examples – Could Harold Shipman have been caught earlier? Who was the luckiest passenger on the Titanic? How many trees are there on the planet? – to bring to life the power of numbers to provide deeper insights that help us to understand the world.

The passion that David Spiegelhalter – Winton Professor for the Public Understanding of Risk at the University of Cambridge – has for his subject is clear in this delightfully light-hearted overview of the basics of how and why to use statistics. It is very much written for a general audience, but will be valuable to students, researchers and clinicians in particular, who might have some training in statistics but would still learn much from this book.

Part of the reason for our poor understanding of statistics is that it can be complex. But Spiegelhalter’s knack for explaining difficult concepts clearly and simply means that those who are simply interested in having a better understanding of why statistics matter will be able to enjoy the book (one chapter is flagged as somewhat more technical). But those who are interested in delving deeper can access more technical material held in reserve – not critical to enjoying the book, but adding another dimension for those who want it.

We are introduced to pretty much all of the major statistical concepts, gradually and in a logical order. We move seamlessly from simple approaches to visualising numbers, to statistical inference and Bayesian methods. A core theme is that part of the reason many of us don’t enjoy statistics is simply because it is not taught well – we plunge too quickly into complex subjects like probability, without a good understanding of the basics. The order of chapter topics reflects how we might do better, and serves as a template for statistics syllabus for undergraduates.

Towards the end the reproducibility debate is discussed, with a focus on how we can do better – new methods and approaches that may serve to improve the quality of the work we do, as well as the reporting of it. This, together with a discussion of the historical debates between the founding fathers of modern statistics – Egon Pearson, Jerzy Neyman and, of course, Ronald Fisher – emphasises that statistics is a living, contested discipline.

Statistics can lead us astray if used badly, but used properly can help us answer questions about the world that we could not answer otherwise. This book will help those who read it – from lay person to academic – to do the latter. It deserves to be required reading for all psychologists.

- Reviewed by Marcus Munafò, University of Bristol

BPS Members can discuss this article

Already a member? Or Create an account

Not a member? Find out about becoming a member or subscriber