Skip navigation

Book Reviews

Last night they got the elephant!

Against the Gods. Peter L Bernstein. John Wiley. ISBN 0-471-12104-5. 337pp. UK£17.99.

One night during the Second World War, in one of the many air raids on Moscow, a distinguished professor of statistics appeared in the local air-raid shelter. He had never been there before, because he used to say that there were seven million people in Moscow, so why should he expect to be hit. His friends asked about the change of heart. "Look," he explained, "there are seven million people in Moscow and one elephant. Last might they got the elephant".

That story could be taken as encapsulating all our problems with risk. But there are more than you think. Peter Bernstein's book is a tour-de-force through numbers, risk and how people perceive and respond to it, investment and the way markets work. It is highly readable and full of great quotes.

Best of all, though, is that this book will explain things in ways that are so different that you will understand things you thought beyond you. Bet you don't believe that. Here's a test - hands up all those who think they understand regression to the mean. Not many of you. But go to the chapter entitled "The man with the sprained brain", and you will. It might just change your life if you play the markets.

Being right

Bandolier particularly liked its introduction to the "Law of Large Numbers". A brainchild of Jacob Bernoulli (one of a tribe of gifted philosophers and mathematicians, not the man who invented the disc drive), this is statistics the other way round. Usually statistics tell us how likely a result is to be wrong: the Law of Large Numbers is more concerned about how likely it is to be right.

Think about this for a moment. What do we want with the result from a trial or meta-analysis? Actually we want to know how likely we are to be right, especially if we sit down for a moment to consider just how right we have to be for any particular circumstance. There is a risk with this book: it might change the way you think about research.

Getting the best from the literature

How to read a paper. Trisha Greenhalgh. BMJ Publishing. ISBN 0-7279-1139-2. 192pp. UK£14.95.

In the introduction Trisha Greenhalgh remembers ward rounds with a distinguished professor, whose clinical acumen and memory were second to none. But that had taken 40 years to acquire. Dipping into the medical literature for those of us (the vast majority) who have never done enough research to be familiar with all the nuances and "tricks of the trade" can be a daunting task.

We hear that much research is not scientifically sound. We know that some research can be biased. We don't know how to make sense of a randomised trial compared with a retrospective case series. Sensitivity and specificity of diagnostic tests are beyond us. And statistics - give us a break!

Not any more. No excuses - because anyone with two neurones to rub together can be at least in touch with a modern scientific paper after putting in some time to read Trish Greenhalgh's superb guide to reading and understanding papers and what goes into them. It is an easy read, and an enjoyable one (we loved the box on page 70 on how to cheat on statistical tests when writing up a paper, and that on page 88 on how companies sometimes over-egg the pudding when marketing their products).

Bandolier has spent many hours reading, and much harsh handling from trial and error to learn only part of the good common sense that is in this book. It will be useful for the "first-year medical student and the grey-haired consultant". Don't just take our word - those are the words of Professor Sir David Weatherall in the foreword!

previous story in this issue