We welcome comments: please click the link at the bottom of each post to add yours.

Wednesday, 2 September 2009

FEATURE: Straight Statistics

Director of Straight Statistics, Nigel Hawkes, on the battle to unearth the truth from the numbers.

As any science writer knows, the world is full of duff statistics, bad surveys, false conclusions, and simple muddle-headedness about numbers. Politicians get away with murder, advertisers lie, and medical researchers use pre-packaged statistical programmes to work out their results, without understanding or explaining what they are doing.

Straight Statistics has been set up to do something about it. I don’t pretend we can cleanse the Augean Stables singlehandedly – clearly we can’t – but we can certainly have some fun pointing out the most egregious errors and maybe, with the help of journalists, inching standards upwards.

The idea came from Lord Lipsey, a Labour peer and former political journalist. The campaign’s board consists of journalists and statisticians, including two well-known to many science writers, Professor Sheila Bird of the MRC Biostatistics Unit, and Professor David Spiegelhalter of Cambridge University. We are supported by a grant from the Nuffield Foundation.

Since I started digging for examples, I’ve been appalled by how easy they are to find. For example, the recent Research Assessment Exercise – supposedly a gold-standard in the measurement of how good the UK’s academics are – was published without a denominator. Universities were allowed to submit as few or as many of their academics as they wished, without disclosing the size of the pool from which they were drawn. That’s a statistical illiteracy that would shame a woman’s magazine reporting a readers’ survey, never mind the summits of intellectual excellence that the universities claim to be.

Crime statistics are a dog’s dinner. Take Penalty Notices for Disorder, on-the-spot fines for not-so-minor crimes (such as disorderly behaviour) and shoplifting. In order to meet its target of reducing the number of young people entering the criminal justice system, the Youth Justice Board doesn’t count these as crimes. But the Home Office, in order to meet its targets for crime cleared up, does. A carefully-collected statistic is thus either counted or discounted, depending entirely upon how well it suits a particular government department.

The outbreak of swine flu has shown how hospitals fail to gather information that matters, while accumulating tons that doesn’t. Claims that putting a ham sandwich in your children’s lunch box will condemn them to colon cancer are reported uncritically. A court rules that pollution from clearing up an old steelworks in Corby contributed to a cluster of birth defects, when the data do not clearly support any such conclusion.

And how about the claim that Britain wastes a third of the food it buys? Not really true, when you look at the figures, which include a stupendous 11,000 tons a year of pheasants allegedly binned every year. Or the claim recently made by the Home Office that violence against women and girls costs £40.1 billion a year. Abuse is a serious issue, but that’s more than all crime put together.

Straight Statistics does its best to expose such false reasoning, through its website. It is collaborating with Sense about Science in a new report, Making Sense of Statistics, due to appear in the autumn.

I welcome tips, ideas, and articles for the website. It’s impossible for me to spot everything but science writers have a good nose for unsubstantiated claims, so please pass on anything that smells whiffy to me. Anonymity is guaranteed if that’s what you want, but I also welcome signed pieces. Alas, any payment would be token, at best. But think of the good you’re doing!

Nigel Hawkes was Science Correspondent of The Observer 1972-80, and of The Times 1990-2000. He is a former Chairman of the ABSW.