Nigel Hawkes: And our survey says...nothing you can rely on

Behind the numbers

Aberdeen is the most sleep-deprived city in Britain, Bristol the unhealthiest, London the smelliest. The average person spends £2,776.75 a year on keeping healthy, and married couples achieve their peak happiness two years, 11 months and eight days after tying the knot.

These "facts" – and let us not forget the disclosure that 48 per cent of women admit faking an orgasm, and 9 per cent do it all the time – were all published by British newspapers last week. They were all the result of surveys, the quickest way to a news editor's heart.

Surveys are proliferating like Japanese knotweed. They borrow the authority of statistics to lend an air of truth to otherwise implausible claims. And, remarkably, all the surveys listed above come from the same survey factory, based in Bristol (Britain's unhealthiest place, remember?).

A regional news agency, South West News Service, offers a complete survey service. For £2,500 they will originate the questions, perform the survey, compile the results, write the press release, then report it as if it were news. That is a bargain if the company commissioning the survey gets a mention in Metro, the Daily Mail, the Daily Mirror, the Daily Express, the Evening Standard, the Press Association, The Sun, the Daily Record, or the Daily Star, all of which ran one or more of these surveys.

The surveys are organised by a subsidiary of SWNS, the PR agency 72 Point, and done by OnePoll, using a panel of volunteers who are paid (modestly) for their participation. They can turn a survey round in a couple of days. For an extra £300 you can motivate the panel to respond even more enthusiastically.

So what's wrong? In statistical terms, plenty. The samples for these online surveys look large – typically, 3,000 or so – but they are neither random nor necessarily representative of the public at large. The results are often broken down city by city, to attract interest from local papers or to compile league tables, when the samples for each city are far too small to make such comparisons valid. No confidence intervals are quoted.

And the results are often given an absurd precision, in the knowledge that journalists are easily seduced by an exact number. Claiming that the average person spends £2,776.75 a year keeping healthy sounds as if somebody counted every penny. In truth, nobody really counted anything at all, or not to an accuracy of better than the nearest £100 or so.

So these surveys are largely meaningless. But SWNS's website has testimonials from Tony Gallagher, Editor of The Daily Telegraph ("Its ability to deliver high-quality stories day after day is simply remarkable"), from Chris Pharo, head of news at The Sun, ("excellent survey-led news stories and stats") and, more worryingly, the Central Office of Information.

You might also worry, if you care for press independence, that quite such a cosy relationship between PR and news-gathering isn't healthy. But it's only a bit of fun, surely? "Poor show, chaps: survey shows that nearly one in 10 women fake it between the sheets", as the Daily Mail headlined one of the surveys. It's not meant to be taken seriously, so aren't my criticisms a bit priggish?

Not really. Almost all national statistics, from crime to the GDP, come from surveys designed with huge care to get as near as possible to the right answer. Newspaper surveys bear about as much relation to these as a chicken nugget does to a three-star Michelin lunch. Newspapers who give them space are selling themselves cheap.

Nigel Hawkes is Director of Straight Statistics