FEATURE9 November 2011

Statistics – lost in translation?



Figures show that crime has been falling in the UK for 15 years. So why do most people still think it’s rising?

According to the British Crime Survey (conducted for the Home Office by TNS-BMRB and bearing the independent UK Statistics Authority’s kitemark), levels peaked in 1995 and fell steadily in the decade after that. In the last few years they’ve fallen further, although more slowly. The number of incidents has halved, as has the amount of violent crime. Recorded crime has fallen, too, since the current counting rules were introduced nearly a decade ago.

But that’s not the impression you’d get from coverage of crime in the mainstream media – which may be why, in the same survey, the percentage of people who believe that crime across the country is rising has stayed stubbornly between 60 and 80 per cent.

The so-called crime perception gap is one example of how statistics can be misunderstood, mangled or just plain ignored by the public, politicians and the press – with far-reaching consequences. At a Royal Statistical Society event yesterday, social scientists discussed why those who produce stats have such a hard time getting their message across.

“You have to believe in the good that statistics can do, and you have to care about the user”

Richard Alldritt, UK Statistics Authority

Mike Hough of the University of London’s Institute for Criminal Policy Research blames the problem with perceptions of crime partly on the Labour government’s tinkering with the way crimes were recorded. By trying to improve counting, they ended up opening the data to question and confusing the public, Hough said. Their opponents in parliament and the media had a field day, and in the end “nobody knew what the hell was going on”.

But the real root of all this, Hough believes, is the toxic relationship between politicians and the media, who are trapped in an “unpleasant dance… and it’s quite often difficult to see who’s leading the dance”.

Simon Briscoe, the former Financial Times columnist, pointed out that the misuse of figures is not a problem confined to politicians – all sorts of other interest groups are at it too. Statisticians, he said, need to “get real” about their role in promoting public understanding of the data they produce.

For Richard Alldritt of the UK Statistics Authority, the communication problem is fundamentally one of language. Although official statistics influence a vast range of decisions, the language of statistics is foreign to most people, he said, and “very few decisions are directly determined by numbers or graphs”. They need to be translated into a language people can use.

“Good translation requires both knowledge of the subject and a real empathy with the user,” Alldritt said. “You have to believe in the good that statistics can do, and [you have to] care about the user.”

But statisticians can sometimes be reluctant to threaten the purity or impartiality of their work by trying to translate stats into plain English. Alldritt says they need to try harder. “Sometimes a general steer in simple English is all that’s required to bring some sense to a big issue,” he said. “The plain English [we use] must take account of the way organisations and individuals are going to use statistics. A little more empathy with the user would go a very long way.”

In defence of politicians, Alldritt pointed out that it was they who decided to set up the UK Statistics Authority in 2007, and that parliament’s select committees have been supportive of the authority in scrutinising how official stats are used.

But he warned that there is no room for “false naivety” on the part of statisticians about the motives and methods of journalists, politicians and others who want to make use of official data. Publishers of statistics have to accept the nature of public debate, and the role of data in it. “There is no option but to tread the narrow, poorly lit, hazardous path” between statistical safety and subjective interpretation.

Whether such improvements to the communication of official statistics will be enough to rebuild public trust, is another question. The University of London’s Mike Hough sees hope in the aftermath of the phone-hacking scandal, in which politicians of all sides were eager to distance themselves from Rupert Murdoch’s News Corp. Might they now be brave enough to bring their unhealthy relationship with the media to an end?

As for the crime perception gap, it surely can’t be impossible to reach a situation where the majority of people believe the truth about crime trends, Hough said. But by then, “crime might be going up anyway”.


13 years ago

Nil desperandum. The Levenson inquiry into press and media standards does offer a window to push for more journalistic probity - and capacity. Getstats, the Royal Stats Soc campaign, is working with journalism educators and media leaders to establish a sort of minimum statistical competence for would-be media folk. We're also drafting a 'crime stats competence' framework for the election of police and crime commissioners in a year's time.

Like Report

13 years ago

Interesting article. I wonder if it would be possible to use statistics to indicate what percent of the problem is due to wilful politicians, what percent is down to wilful journalists and what percent is down to general ignorance of statistics. Otherwise we just keep on going on the same loop.

Like Report