LEADING ARTICLE : Prison is not a numbers game

The Home Office would like us to think that it is a sophisticated place, at the cutting edge of crime control. So it has called in academics to find a clever way to predict when convicts will re-offend.

As we report today, statisticians at Warwick University have worked out a magic equation that is supposed to sort the incorrigible from the redeemable. It's a fiendishly complicated sum. Those who try to use the equation will certainly find it more difficult than filling in a Cosmopolitan questionnaire on "Is your man a psychopath?" But the principle is simple. You feed in relevant numbers, including age and previous offences, do a bit of multiplication, a square root and some adding up. Out comes a total. If it's high, the individual is best locked up. A low figure probably indicates community service digging up old Mrs Brown's garden.

All very impressive. The "Offender Group Reconviction Scale" could eventually be turned into a Christmas board game, renamed Go to Jail! Yet close examination of the proposal reveals that the Home Office's new method is about as discriminating as a policeman's truncheon at a football riot. It turns out that the dominant number in the equation is - surprise, surprise - the frequency of previous convictions. So when you strip away the pseudo- scientific technique, the process produces what is virtually a truism: thugs are likely to continue being thugs.

This is not much of an insight. It tells us very little about what actually causes recidivism. The problem is that the academics failed to build into the equation factors that even a rookie probation officer knows are important. A criminal is less likely to break the law again if he or she has a home, a job and is off drugs. None of these variables has been included in the magic equation. Presumably the data was not available to the Warwick researchers.

The narrowness of their approach is symptomatic of wider ills. Criminal justice policy has proved reluctant to examine the deeper causes of crime. So, for example, Sir Paul Condon, the Metropolitan Police Commissioner, recently drew attention to evidence that most muggings in London are committed by young black males. This may be true. But Sir Paul's statement did not help us to understand the causes of such crime. People do not become muggers simply because they are black. There are deeper reasons, presumably to do with economic and social circumstances.

Statistical association can be amusing and harmless in its absurdity. There is, for example, a well-known correlation between the population of China and the divorce rate in Sweden. Likewise, according to a paper published in the journal of the British Astronomical Association, an increase in sun-spot activity coincides with Tory general election victories.

Statistics can also be dangerous, even when accurate. Sir Paul's comments have smeared a whole group of people to no good end. The crude research commissioned by the Home Office amounts to little more than a charter for locking up those who have been inside many times before.