# Statistics and the law

Updated Tuesday, 12th April 2011
With clean-up rates and reoffending numbers, discussions of crime often circle back to statistics. But are these 'facts' really such reliable witnesses?

This page was published 10 years and 9 months ago. Please be aware that due to the passage of time, the information provided on this page may be out of date or otherwise inaccurate, and any views or opinions expressed may no longer be relevant. Some technical elements such as audio-visual and interactive media may no longer work. For more detail, see our Archive and Deletion Policy.

Professor Gary Slapper, The Open University’s Professor of Law, has written about statistical aspects of the law several times. Here is one of his contributions, originally published in 2002 in the New Law Journal.

### Figures of fun

According to Benjamin Disraeli there were three kinds of lies: "lies, damned lies and statistics". This week, a report of the inspectorate of constabulary has gone some way to corroborate that view. It suggests that the annual crime figures under record crime by about 1.4 million offences a year.

This year's figures were recently published to the customary cacophony of calls for resignations, due credit, more police officers, more initiatives, more public patience and fewer statisticians.

Beyond the initial rhetoric, surprisingly little in the way of policy changes or legislation is actually made as a result of statistics, which is ironic as statistics were originated precisely to permit governments to make sound social policy in accordance with hard evidence.

The first formal use of statistics was in the seventeenth century. Statistics (etymologically, a science dealing with facts of the State) were pioneered by the writer William Petty who really began the process of gathering and using numerical data which he thought might help in the making of policy decisions and in judging the "moral health" of the nation.

In his work Political Arithmetic, in 1699, he became the first person to advocate collecting crime figures as a better way of guiding policy than the futility of "using only comparative and superlative words". He wanted to make intellectual arguments about policy superfluous by "finding the facts".

Agreeing upon "the facts", however, is not as simple as someone like Dickens’ Mr Gradgrind ("Facts alone are wanted in life") would have liked.

As the historian E.H. Carr once observed: "It used to be said that the facts speak for themselves. This is… untrue. The facts speak only when the historian calls on them. It is he who decides to which facts to give the floor, and in what order or context".

Much crime is unreported to the police and so never ends up in the statistics.

The Home Office’s periodic British Crime Survey estimates that the true level of crime (the sorts, anyway, which inform the official figures) is about four times that which is registered in the annual statistics.

Quite often, especially in the financial services sector, businesses do not report crimes against themselves for fear of lowering their public image. Many citizens today are not insured against car theft or property loss (because they cannot afford the premiums) so they have no incentive to tell the police if they become victims.

A steep statistical rise in crime can sometimes arise not from a real growth in a particular type of conduct but from a new policing policy - offences of "lewd dancing" rose by about 300 per cent during 12 months in the 1980s in Manchester, but only because the zealous Chief Constable James Anderton had deployed a great many officers in gay night clubs

Old crime, new temptation: Stealing a mobile

Sometimes the enactment of a new range of offences or the possibility of committing old offences in a new way (like computer offences involving fraud and deception) can cause an upward jolt in crime levels. The figures just released show a startling jump in street robbery but much of this seems to be a very particular crime: the theft of the now ubiquitous mobile phones.

Conversely, if crimes like joyriding and some assaults are kept out of the categories measured in the annual statistics, as is the case, the official figures do not reflect even what is reported to the police as criminal.

The way that criminal statistics are compiled by the Home Office is also relevant. From April 1998, police forces started to count crime in a way which, according to the government, will give "a more robust statistical measure".

Under the new rules, crime is recorded as one crime per victim. Some crimes, like assaults, have always been recorded in this way, so the main impact of the change will be in the area of property offences. Shop thefts, for example, were the old rules counted offenders, will now count victims. Multiple thefts from cars in a car park with a barrier were previously counted as one offence but are now counted as separate offences.

Yet, despite all the uncertainties with some aspects of the figures, many statistics present such a clear, strong steer that we ignore them at our peril - if tests show that, in the event of a car accident, you are 80 per cent more likely to survive if you wear a seat belt, the lesson is obvious. The political will of some governments, however, has seemed perversely at variance with "the facts".

Consider the use of prison.

Eight out of ten people sent to prison have been convicted of a non-violent crime, and rates of recidivism are high: five out of ten people discharged from custody in 1992 were reconvicted after two years.

It costs about £36,000 to keep a prisoner in prison for a year.

Yet, prison building continues apace, and one Home Office figure estimated that by 2005 we shall have over 90,000 people in jail - a doubling of the prison population in twenty years. That, surely, would not be something about which a society should be proud?

At time of writing, the latest official criminal figures show a rise since last year of 192,000 offences, bring the total of notifiable offences to 5.3 million. The real figure would be closer to seven million if you accept the conclusions of the inspectorate of constabulary.

Let us hope that when we make social policy in future there is a good chance (evens?) that it will be based upon the available statistics.

### A perspective from 2011

Although this piece is several years old now, the general points made in it have stood the test of time. The detailed numbers in the crime statistics have changed – you can find the latest numbers on the Home Office website. Crime recorded by the police (in England and Wales) has fallen compared to 2002; the latest annual figure for England and Wales (for the year to September 2010) is 4.2m offences. This is still considerably different from the British Crime Survey figures (9.4m crimes in the same year) – different questions, different answers.

Figures on prisoners and reoffending now come from the Ministry of Justice (for England and Wales) and the devolved administrations in Scotland and Northern Ireland. The prison population in England and Wales stood at 85,600 in August 2010; including prisoners in Scotland and Northern Ireland puts the total up to about 95,000. The position on reoffending is better than it was but hardly encouraging; of those released in 2002, about four in ten had reoffended within two years.

(There’s even been more research on who actually made the remark about "lies, damned lies and statistics". It probably wasn’t Disraeli; it might have been Sir Charles Dilke, according to research by the University Of York.)

Probably not the man who said 'lies...': Disraeli

### Equals before the law

As an afterword, another piece by Professor Gary Slapper on another aspect of statistics and the law, first published in The Times in 2005.

Recalling the comical line that 80 per cent of statistics are misleading, a recent report announced an official finding that "60 per cent of Britons do not trust government statistics".

The relationship between law and statistics is close and important. The word "probability", prolifically used in both fields, derives from the Latin probare to prove, or test. Many pioneers of statistics in the 18th century, including Condorcet and Laplace, were concerned with problems arising from legal evidence. The study of forensic statistics intensified after an American case in 1968 in which the prosecution presented a fallacious statistical argument regarding eyewitness identification evidence, and again after 1985 with the advent of DNA profiling.

Today, statisticians often give expert testimony about theories every bit as complex as the law which the judge must master.

In one case, Peter Donnelly, Professor of Statistical Science at Oxford, was taking a court and jury through some points on calculators, which all had been given. He went through a worked example with them and then said: "Your calculator should now show the value 31.6." The jurors all nodded, then the judge said plaintively: "But mine shows zero."

### Find out more

Discover how to study statistics or law with The Open University.