Classifying offenders could lead to criminal justice revolution

By Juliet Rix, The Guardian, 2 March 2011:

Imagine being able to predict which offenders will go on to commit serious crimes. Juliet Rix on a new system that could have a drastic impact on the justice system

A new approach to classifying crime could revolutionise the way the criminal justice system deals with offenders. According to Peter Neyroud, the recently retired chief of the National Policing Improvement Agency and a former chief constable, such a revolution is urgently needed. Unless the UK dramatically changes the way it does justice, he says, the forthcoming 20% cuts in the criminal justice system budget will be "disastrous". So the requirement to save money is an opportunity to persuade politicians of all parties to embrace a new and more effective way of dealing with offenders, Neyroud believes. He and Lawrence Sherman, Wolfson professor of criminology at Cambridge University, are together working on a scheme they believe could achieve an irresistible trinity: cutting prison numbers by 50%-70%, saving money and keeping the public safer.

The prison population has soared in recent years – almost doubling since justice secretary Kenneth Clarke was last home secretary in 1992-93 – with a commensurate rise in the budget. Some 85,000 people are now locked up (the highest rate in western Europe) at a cost of some £45,000 a year each. The UK's criminal justice system is the most expensive per capita in Europe, costing more even than in the US, yet half of released prisoners are reconvicted within a year, three-quarters in the case of juveniles. "The best study we have on the effect of prison," says Sherman, "suggests that the first time an offender goes to prison it triples his offending rate, so the advantage of [temporary] incapacitation is quickly lost." As a tool for the overall reduction of crime, he says, "prison doesn't work".

Sherman and Neyroud hope all this will change if their new approach to classifying and dealing with crime takes off. Sherman has developed a Crime Harm Index (CHI) – a way of consistently measuring the harm done by crime rather than simply counting offences. Clearly, a murder does a lot more harm than shoplifting and so should carry more weight. Most crime measures – as well as government targets – focus purely on numbers of crimes, so a conviction for smoking a single joint of cannabis counts the same as one for rape.

The index would completely turn on its head the way crime is measured, by predicting the harm likely to be caused by individual offenders. This is crucial information for those deciding how to deal with them and a key tool, Sherman and Neyroud believe, in reducing the prison population – and budget – without compromising public safety.

Cambridge University's statistical laboratory is designing an automated system for offender forecasting based on the sophisticated analysis of 100,000 criminal histories from the Police National Computer. Using all these past patterns of offending, the system will be able to forecast how much harm each arrested person is likely to cause over the next two years.

The system will soon be tested in a pilot scheme around the country. Fourteen police forces have expressed an interest in trying it out and, with a grant from the Monument Trust, it is hoped that the system will be installed in at least some police stations by the end of this year for a "clinical trial". When someone is arrested, they will be randomly allocated either to the old procedure – which has usually meant prosecution – or to the new CHI approach.

In the latter case, the custody officer will put the individual's name into a computer (which will hold current criminal records as well as the forecasting software) and receive a response: red, yellow or green.

Red would mean that here was a potentially dangerous offender likely to commit seriously harmful crime in the next two years. This would signal to the police to go ahead with investigation and prosecution to achieve the maximum possible prison sentence to protect the public – even if the offence committed was non-violent (Al Capone, Sherman points out, was jailed for tax fraud). "This is what prison is good for," says Sherman, "keeping really dangerous people off the streets."

At the other end of the scale, green would indicate someone who is a minimal risk to others. Crimes such as shoplifting, being drunk and disorderly and minor assault (verbal or pushing and shoving, not violence) would generally merit a green. If offenders had previous convictions for more harmful crimes, they would come up yellow or red. Greens, Neyroud says, are, "people who have blundered into crime or made one big mistake they are unlikely to repeat." They are "not bad enough to waste money prosecuting", says Sherman and, crucially, evidence suggests that putting them into the criminal justice system does more harm than good – not just to the individual but to society as a whole. So the green light recommends diversion away from prosecution.

The yellows are the tricky cases. Neyroud, a former custody officer, describes them as "real offenders with the potential to commit further crimes but probably not of the worst kind". They might have been responsible for low-level mugging (without violence or a weapon) or burglary. Many will have problems with substance abuse or mental ill-health. All yellows will require careful intervention but not necessarily court or custody.

Out of the roughly 750,000 offenders dealt with each year, Neyroud estimates that perhaps 5% of offenders will fall into the red category, around 20% into the yellow and up to three-quarters into the green. If this is the case, it is not hard to see how the scheme could cut prison numbers by the 50-70% the pair claim.

Those not prosecuted (green and yellow) would be referred to an offender management team of trained police and partner agencies with a range of interventions at their disposal including treatment for drugs or mental illness, curfews, restorative justice, and close monitoring by the police with the omnipresent threat of prosecution if necessary.

The index would also allow payment-by-results to be introduced in offender management, a key plank of Clarke's proposed reforms to prison and probation. But Sherman would like to see this applied to the police as well – and the CHI used to measure the results in question. He points out that the Ministry of Justice's proposed payments-by-results system may merely count how many ex-prisoners reoffend – encouraging programmes to concentrate on the easy offenders rather than the serious or persistent ones. A CHI-based system would, in contrast, be able to identify the considerable achievement of turning a murderer into a mere shoplifter.

In the case of the police, Sherman says, total "harm" committed in their area would become the key measure of results. Areas with more harm would need greater starting budgets, with forces that reduce harm rewarded with extra money to continue and expand the good work – calculated as a proportion of that saved by reducing prosecution and prison costs. While Sherman applauds Clarke's stated desire to increase rehabilitation of offenders, he points out that there is little good evidence for what works in rehabilitation after prison, and none for what could work while cutting costs. So why not start, he argues, by introducing change at the outset, keeping substantial numbers of people out of the judicial system altogether? There is good evidence, Neyroud points out, "that quick and certain justice – starting in the custody suite within hours of arrest rather than waiting weeks or months for a court appearance – reduces reoffending".

In America, the political left and right are already coming together over similar problems of rocketing prison numbers, spiralling costs and high rates of recidivism. Several states are already successfully transferring resources from prisons to community programmes.

Texas has led the way. Starting in 2005, the state strengthened its probation service, introduced incentivised funding and shifted money from prison building to community programmes, addressing such issues as drug abuse and mental health. The reforms are expected to save $2bn (£1.2bn) in prison costs over five years, prison numbers are down, for the first time there is no waiting list for drug treatment in the state, and crime has already dropped by 10%.

Sherman and Neyroud hope that Clarke's rightwing backbenchers are taking note, but they are not expecting anyone to take America's or anyone else's word for it. Their pilot scheme of a harm-reduction approach to criminal justice is being set up as a scientific experiment. There will be control groups, careful data collection on actions and outcomes and a three-year follow-up, and they expect to have significant results before the end of this parliament.