Last week, Priti Patel, the new British Home Secretary, provoked a media stir when she announced that she thought the criminal justice system should aim to strike fear into the heart of criminals. Critics combined her new interview with her previous support for the death penalty, banned in the mainline UK since 1965, to suggest that Patel represents a draconian and reactionary turn in British law enforcement.

Then a couple of days ago, a YouGov survey showed, that 72 per cent of the British public agreed with her. Media commentators can forget quite how high support is for law and order among ordinary citizens. Support for the death penalty itself still attracts almost half of the population.

Are the public right? The meat of the Government’s new policy is an increase in the number of police officers; this at a time of increasing violent crime and concerns about rising knife crime in London. On that front, the evidence points in Patel’s favour. More police often reduce crime and do so through a variety of mechanisms, including situational deterrence (for example, patrolling in high-crime areas) as well as increasing detection rates. There is general agreement that increasing the certainty of apprehension contributes to deterrence.

What about punishment severity? There the evidence is decidedly more mixed. There is remarkably little evidence, for example, that the death penalty deters crimes like murder more than an appropriate prison sentence.  Using a new data set of sentencing practice in all police force areas in England and Wales, myself and some great colleagues at the Centre for Crime, Justice and Policing at the University of Birmingham produced a study just printed last month: ‘Alternatives to Custody’. We compared the way a previous year’s sentencing influenced the subsequent year’s recorded crime.

What we found was that for property crime, our largest category, and robbery, community sentences generally reduced crime more than prison. In fact, one of our models suggested increased use of prison caused subsequent crime to go up. On the other hand, prison seemed to work (and was the only thing that worked) to reduce violent crime and sexual offences. (We summarised our results for the LSE British Politics and Policy blog.)

The lesson that we draw is that deterrence isn’t an overwhelming explanation of the impact of sentencing. Harsher sentencing probably works to deter some offenders. But at the same time carrying out punishments can have criminogenic effects. Experience of prison often makes convicts less employable and can effectively socialise them into having an enduring criminal identity. Of course, many offenders in the real-world are not particularly well informed about the criminal justice system. They may also have less self-control than a typical member of the public. So information about an increased penalty for a crime may never effectively filter into the deliberation and reflection of some offenders until they are sentenced, at which point you get the high financial and social costs of prison kicking in.

Getting caught by the police, perhaps on a few occasions,  is a more immediate sign to an offender that their behaviour is unlikely to pay off in the long-term. What does this mean for Patel? It suggests that fear of the consequences can play a role, but what we really need is graduated sanctions, avoiding prison when possible. This gives offenders plenty of options to exit a criminal career path. Relying on terror, by contrast, can lead to a large prison population producing a lot of stigmatized and harmed individuals who quite possibly will re-offend when they are released.