Do short prison sentences work?

Do short prison sentences work?


We need to lock more people up 

“Prison Works” said Michael Howard. He is a QC, so he must know what he’s talking about, right? No. Wrong. People who are sentenced to sentences in the community are less likely to offend than those getting a short prison sentence.


Says who, you lily-livered liberal?

Those notorious communists in the Ministry of Justice, that’s who.

They did a study looking at the data of re-offending rates between 2008 and 2011 and found that those who get a non-immediate custodial sentence commit fewer offences after sentencing that those who are locked up.


But surely locking up people cuts crime?

Well. The MoJ asked the following questions (with answers below):

Q : Are ‘court orders’ more / less effective at reducing re-offending than short term custody (prison sentences of less than 12 months)?

A : The one year re-offending rate was higher for those sentenced to short term custody than for those given ‘court orders’ overall (around 4 percentage points), community orders (around 3 percentage points) and suspended sentence orders (around 7 percentage points)

Also, the one year average frequency of re-offending per person was also higher for those sentenced to short term custody than those given court orders (by slightly under 1 re-offence on average)

This gap narrows, but is not extinguished and is still statistically significant, over time.

Q : Are there any specific requirements that are particularly effective, in terms of reducingre-offending? What is the impact of multiple requirements on re-offending?

In relation to Community Orders, adding requirements reduced recidivism rates. The data is not clear as to exactly which requirements are best. It does show that multiple requirements are more effective.

A similar, but less significant, effect is seen with suspended sentences. 

There is a nice table that shows all this on page 23.


Does this account for the fact that those who get a non-custodial sentence are less likely to have previous, so it’s more likely to be a one off?

Yes, it does. Have a look at Part 3.


Yeah, but I don’t like facts, I prefer my gut instinct – I know prison works – take that science!

Welcome to the House of Commons, and government policy making in general – you’ll fit in well.

Dan is a barrister at 2 Dr. Johnson’s Buildings practising in crime.


  1. Correlation does not imply causation.

    Causation can be in either direction. So, it could be the judiciary’s successful attempts to assess accurately the likelihood of particular offenders reoffending that cause short custodial sentences.

    My GP told me the other day that most men over seventy have prostate cancer, but that very few of them are prescribed any treatment at all for it. I would guess that the minority with diagnosed prostate cancer who are prescribed treatment for their particular cancers, are those most likely to die from their cancers. Would that mean that the treatments were *worsening* their prognoses? Of course not!

    • I agree with John Allman.

      It should be obvious that if the courts consider an offender will to go on to commit further crimes then they are more likely to send him or her to prison. On the other hand if they believe that the offender is unlikely to re-offend then a community punishment is rather more likely.

      The better the courts are at deciding which offenders pose the greatest risk of re-offending and should therefore be punished by prison, the greater the disparity between the recidivism rates of those sent to prison and those punished in the community.

      • Nice try, but not sure that’s right.

        Firstly, the MoJ have accounted for that. It will never be possible to have a ‘proper’ randomised trial which would be the gold standard I accept, but it seems that they have done their best to compare two similar groups.

        Secondly, it is not the case that the Courts pass a custodial sentence dependent on the perceived likelihood of reconviction. It may be that some judges subconsciously take this into account, but not enough to account for the difference.

        I don’t know either of your views, but stepping back the result is not a surprising one – someone who spends 6 months or less in prison won’t have anything by way of supervision/assistance/intervention etc, but have their life turned upside down. In those circumstances it would be very surprising if they didn’t have a higher rate of recidivism.

  2. My heart tells me that the “short sharp shock” doctrine is every bit as counter-productive as you think it is. I was simply pointing out that one cannot prove this, merely from correlation.

    • I know, and I’m normally the one to point that out. It seems that this was about as good as it would get for research on this topic!

  3. For once, a rather well-presented and statistically sound piece of work from the MoJ. This means of course that it is something which Failing Grayling hasn’t managed to get his mitts on.

    Short custodial sentences are essentially an admission of failure by courts when everything else has been tried, but no-one is under any illusion whatsoever that they work, even if they become inevitable after one breach too many, or the umpteenth conviction for theft, etc.

    Custodial sentences for certain offences are more justified than for others. The courts are required to begin at custody level for knife crime for very good reason. Large scale benefit fraud usually leaves the courts with no realistic alternatives.

    What is perhaps dispiriting in this as in other analyses is just how far courts will go to avoid sending someone into immediate custody, because they know that they won’t get any drug rehab, detox or even any meaningful intervention whatsoever. Judges and magistrates do believe, for example, that purposeful restorative justice, even in cusp of custody cases, can pay enormous societal dividends all round, but the only area in the country to still have the possibility of making an adult RJ disposal is the Thames Valley.

    The “rehabilitation revolution” has seen Provation starved of resources and the
    courts denuded of effective requirements to impose.

    “Payment by results” will simply mean that more money will be paid to the same old consortia that have fiddled the books on tagging and so many other fronts. What we really need is to find a way of allowing Probation to engage proactively with residential drug rehab centres, small local charities that have the expertise in and understanding of housing issues, money management, and all the basic tools for living, and can work with other voluntary groups, ex-offenders and the like, and to bring them all together with local authorities, the healthcare system (including community mental health teams) and offender managers in a coordinated manner. In fact, something rather like that now forgotten notion of the “Big Society” that was floated by a certain Dave Cameron only a few years ago…

  4. You have to define “works” and that means knowing what the purpose of the sentence is: what about deterrence, protect the public or punishment ? These are just as valid purposes as rehabilitation. Most people sent to prison for the first time have had several previous attempts at rehabilition that didn’t work. Eventually courts give up trying to stop reoffending and revert to punishment.

    • True. Although all the evidence is that deterrence doesn’t work. If it’s a sentence under 12 months, in theory the public get a break for a few months but that’s it. And the public may prefer to have people out in the community not committing crimes rather than away for a bit and then back committing crimes.

      So we’re left with punishment.

      • It is true that who get court orders are less likely to re-offend but if you think from the victim’s point of view it is logical to say that there should be a balance between crime and punishment.

        • Of course. But nobody is saying a choice between punishment or no punishment. It is punishment in prison versus punishment in the community.

    • J2biscuit says that most people sent to prison have previously had several attempts at rehabilitation that did not work. Is that true? If so then does anyone know how big the percentage is? Is it 51% or is it 90%?

      If we take into account cautions then I suspect that the figure will be rather nearer to 90%; and that only a small proportion of those who are sent to prison are first offenders who have never been given a chance by the system to sort themselves out after being found committing a crime.

      There should be little surprise if, in the case of people who have already been given cautions and community punishments, in order words recidivists, prison does not stop all re-offending.