Some things in life are complicated. Take, for example, deciding the causes of the August riots. The government, Metropolitan Police and the Guardian/LSE are just three bodies who have published their analysis recently.
Depending on who you listen to, the root cause of the disturbances was:
- Broken Britain, a lack of family values and feckless parenting OR
- Social media – particularly the Blackberry messaging service OR
- The antipathy between young people and the police OR
- The slow response by the police/government OR
- The recession and cuts in public services.
The key, of course, is that there is some truth in all these explanations although we will never know exactly how they interacted to cause the events of the 6 – 10 August 2011; it’s just too complicated.
Some things in life need to be complicated. Measuring reducing reoffending outcomes in payment by results schemes is one of those things. However, not everybody agrees.
It is clear from his recent speeches and, in particular, the publication of the roundtable discussion organised by @reformthinktank that Crispin Blunt and the government favour a simple binary approach to measuring reoffending for the range of PbR criminal justice initiatives (full details in my free-to-download resource pack).
“Binary” is a term of jargon that I don’t find particularly helpful. In this case, it simply means that a project’s success would be measured on how many offenders committed further offences – typically in a one year period following their release from prison.
A binary approach is very attractive, predominantly because of its simplicity – easy to measure, and easy to understand. If a released prisoner does not commit a crime, (or, rather, does not get caught and convicted), the project gets paid. If s/he is convicted, the project doesn’t get paid. A binary measurement is also particularly attractive to government because anyone who does not reoffend incurs no costs whatsoever to the criminal justice system.
The problem with a binary approach is that it is unlikely to reward payment by results projects which truly focus on those offenders who we all want most to stop offending – those with entrenched criminal histories who commit large numbers of crimes. We have an increasing evidence base on what we now term “desistance” but which, in normal English, means the process by which people give up committing crime. All the research (which mirrors what we know about the processes of recovering from addiction) clearly shows that “desistance” is an uneven process typically marked by a reduced frequency of offending and/or a reduction in offence seriousness before someone gives up crime for good. You can find good explanations of desistance via @ben_jarman or by reading: Changing Lives: Desistance and Offender Management by @fergus_mcneill and Beth Weaver.
There is an alternative to the binary approach, known by the shorthand of “frequency”, this measures the reduction in the number of “reconviction events” (sometimes several offences are dealt with in one court appearance) compared to either what is expected from a profile of the client group that the project is working with or compared to a similar cohort in another area. Although the “frequency” approach is a more reliable indicator since it accurately measures a reduction in crime, it is more complicated and, therefore, more expensive to measure and not so straightforward to communicate to the general public.
Interestingly, the best known payment by results project, the ONE Project which provides a resettlement service to short-term prisoners at HMP Peterborough (and whose one-year report was published last week and is reviewed here) is being measured by the frequency approach.
However, the payment by results contract operated by Serco and its partners, Turning Point and Catch-22, at HMP Doncaster will be measured using the binary method. Helpfully, Kate Steadman, from @Sodexo_UK characterised the Doncaster model as a “penalty by results” rather than payment by results system since Serco needs to achieve a 5% reduction in reoffending rates in order to keep its contract.
I am aware of a number of projects which use both measures and the most effective ones tend to succeed on both binary and frequency measures. There is a strong argument for applying the binary measure to projects who work with high risk offenders where the priority must be to avoid further victims at all costs. But if we apply that approach to schemes aimed at short term prisoners or drug using offenders who characteristically commit large numbers of minor offences, we will be building in a disincentive to work with large numbers of this group the moment that they commit one further offence.
Rather than binary of frequency, perhaps we should be adopting a “horses for courses” approach.
Subsequently, I wrote another post on this issue, looking arguing the particular inappropriateness of binary measures when Prolific and Priority Offenders (PPOs) are included in PbR cohorts.