If you haven’t heard of it, it’s a way for organisations working with offenders to get data on re-offending rates for their service users compared to a statistically matched control group. This gives them an indication of the impact of their work.
We are very proud to be involved with the Data Lab because it resolves an almost insurmountable problem: how do small charities with limited research budgets find out what happens to their service users in the long-run? For me, all the frustrations and debates around impact measurement can be boiled down to this question. The Justice Data Lab is the first genuine, workable solution; and in being so represents the most realistic way for impact measurement to become routine and for us to learn ‘what works’.
For this reason, NPC argues that, wherever there is administrative data on long-term impact, the Justice Data Lab should be the default model for helping organisations access this data. And this has begun to happen. It now has a cousin in the form of the UCAS Strobe service, and similar services are being piloted by DWP and DfE (via the Education Endowment Foundation). There’s also a lot of interest from abroad; particularly in the US, Canada and Australia.
Some opportunities, like a Data Lab for NHS data, still seem some way off. We have learned that getting Data Labs off the ground requires a lot of effort from analysts and commitment from politicians and senior officials—and these things rarely come together.
Which is why so much credit should go to the Ministry of Justice (MOJ). NPC was involved in the conception of the Justice Data Lab, but it is the team at MOJ that have done all the hard work. They should be lauded for the way the Justice Data Lab has matured into a service which offers increasingly clever analysis and links to more and more data sources. We should also applaud those holding the purse strings at MOJ who have continued to support it.
What about the results? So far 139 analyses having been published ranging from small projects like the West Yorkshire Community Chaplaincy Project (30 service users) through to major programmes like the HMPPS Co-Financing Organisation (CFO) Employability Programme (c.20,000).
Across all these projects—and including both statistically significant and insignificant results—a key finding is that the mean effect-size on the one-year reoffending rate is 7 percent or 2.2 percentage points. This means an averagely performing project should expect to reduce the proportion of people reoffending by this amount. Although not every project has achieved this, results have ranged from a reduction in reoffending of 20 points to an increase of 12 points.
Looking at individual organisations, there are three organisations I want to highlight where the data shows they worked with people who were more likely to reoffend than average but still showed a large and significant positive effect:
- Brighton & Hove City Council Preventing Offender Accommodation Loss Project (associated with a reduction in one-year reoffending from 53% to 33%)
- Warwickshire Youth Justice Service (51% to 38%)
- Two projects run by Safe Ground: Fathers Inside (40% to 24%) and Family Man (47% to 39%)
Also, two projects / organisations that recorded very large percentage reductions.
- Working Chance (associated with a 53% reduction in one-year reoffending from 13% to 6%)
- The resettlement and employment programme at HMP Kirklevington Grange (50% from 16% to 8%)
And the Prisoner Education Trust (PET), which worked with a large number of people (5,846) and was associated in a reduction in reoffending from 25% to 18%. Indeed, PET have gone a stage further and commissioned an economic analysis based on their results to estimate the value for money their services offer.
I am a bit reluctant to report on results like this because we don’t want the Data Lab to be a league table but, as it’s a birthday, you will hopefully forgive us.
The truth is that the real value of the Data Lab will come through meta-analysis that looks for wider conclusions and help us understand why these projects are so effective. As such, I’m really hoping that some keen researchers will continue the process of synthesising results we began with the University of Middlesex a couple of years ago.
Our priority now is to persuade organisations to launch more Data Labs in different sectors and countries, and to get more organisations using the ones that exist. It is a long process, but we are confident of success because the arguments are so compelling.