Measuring Reoffending & Payment By Results

Key sources

1)    Ministry of Justice Statistics

  • Response to consultation on reoffending statistics[1] (MOJ, 2010b)
  • Proven Re-offending Statistics Quarterly Bulletin (PRSQ) – January to December 2009, England and Wales[2] (MOJ, 2011g)
  • Early estimates of proven reoffending – January to December 2010, England and Wales[3] (MOJ, 2011h)
  • 2011 Compendium of Reoffending Statistics and Analysis[4] (MOJ, 2011c)
  • Assessment of compliance with Code of Practice for Official Statistics – Statistics on Re-offending in England and Wales[5] (UK Statistics Authority, 2011)

2)    Payment by Results

  • See PRSQ (MOJ, 2011g) and 2011 Compendium (MOJ, 2011c) for caveats regarding re-offending effectiveness comparisons.
  • Disley, et al. (2011)[6] describes implementation issues for the Peterborough Social Impact Bond
  • Breaking the Cycle[7] provides ‘rehabilitation revolution’ policy context background (MOJ, 2011i)
  • The Green Paper Evidence Report for Breaking the Cycle (MOJ, 2010c) provides a summary of experience of and issues for PbR
  • Government Response to the Justice Committee Report: The role of the Probation Service (MOJ, 2011j)[8]
  • Fox (2011)[9] discussion on PbR and social impact bonds

Key Points

1)    Changes to the way re-offending statistics are calculated and presented

Following consultation in 2010 there have been substantial changes to the basis of re-offending statistics. These changes have been found to be compliant with regulations for National Statistics. Changes include:

  • Combination of 6 previous measures to a single measure
  • Including adults and juveniles from age 10 in the same statistic
  • Using a cohort follow-up methodology for all measures, and so making national and local measures consistent[10]
  • Following up for 1 year, plus 6 months for court processing
  • Presentation of an adjusted proven re-offending rate to control for changes in the composition of the offender group.
  • Enabling of geographical area and offender type comparisons
  • Enabling measurement of prison and probation re-offending

Generally re-offending rates appear lower than with the previous measures because of changes to the range of offenders and disposals included in the single measure, and changes to the follow up period. Trends under the old and new measures are broadly similar.

The percentage of offenders re-offending, and the number of offences they commit have now been made accessible on the Ministry of Justice Website Making Sense of Criminal Justice (MOJ, 2011k)[11]. Data can be viewed at local authority level.

2)    Overall trends in re-offending

  • Recently (2008 to 2009) the re-offending rate fell from 26.9% to 26.3%.
  • This continued the longer term trend (2000 to 2009) which has fallen from 27.9%.
  • There has also been a recent fall (2008 to 2009) in the average number of proven re-offences from 2.89 to 2.79 per offender.
  • This continued the longer term trend (2000 to 2009), which has fallen from 3.37 offences per offender.

3)    Age and re-offending

  • Proven re-offending was highest for 15-17 year olds, falling with increasing age.
  • From 2008 to 2009 proven re-offending fell for 15 to 29 year olds and rose for other age groups.
  • The largest falls were for those aged 21 to 24.

4)    Previous offences

  • Although proven re-offending was highest for both adults and juveniles, the largest fall since 2000 was for offenders with more than 25 previous offences (a fall of 3.8 percentage points).
  • Adults with more than 25 previous offences represented just over a quarter (27.8%) of adult offenders in 2009, but committed almost two thirds of proven re-offences.
  • Juveniles with more than 25 previous offences represented only 4.8% of juvenile offenders in 2009 but committed almost one fifth proven re-offences by juveniles.

5)    Index offence and re-offending

  • Burglary has continued since 2000 to have the highest re-offending rate (48.1%).
  • Sexual offences against children continued to have the lowest (9.8%).

6)    Comparing prison and court orders

  • PRSQ Bulletin January to December 2009 gives re-offending for prisoners and adults starting community orders, but these data cannot be used to evidence differential effectiveness as the groups are not matched[12].
    • All prisoners: 46.8% (average 4.03 offences)
    • Less than 12 month sentences: 56.8%
    • Community Orders: 35.6% (average 3.15 offences)
    • Suspended Sentence Orders: 31% (average 2.86 offences)
  • Compendiums produced in 2010 and 2011 compared ‘relative effectiveness of different disposals’.
  • The 2011 Compendium broadened the range of comparisons and improved some of the matching procedures.
    • Community Orders and Suspended Sentence Orders were more effective than custodial sentences of less than 12 months by between 5 and 9 percentage points, for data up to 2008
    • This reinforced the 2007 finding.
    • Generally, sentences up to 4 years were found to have lower re-offending than shorter sentences.
    • Compared to Community Orders, SSOs and conditional discharges had lower re-offending (1.4 and 5.6 percentage points respectively).
    • There was no evidence of a difference in re-offending between community orders and fines.
    • Conditional discharges showed lower re-offending than fines.
  • 2011 Compendium warned of the limitations of current Ministry of Justice administrative data to determine relative effectiveness  (MOJ, 2011c, p.4)

7)    Offender needs and reoffending

  • Compendium (2010) provided the analysis from the research study: Surveying Prisoner Crime Reduction.
  • Higher reconviction was associated with: problems in early life, family and schooling (violence in the home; abuse as a child; school expulsion or exclusion and lack of qualifications.)
  • Homelessness and joblessness were more prevalent than in the general population.
  • Drugs and alcohol use were more prevalent in than in the general population and were associated with higher reoffending.
  • Reconviction rates were similar for offenders reporting treatment for mental health problems in the year before custody compared to those who did not report treatment.

8)    Performance comparisons[13]

  • Early Estimates of Proven Reoffending – January to December 2010, using 3 month follow up and 3 month waiting period, showed more increases in reoffending than decreases compared to the previous year:
    • Probation Trusts: 15 had lower rates; 20 had higher; none remained the same. (Trusts were also compared to expected rates, showing 2 lower and 3 significantly higher.)
    • PPOs: 37 local authorities showed lower estimates; 42 showed higher.
    • Drug Action Teams: 41 had lower rates; 123 had higher.
    • Young Offender Teams: 48 showed lower rates; 109 higher.

9)    Payment by results (PbR)

  • PbR is part of the government’s focus on reoffending, cost saving and decentralisation. PbR aims to shift the risk for innovation and effectiveness away from the taxpayer and towards private and voluntary sector organisations.
  • PRSQ cannot be used to examine the effectiveness of sentence types[14].
  • 2011 Compendium presented more detailed sentence comparisons, but how the new reoffending statistic will be used to evidence PbR is not yet clear.
  • Pilot PbR schemes are underway at Peterborough and Doncaster prisons. There are also local ‘justice reinvestment’ projects in London and Manchester.
  • Early implementation evaluation of the social impact bond pilot at Peterborough (Disley, 2011) has indicated the following issues for outcome measurement:
    • Time consuming and complex analytical process
    • Difficulty of attributing change and sharing outcome payment where several agencies involved
    • Need for ‘intention to treat’ analytical model to avoid ‘cherry picking’
    • The Peterborough control group comparison model for evaluation cannot be rolled out nationally for outcome measurement as there would be no comparison group to provide a counterfactual.
    • Availability of robust cost data
    • Importance of size of project in delivering ‘cashable’ benefits.
  • Government response to the Justice Committee report on the Role of the Probation Service located responsibility for developing appropriate measures for PbR with the wider research community, rather than solely MOJ.
  • Fox and Albertson (2011) conclude that the case for PbR in the criminal justice sector is not yet proven, given lack of agreement on evidence base and difficulty of quantifying savings.

In the pipeline

1)    Proven Reoffending Statistics Quarterly Bulletin April 2009 to March 2010[15]

2)    May 2012 Compendium will include:

  • 2 and 5 year follow up for headline reoffending rates
  • Penalty Notices for Disorder
  • Analysis  of reoffending from end of sentences

3)    Future plans to publish include quarterly comparisons by prison, Community Safety Partnership, Local Authority and Probation Trust, and a measure of re-offending while on license.

4)    Reconviction analysis of the impact of the Peterborough Scheme expected 2013-14.

5)    Full roll out of payment by results principles is planned to begin in  2015.