Summary of the impact

In the mid-2000s the Ministry of Justice (MoJ) devised a new measure to compare area variations in reconviction rates across the Probation Service in England and Wales so that these differences could be taken into account when allocating resources. A number of Probation Trust Chief Executives have used Hedderman's research successfully to argue for revisions to the reconviction 'performance measure'. Her findings also influenced the Justice Select Committee's recommendation that the original measure should be replaced, as she showed that it led to unfair comparisons, was easy to manipulate, and failed to provide information which could be used by areas to improve their impact on reoffending. She has since worked directly with Kent, London and Hertfordshire Probation Trusts to address this last point.

Underpinning research

Probation statistics routinely show considerable geographic differences in reoffending rates. For example, the range in 2010 for those commencing supervision was from 26.6% in Warwickshire to 44.7% in Durham and Tees Valley. However, before these figures can be interpreted as evidence of differences in their effectiveness, the extent to which areas supervise different types of people needs to be taken into account. A further difficulty is that there are no simple, agreed ways of measuring reoffending although, over time, best practice lessons have developed (1). Hedderman (who joined the Department of Criminology as Professor in 2004) conducted analyses which showed that the performance measure developed by the MoJ failed to fully reflect those lessons. Consequently, it did not produce fair or useful comparisons between areas; nor did the results help Probation Trusts to understand what they could do to improve their performance.

Hedderman's previous work on sentencing and reoffending (1) led the five East Midlands Probation areas (Derbyshire, Leicestershire and Rutland, Lincolnshire, Northamptonshire and Nottinghamshire) to commission her in 2006 to investigate why they had different reoffending rates to each other, and why these changed over time, so that they could target their efforts to reduce reoffending more effectively. They also wanted expert advice on how fairly the MoJ measure assessed their performance.

Based on a quantitative analysis of just under 50,000 probation and police records, the initial analysis of East Midland's data demonstrated that the MoJ method (based on caseload rather than commencement samples) was a poorer measure of reoffending than other existing reconviction measures (2). It undercounted those on orders of less than three months and over-counted those who were at low risk of reoffending (having been on the caseload for very long periods without doing so). Variations in these factors between areas were not taken into account, leading to unfair comparisons being drawn between areas. The measure also created perverse incentives. For example, a Trust could improve its performance by encouraging the courts to use short prison sentences which were not included in the calculations. Not only are such sentences more expensive in financial terms, they may increase rather than decrease the risk of further offending (3).

These criticisms led the MoJ to amend their preferred measure to include a variable which took account of the differing lengths of time offenders were under supervision. However, they argued against replacing the measure on the grounds that their approach was the only practical one. To combat this claim the same five areas commissioned a new reoffending study from Hedderman and Jolliffe (who worked at the Department of Criminology as a Senior Lecturer until 2012). Based on a sample of over 9,000 cases, this study demonstrated that it was both practical and preferable to adopt a conventional longitudinal approach by taking quarterly samples of cases commencing supervision. In line with best practice in the field of reoffending studies, this approach focused on the period when the risk of reoffending was greatest (within six months of commencing supervision). The longer follow-up timescale minimised the impact of factors such as variations in court processing times so the results were a more genuine measure of underlying offending behaviour. Using this approach it was also possible to identify action which might be taken to reduce reoffending. For example, one area was advised to speed up first appointments and another is now targeting those convicted of theft offences with employment and debt advice services.

References to the research

Grants:

The initial research was funded by two research grants from five Probation Trusts (Derbyshire, Leicestershire and Rutland, Lincolnshire, Northamptonshire and Nottinghamshire):

£24,000 commencing 28 March 2006; Principal Investigator Carol Hedderman

£32,635 commencing 1 December 2008; Principal Investigator Carol Hedderman

Details of the impact

Hedderman's initial analysis of reconviction in the East Midlands demonstrated that the MoJ method for comparing different probation areas, which was based on caseload rather than commencement samples, was a poorer measure of reoffending than other existing reconviction measures (2). Initially, the MoJ changed their measure to make some allowance for the length of time offenders were under supervision, but they resisted making more substantive changes. Subsequently, the government's position changed. This reflected pressure from within and outside the Probation Service following the dissemination of the results of Hedderman's second study (with Jolliffe). For example, the study was cited in a House of Lords debate in 2010 and showcased in an invited presentation to the Probation Chiefs Association National Conference in 2011. Hedderman was then invited to give evidence before the Justice Select Committee's inquiry into the Probation Service in 2011 which led to the Justice Select Committee (A p.114) being able to 'welcome the MoJ's review of the local snapshot measure of re-offending...' and the MoJ (B p.31) responding that they were 'investigating whether meaningful data on the re-offending of offenders supervised by individual probation trusts can be produced after controlling for changes in offender characteristics'.

In 2011 the MoJ created a new local 'proven reoffending' measure which is based on commencement samples and includes pre-court disposals which again increases the degree to which the measure assesses underlying offending behaviour rather than criminal justice processing decisions. Ultimately this measure is expected to form the basis of a replacement for the current probation performance measure.

A number of Probation areas subsequently contacted Hedderman for advice on how to use reconviction information to improve the work they do with offenders to reduce reoffending. This led to formal consultancies with three Probation Trusts:

£10,350 (ex VAT) commencing 2010, Kent Probation Trust.

£19,000 (ex VAT) commencing 2012, London Probation Trust (renewed April 2013 for the same amount for a further year).

£10,450 (ex VAT) commencing 2013, Hertfordshire Probation Trust.

The work for Kent Probation Trust involved Hedderman and Jolliffe analysing reconviction data to identify ways to improve their work on reducing reoffending. In 2012, Hedderman was appointed as Research Advisor to London Probation Trust (which manages 20% of all offenders under probation supervision) to advise them on which further analyses to conduct in relation to reoffending (as well as other matters). This relationship proved so valuable to them that this consultancy was renewed in April 2013 for a further year (C).

[text removed for publication]

Since the results were provided, reoffending rates in the East Midlands Region (even using the Government's preferred measure) have outperformed the predicted level (8.76% proven reoffending as opposed to a predicted figure of 9.03% in the most recent year) (E).

Hedderman and Jolliffe were also invited to be the key note speakers at a specially convened seminar `Sentencing and reducing reoffending: some reflections on the British Experience' in Santiago, Chile on 5 July 2011. The event was sponsored by the British Embassy in Chile, the Chilean Ministry of Justice and Chilean Public Criminal Defender. Hedderman was subsequently appointed advisor to a project funded by the Organisation of the American States (OAS) to develop outcome measures for comparing the impact of prison and community-based interventions across a number of countries in Latin America and the Caribbean (Chile, Uruguay, Costa Rica, Barbados, Panama, Dominican Republic and Jamaica). In May 2013, the OAS Director of the Department of Public Security wrote to Hedderman noting that OAS 'are convinced that this project will make a significant contribution to the efforts of our Member States' (F).

Sources to corroborate the impact

Evidence relating to change in measure

A. Justice Select Committee (2011) The role of the Probation Service, Eighth Report of Session 2010-12,
http://www.publications.parliament.uk/pa/cm201012/cmselect/cmjust/519/51902.htm

B. Ministry of Justice (2011) Government Response to the Justice Committee's Report: The role of the Probation Service, http://www.justice.gov.uk/downloads/publications/corporate-reports/MoJ/government-repsonse-role-of-probation-service.pdf

Evidence relating to the value in focusing probation efforts to reduce reoffending

C. Contact details for Chief Executive, London Probation Trust.

D. Contact details for Chief Executive Officer, Hertfordshire Probation Trust.

E. Contact details for Leicestershire and Rutland Trust Probation.

Evidence of international impact

F. Contact details for Director of Public Security, OAS