The measure of reoffending used by the Ministry of Justice is called ‘proven reoffending’.
When someone is cautioned, given a ‘final warning’ (the case for young offenders), a conviction that doesn’t involve prison, or is discharged from prison, their records will be monitored for a year. Any offence they commit within the year (that isn’t simply a breach of their probation rules but receives a court conviction, caution or warning) is counted as a proven reoffence.
Proven reoffending can be measured in a number of ways. Figures for the number and proportion of people who reoffend are provided every three months in proven reoffending statistics. This data is broken down by: age, gender, ethnicity, offence, offending history, sentencing history, sentence type and length, local authority and prison.
There are also figures on how much time passes before people reoffend and how their new offence compares to their original offence. This information is published less regularly in the annual release.
Some people already have a criminal record but, because they finished their sentence for their most recent offence more than a year ago, they’re not counted in reoffending statistics.
More detailed figures on offending histories are published as part of the criminal justice statistics release. This is published every three months, with more detail provided in the annual release. This data is broken down by age, gender, type of offence and the length of criminal history.
Policymakers are always looking for ways to reduce the number of people who reoffend. However, the actual reoffending figures aren’t very helpful indicators of success.
For instance, it’s not possible to compare the reoffending rate of people who serve community sentences with those for people who serve prison sentences. These are different ‘sorts’ of people and, when it comes to their likelihood of reoffending, their varying characteristics might be playing more of a role than the type of sentence they’re given.
In response to this, the Ministry of Justice creates its own models which, in defining certain offender characteristics, ‘predict’ how likely people are to reoffend. To define the success of a reoffending rate, the actual value is compared to the model’s ‘expected value’.
There are detailed figures that account for offender characteristics and offer some information on ‘what works’ in reducing reoffending. These are published as part of the annual compendium of reoffending statistics and analysis.
Information on the wider ‘what works’ debate is published by the House of Commons Library.