Background

Within the Probation Board for Northern Ireland (PBNI) the Assessment, Case Management and Evaluation system (ACE) assessment tool is currently the main generic assessment tool used to measure the likelihood of general reoffending over a two year period. ACE was chosen because of its completion in partnership with the service user, its relative ease of use, and its current embedded use in probation practice. It is applied to all service users, except for those convicted of solely sexual offending. It is used at an individual level to routinely measure changes (reductions or increases) in the overall likelihood of reoffending. ACE is completed at the beginning of PBNI’s intervention and at regular intervals and/or at the end of probation supervision.

In summer 2024 PBNI commissioned the Northern Ireland Statistics and Research Agency (NISRA) to complete a data linkage project to help determine the effectiveness of the Assessment, Case Management and Evaluation system (ACE) assessment tool.

Data for individuals supervised by PBNI during 2019/20 and 2020/21 was linked with the Department of Justice (DoJ) reoffending dataset and analysis carried out in relation to gender, age, disposal type and number of reoffences.

This report summarises the findings from the data linkage project. The full detailed report can be accessed on the PBNI website.

Key Findings

Overall, it was determined ACE assessments are a significant predictor of reoffending behaviour.

  • Both at the base date (date of release from custody or court date) and the end of Order/Licence a greater proportion of individuals categorised by ACE as being at high risk went on to reoffend. Just over one-quarter of those who were categorised as medium risk reoffended, whilst those categorised as low risk had the smallest proportion of individuals who reoffended.

  • Across genders, a greater proportion of individuals categorised as high risk at the end of their Order/Licence went on to reoffend, followed by medium and low risk respectively. This link was found to be statistically significant with a strong association across both genders.

  • Similarly, across all age bands a greater proportion of those assessed as high risk went on to reoffend, followed by medium and low risk respectively. This link was found to be statistically significant with a strong association across all age bands.

  • There appears to be a pattern among ACE risk assessment and number of reoffences. Over four-fifths of individuals categorised as high risk (82.5%) went on to commit more than one reoffence; 12.8% committed eleven or more offences. Of those categorised as low risk who went on to reoffend, almost four fifths (79.6%) committed three or fewer reoffences.

  • For both custodial and non-custodial disposals, the greatest proportion of individuals who reoffended were categorised as high risk, followed by medium and low risk respectively; the same pattern was seen when looking at each of the different order types individually.

  • Across all ACE bands, a smaller proportion of those released from custody went on to reoffend compared with non-custodial disposals.

Methodology

Data from the PBNI electronic case management system (ECMS) was linked with the DoJ reoffending dataset which is sourced from the Causeway Data Sharing Mechanism (primarily the Criminal Records Viewer (CRV) on Causeway) and supplemented by information from the Northern Ireland Prison Service case management system (PRISM) and the Youth Justice Agency (YJA).

The initial PBNI data extract contained custodial cases (Determinate Custodial Sentences, Extended Custodial Sentences and Juvenile Justice Centre Orders) where the actual date of release (ADR) was within 2019/20 or 2020/21 and non-custodial cases (Custody Probation Orders, Combination Orders, Enhanced Combination Orders or Probation Orders) where the order start date was within 2019/20 and 2020/21.

For each case, two key ACE assessments were identified: a ‘base date ACE’ which is the ACE assessment closest to (either before or after) the actual date of release or the order start date of the order/licence, and an ‘end ACE’ which is the ACE assessment closest to (either before or after) the order end date. The overall offending related score (ORS) for each of these ACE assessments was included in the data extract.

Data linkage was carried out by statisticians within the Analytical Services Group (ASG) within DoJ who have responsibility for the reoffending datasets. Records were matched using the unique reference number (PartyURN), which is distinct to an individual service user, and the base date (defined as the release date for custodial disposals and the court date for non-custodial disposals). The overall match rate between the two datasets was 75.2%.

Initially comparison of ACE risk bands and reoffending was carried out for both the base date ACE and the end ACE. While assessments at both points indicated a statistically significant link with reoffending, looking at effect size, there was a stronger association at the end of the order/licence. For this reason, all subsequent analysis was based on the ‘end ACE’.

The chi-square test was used to determine the statistical significance of the results whilst Cramer’s V was used to determine the strength of the association.