Measuring the Dynamic Risk of Further Offending: A Feasibility Study

Abstract

Background
Young people who have offended were, until recently, assessed using the Core ASSET Profile – a tool which determined their likelihood of reoffending based on their criminal history and practitioner scores across 12 domains. The repeated assessments provide a set of data for each individual which can be used to model how their risk of further offending varies over time. Having conducted an initial proof of concept study, this work explores the potential of extending the range of ‘risk’ and ‘protective’ factors using anonymised linked data held within the SAIL Databank.

Main Aim
The feasibility study is designed to establish the potential for using administrative data to develop a more sensitive assessment tool for use in the youth justice system. Specifically, the study explores the impact of being care experienced and of subsequent system contact in elevating the risk of further offending.

Approach
A series of Bayesian hierarchical models will be generated which mimic the features of the Core ASSET Profile under the Scaled Approach. These include a range of time-varying and non-time varying variables matched to the individual, drawn from education, health and social services datasets as well as their court and offending records.

Results
The anticipated findings will advance our understanding of how the likelihood of further offending varies over time for different groups, and how further system contact increases the risk. This will enable the complexity of young people’s real lives to be explored, and hence appropriate and timely interventions to be developed.

Conclusion
Modelling under a Bayesian framework affords the opportunity to generate robust analysis based on smaller datasets. Findings have significant implications for policy and practice, particularly in the context of assessment processes across the justice system and social welfare.

Background

Young people who have offended were, until recently, assessed using the Core ASSET Profile – a tool which determined their likelihood of reoffending based on their criminal history and practitioner scores across 12 domains. The repeated assessments provide a set of data for each individual which can be used to model how their risk of further offending varies over time. Having conducted an initial proof of concept study, this work explores the potential of extending the range of ‘risk’ and ‘protective’ factors using anonymised linked data held within the SAIL Databank.

Main aim

The feasibility study is designed to establish the potential for using administrative data to develop a more sensitive assessment tool for use in the youth justice system. Specifically, the study explores the impact of being care experienced and of subsequent system contact in elevating the risk of further offending.

Approach

A series of Bayesian hierarchical models will be generated which mimic the features of the Core ASSET Profile under the Scaled Approach. These include a range of time-varying and non-time varying variables matched to the individual, drawn from education, health and social services datasets as well as their court and offending records.

Results

The anticipated findings will advance our understanding of how the likelihood of further offending varies over time for different groups, and how further system contact increases the risk. This will enable the complexity of young people’s real lives to be explored, and hence appropriate and timely interventions to be developed.

Conclusion

Modelling under a Bayesian framework affords the opportunity to generate robust analysis based on smaller datasets. Findings have significant implications for policy and practice, particularly in the context of assessment processes across the justice system and social welfare.