Agent-Based Modelling of Burglary

• In Event: Spatial and Temporal Analysis

Abstract

Crime research is often focused at the aggregate level; consequently crime prevention policies derived from this type of research are predominantly based upon such aggregated patterns. However, this can lead to a misunderstanding of crime patterns at lower (micro and meso) levels. The use of computational-modelling seeks to overcome these issues. In particular, Agent-Based Modelling enables researchers to model human behaviour at the level of the individual person. The current paper presents an Agent-Based Model (ABM) of residential burglary. Whilst the behavioural preferences of offenders in previous ABMs of burglary have been based predominantly upon criminological theory, this model has been directly informed through semi-structured interviews with a sample of burglars from the city of Leeds, in Northern England. From these interviews, offender traits were classified using a quantitative, cluster-based (‘Latent-Class’) analysis, developed using offenders' behavioural preferences and target selection criteria. These classifications help to illustrate the heterogeneity of offenders within Leeds. The model developed will help to explore the impact of different crime reduction initiatives and Policing styles with different offenders, whilst assessing the extent to which offending by different offender groups may be estimated.

Authors

• Nicholas Addis, University of Leeds, UK

• Andrew Evans, University of Leeds, UK

• Susanne Karstedt, University of Leeds, UK

• Nicolas Malleson, University of Leeds, UK