State of the Art in Agent-Based Modeling of Urban Crime: An Overview

Abstract

Objectives

Agent-based modeling (ABM) is a type of computer simulation that creates a virtual society and allows controlled experimentation. ABM has the potential to be a powerful tool for exploring criminological theory and testing the plausibility of crime prevention interventions when data are unavailable, when they would be unethical to collect, or when policy-makers need an answer quickly. This paper takes stock of the current literature to discuss the potential contributions of ABM, assess current practice, identify shortcomings that threaten the validity of findings using ABM, and to make suggestions regarding the construction and communication of future work using ABM.

Methods

We systematically searched major databases to find all publications using ABM to simulate urban crime patterns and coded publications to quantify the following information: (1) characteristics of the publication, the model and the agents, (2) model purpose, (3) crime type investigated, and (4) interrogation of the model via sensitivity testing and validation.

Results

After sifting papers according to our inclusion criteria, we identified and reviewed 45 publications. Models informed by the opportunity theory framework dominated. Most publications lacked detail sufficient to enable replication. Many did not include clear a rationale for modeling choices, parameter selection or calibration. Rarely were parameters calibrated using empirical data. Model validation was limited and inconsistent across papers.

Conclusions

ABM offers significant potential for criminological enquiry. However, at present, the lack of model detail reported in publications makes it difficult to assess where sufficient evidence exists to support—and where gaps limit—the development of models that reflect extant conditions and offender decision-making. For the field to progress, as a minimum, standardized reporting that encourages transparency will be necessary.