Objectives
This study investigates the impact of three police deployment strategies on emergency response times using agent-based modelling (ABM). Specifically, it evaluates the effectiveness of random patrol, stationary deployment (optimal spreading), and last location deployment (idling at last incident location). It further examines how key variables – urbanisation, call volume, and police capacity – moderate these effects.
Methods
A detailed ABM was developed using NetLogo, integrating real-world data: historical calls-for-service (CFS), jurisdiction shapefiles, and street network data from the Netherlands. The model simulated police travel and response dynamics across 120 scenarios, varying deployment strategies, urbanisation levels, call volumes, and police capacity levels. Outputs were analysed to assess response times, fast response rates (< 3 minutes), and late response rates (>13 minutes). Methods were pre-registered at https://osf.io/yrwdp/.
Results
On average, stationary deployment reduced response times by 35% (SD ±14%), increased fast responses by 74% (SD ±40%), and decreased late responses by 66% (SD ±33%) compared to random patrol. Last location deployment also outperformed random patrol, reducing response times by 13% (SD ±9%), increasing fast responses by 22% (SD ±14%), and reducing late responses by 42% (SD ±36%). The advantages of stationary and last location deployment were most pronounced in rural areas and at lower police capacity levels. Urbanisation reduced the performance gap between strategies, while higher call volumes modestly diminished the relative benefits of stationary deployment.
Conclusions
This study highlights the significant impact of police deployment strategies on response times and rapid interventions. These findings underscore the need for further research on rapid response. The modular ABM framework offers a valuable tool for adapting investigations to different policing contexts, enhancing external validity.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)