Department of Computational Social Science Seminar Abstract

Wednesday, May 9 - 2:00 p.m.
Sub II (The Hub), Meeting Room #1

Computational Modeling of Neighborhood Crime from Sociological and Economic Perspectives

Steven P. Wilcox, PhD
Computational Social Science
George Mason University

This research extends the understanding of violent crime from economic, complexity and sociological perspectives, using agent-based modeling as its methodology. Much has been done with agent-based models of violent crime “hot spots”, modeling them in terms of cat-and-mouse games, intelligent agents, and the importance of place, but economic and sociological perspectives also have promise for enriching the modeling of crime. To fill the gap, this research examines the utility of economic models by Glaeser et al. (1996) and Calvó-Armengol and Zenou, (2004) for replicating the intriguing variability properties of city-level crime rates, and then examines the robustness of these models and the options for making them correspond better to real-world topologies. A concern with the economic approach is that the model of Glaeser et al. may not track with the variability-scaling properties of actual urban crime rates as seen in the FBI data. Then it builds a bridge to the sociological literature of crime at the neighborhood level by considering how to extend these models to cover social influence by criminals and anti-crime neighbors. Using agent-based modeling, an approach to modeling norms as a form of collective action based on game theory will be developed. Restricted data from the Project on Human Development in Chicago Neighborhoods (Sampson et al., 1997) will be requested for the purpose of validating the model at the neighborhood level. By incorporating social control theory into an effective agent-based model of neighborhood crime that also has the correct scaling properties at the macro level, I hope to lay a new foundation for advancing computational research on criminological theory and the management of the crime problem.