Comparing Mobility and Spatial Pathways in the Hourly Prediction of Violent Crime

Abstract

Much research has documented the contagiousness of violence. Some of this work has focused on contagiousness as operationalized by the spread across geographical space, while other work has examined the spread within social networks. While the latter body of work struggles with incomplete network data, the former constitutes a theoretical mismatch with how violence should spread. Theory instead strongly suggests that violence contagion should diffuse through everyday mobility networks rather than just adjacently through geographical space. Beyond contagion itself, I argue that neighborhoods connected through mobility networks should serve as useful short-term sensors in predicting imminent violence because these sets of residents tend to experience shared environmental exposures, which may induce synchrony in the likelihood of violence. I explore this topic and these relationships using violent crime data from the three largest U.S. cities: New York City, Los Angeles, and Chicago. Using two-way fixed effects models, I test whether or not violence in mobility-connected alter neighborhoods in the preceding hour predicts violence in an ego neighborhood in the next hour. Across all three jurisdictions, I find that recent violence in the neighborhoods to which a neighborhood is connected through mobility ties can strongly predict that neighborhood’s odds of subsequent violence. Furthermore, spatial proximity has no significant effect on the likelihood of violent crime after controlling for mobility ties. I conclude by arguing that mobility patterns are an important pathway in the prediction of violence.