Re-spatializing Gangs: An Exponential Random Graph Model of Twitter Data to Analyze the Geospatial Distribution of Gang Member Connections
Keywords:gangs, location, Twitter, exponential random graph model (ERGM)
Gang studies often use location-based approaches to explain gang members’ interconnectedness. Although this perspective remains consistent with the proximity principle that the smaller the geographic space, the greater the likelihood of observing connections between individuals, location-based studies limit our understanding of gang member connections to narrowly defined geographic spaces at specific points in time. The advent of social media has re-spatialized gang member interconnectedness to unbounded geographic spaces, where the preservation of online activity can extend indefinitely. Despite having an online presence, most research examining the digital footprint of gangs tends to be descriptive. This study collects Twitter data to analyze the geospatial distribution of gang member connections using an exponential random graph model (ERGM) of location homophily. An ERGM analyzes network substructures to determine the patterns of relationships between vertices. In this case, the extent to which homophily by city, state, and gang affiliation determine gang member connections. The results of this study support the proximity principle but challenge the assertion that gangs are strictly localized.