The Friendship Paradox Can Get you Killed: The Predictive Value of Network Centrality on Gunshot Victimization

• In Event: Criminal Network Analysis: Applications of Social Network Analysis to the Study of Co-Offending, Gangs and Violence

Abstract

The friendship paradox is a well-known mathematical feature of almost all networks; your friends to have more friends than you do. This paradox has also been proven useful to understand contagious diseases: your friends are more likely to be sick than you are. This phenomenon has been leveraged by using people more central to the network as early censors for contagious diseases outbreaks. Research on violence suggests that victimization is highly concentrated among offenders more central to the network, and follow diffusion patterns, similarly as a disease would do. In the present paper, we observe if the friendship paradox applies to co-offending networks and if this phenomenon could be useful to predict upcoming violence outbreaks using co-arrest network and shooting victimizations from Chicago (2006-2013). Point process modeling of time before victimization suggests that more central victims can help to predict a violence at the city level earlier than the number of victims with low degree or the number of victims that were not part of the network. These results suggest that victims’ network centrality could be use as early censors for violence outbreaks that could guide interventions.

Authors

• Yanick Charette, Université Laval

• Andrew V. Papachristos, Yale University