Five Handshakes · Networks and the Risk of Gunshot Injury
AFTER PAPACHRISTOS, BRAGA & HUREAU (2012) · BOSTON · SYNTHETIC NETWORK, PUBLISHED STATISTICS
Shading gives each person's distance, in ties, to the nearest gunshot victim. Risk falls off with every step.
Tap any person in the network, or press pick someone at random. The panel reports their position and traces the shortest chain of ties from them to the nearest person who has been shot.
Slopes are the published odds ratios. The intercept is set so each curve passes through the sample base rate at the sample mean distance; the authors do not report one, so absolute levels here are illustrative and only the shape is faithful.
Individual and neighbourhood risk factors explain why some places have more gun violence than others, yet most people in a high-risk population are never shot. Papachristos and colleagues asked a structural question instead: where does a person sit relative to others who have been shot? In Boston, 85% of the gunshot victims in a 763-person network fell inside a single connected component, and each additional tie separating someone from a victim cut their odds of being shot by about a quarter. The Chicago replications found the same shape at city scale.
The network shown is synthetic, generated to reproduce the published structural statistics. It is not the authors’ data, which are not public. Click any badge above for the measurement details, the limits of police-observed ties, and the contagion-versus-homophily problem.
References: Papachristos, Braga & Hureau (2012) · Papachristos & Wildeman (2014) · Papachristos, Wildeman & Roberto (2015) · Green, Horel & Papachristos (2017) · Papachristos (2009) · Braga, Papachristos & Hureau (2010) · Shalizi & Thomas (2011) · Saunders, Hunt & Hollywood (2016)
Full text of the Boston study is open access at PubMed Central.