Chicago native and Yale sociology professor Andrew Papachristos has discovered that social networks are the single most predictive factor in violent crime in his home city.

Social networks constructed from arrest records explained violent crime better than gender, race or gang affiliation in a subset of the Chicago population. While other studies have looked at crime as a disease, this study is among the first to rigorously model crime as a disease. Papachristos is now collaborating with police departments to identify those at most at risk for committing a crime, as opposed to more traditional methods of crime prevention that target entire demographics.

“This is the anti-stop-and-frisk,” Papachristos said.

The researchers used arrest records to construct the social network, assuming that those arrested together knew each other. The analysis revealed that for every social tie separating an individual from a homicide victim, one’s chance of being killed decreased by 57 percent. While this result came as no surprise to Papachristos, who has been working with social networks for years, study coauthor Yale sociology professor Christopher Wildeman said he did not expect the influence of social networks to be so great.

Both authors said they were surprised by the additional finding that gang affiliation had a nearly negligible effect on one’s likelihood of being the victim of a homicide. However, Papachristos warned that gang affiliation may have been misreported through the police data.

While the data for the Chicago analysis comes from a socioeconomically homogeneous segment of Chicago, Papachristos said the results seem to hold in the early stages of a citywide analysis.

Papachristos added previous research he did in Boston came to the same conclusion that social networks are the most important factor for predicting violent crime, and is now working to expand the study to cities in Connecticut and California.

He said he is collaborating to create data solutions for police departments so that they can perform similar network analysis with their own data.

“It’s actually taking what good cops and case workers do and aggregating it,” he said.

There were 506 homicides in Chicago in 2012.