Selectivity Bias in the Social Networks of Deceased and Non-Deceased in Serious and Organized Crime

• In Event: Recent Advances in Network Approaches to Study Gangs and Criminal Organizations

Abstract

Where do killings happen in criminal social networks? Social network analyses of gang violence shows that fatal and non-fatal gunshot incidents are non-random events that are reciprocated with violence, and that risk of injury or death from residual violence increases with ‘closeness’ to gang members. The accuracy of predictions of who gets shot and/or killed relies on the relative ‘completeness’ of criminal social networks that are subject to missing data bias. Because (gang-related) homicides are rare events, it’s reasonable to expect a ‘spotlight effect’ on the deceased and their social networks that counters the biases induced by missing data. This study considers differences in the social networks of deceased and non-deceased career criminals using data sourced from police intelligence records of serious and organized crime in the Canadian province of Alberta. Our data is a social network of 1,700 people, with nearly complete coverage of gang-related homicide victims, connected by 3,100 co-offending incidents, prison contacts, business partnerships, family members, and social events. Knowing who is recorded within and across police intelligence records makes it possible to test for selectivity bias in the social networks of the deceased versus a random control group and gang leaders.

Authors

• Mitchell Macdonald, Simon Fraser University

• Martin Bouchard, Simon Fraser University