Empirical analysis uncovers incident patterns and victim profiles in U.S. houses of worship, suggesting urgent safety measures.
Houses of worship (HoW) are viewed as sanctuaries for American communities, yet recently have become settings of increased targeted violence. Despite extensive research on school shootings, empirical studies of attacks at HoW are scarce. This study draws on the Violence Prevention Project Research Center's House of Worship Homicides Database (N = 380 incidents, 2000-2024) to provide an empirical portrait of perpetrator profiles and incident patterns. Results reveal sharp concentration in the high-harm tail: 96.8% of incidents fall within the non-ideological/uncoded cluster, most incidents occur outdoors (70.8%), and most produce single-death outcomes (88.4%). The strictly coded ideological subset represents only 3.2% of events but accounts for 19.7% of total harm; its share rises from 11.1% among incidents harming at least two people to 69.2% among incidents harming at least six. A Bayesian negative binomial comparison indicates that ideological incidents are expected to produce about 7.5 times more total harm than other incidents (95% credible interval: 5.53-9.99), with limited posterior predictive overlap (15.0%). Unknown/Uncoded incidents do not display the same upper-tail concentration, declining from 28.4% of all events to 7.7% of incidents harming at least six people. Among codable perpetrator records, criminal histories appear in 88.6% of cases.
No takes yet. Share an insight, caveat, or question.
Dulisse et al. (2026) studied this question.