Random Acts of Violence? Examining Probabilistic Independence of the Temporal Distribution of Mass Killing Events in the United States

Abstract

Recent mass killings, such as those in Newtown, Connecticut, and Aurora, Colorado, have brought new attention to mass killings in the United States. This article examines 323 mass killings taking place between January 1, 2006, and October 4, 2016, to assess how they are distributed over time. In particular, we find that they appear to be uniformly distributed over time, which suggests that their rate has remained stable over the past decade. Moreover, analysis of subsets of these mass killings sharing a common trait (e.g., family killings, public killings) suggests that they exhibit a memoryless property, suggesting that mass killing events within each category are random in the sense that the occurrence of a mass killing event does not signal whether another mass killing event is imminent. However, the same memoryless property is not found when combining all mass killings into a single analysis, consistent with earlier research that found evidence of a contagion effect among mass killing events. Because of the temporal randomness of public mass killings and the wide geographic area over which they can occur, these results imply that these events may be best addressed by systemic infrastructure-based interventions that deter such events, incorporate resiliency into the response system, or impede such events until law enforcement can respond when they do occur.