Temporal analysis of the clustering and hypothesized social contagion of mass killing events in the United States

Mass killings (e.g., 2016 Pulse nightclub shooting, 2017 Las Vegas shooting) are tragedies that devastate the victims’ families and harm the local communities and the nation at large. Amid an increase in mass-killing research, the idea that these events may be contagious has emerged among scholars and been publicized in popular media. This paper implements a three-phase methodology to evaluate the social contagion hypothesis for US mass killings that (a) detects prominent contagion effects, (b) identifies mass killing clusters, and (c) detects subtle contagion effects. Evidence of a prominent contagion effect was not found, utilizing a periodically-observed time-homogeneous Poisson process framework. It is shown that the occurrence of family mass killings and the occurrence of felony mass killings were homogeneous and temporally random between 2006 and 2023, whereas the rate of public mass killings during this timeframe approximately doubled starting in late 2015. The occurrence of public mass killings was homogeneous and temporally random when separated at this arrival rate changepoint. Event clusters were identified and compared to Poisson bursts with respect to three attributes: number of clusters, duration, and surprise. The relationship between event notoriety and the time until the subsequent event was also evaluated. Analysis of the relationship between event notoriety and the time until the next mass killing revealed some irregularities that, while not consistent with a subtle contagion effect, invite future qualitative research investigating specific clusters for evidence of behavioral transmission. The ten highest-density clusters for each mass killing type are reported to facilitate future research.