ABSTRACT
This study uses the routine activities framework to identify motivation, target, and guardian characteristics influencing the severity of mass shooting fatalities and injuries. Significant findings indicate media-driven motivations, particularly fame-seeking perpetrators, produced more casualties. Open-spaces and schools provided more suitable targets, with open-spaces incurring more fatalities and schools incurring more injuries. Guardianship variables indicated perpetrators with a history of mental illness, as well as incidents involving rifles, more than one gun, and ending in the perpetrator’s death, all resulted in higher rates of victimization. A discussion of findings highlights targeted policy and security strategies aimed at reducing the victim-counts attributed to mass shooting attacks.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1. Open-spaces can include such locations as malls, restaurants, clubs, bars, and events (Silva & Capellan, Citation2019b).
2. Scholars suggest a time period this extensive may skew the reality of the phenomenon (i.e. publicity and time-period effects) (Silva & Greene-Colozzi, Citation2019a). As such, it is important to consider the rate of victimization between 2000-2009 and 2010-2018, which would not be influenced by publicity and time-period effects. Only taking into account victimization rates at the turn of the century, findings indicate 65% of fatalities (n = 622) and 78% of injuries (n = 1,213) occurred in the most recent nine years of this study.
3. The 2017 Las Vegas shooting was removed from all bivariate injury analyses. This incident produced an enormous number of injuries (n = 500), which was 100x higher than the overall average (i.e. 5 injuries). This outlier was skewing, and even reversing, the mean scores. As such, this work followed previous mass murder, mass shooting, and extremist violence studies that remove extreme outliers “because [their] enormity and special character would grossly distort the statistical results” (Fox & Levin, Citation1998, p. 432).
4. It is important to note, the Poisson regression also allows for consistent estimates using count data, but it was not used because it has more restrictive distributional assumptions than the negative binomial regression. The Poisson model requires means and variances to be equal, and initial analyses found the variances of both fatalities and injuries were much larger than the means. As such, the dependent variables were over-dispersed, and the negative binomial model was determined to be a better fit (see Blau et al., Citation2016; Yelderman et al., Citation2019).
5. See the Department of Homeland Security’s “See Something Say Something” campaign.
6. California, Maryland, and New Jersey.