Unveiling the Unseen: Documenting and Analyzing Nonfatal Shootings by Police

The 2014 fatal shooting of Michael Brown by Officer Darren Wilson in Ferguson, Missouri, prompted what criminologist Lawrence Sherman called the Second Great Awakening of “both public and scholarly sentiment against avoidable police shootings.”^1(p424) (The First Great Awakening occurred in the 1970s and 1980s, when 50 large cities prohibited officers from shooting nonviolent, fleeing suspects.) Since then, an interdisciplinary array of scholars have drawn on newly available, more comprehensive databases tracking deadly police‒citizen interactions (e.g., Fatal Encounters, Mapping Police Violence, and The Washington Post’s Fatal Force database) to test hypotheses about the causes^2,3 and public health consequences⁴ of these interactions, as well as disparities therein.^5,6 This body of research has revealed an annual average of approximately 1000 fatalities attributable to police gunfire in the United States. However, the scope of these studies has been limited by the paucity of data on nonfatal police shootings.⁷ Though nonfatal police shootings, by definition, do not involve fatalities, they nevertheless constitute uses of deadly force (i.e., force likely to cause death) by police officers.

A WELCOME ADDITION TO THE LITERATURE

Ward et al. (p. 387) have addressed this limitation by meticulously abstracting and manually verifying data from the Gun Violence Archive (GVA) to compile the first multiyear, nationwide analysis of injurious shootings by US police officers. Several key findings emerged. First, their work reveals that from 2015 to 2020, 4741 people were nonfatally injured by police gunfire—an average of 790 per year. Second, the authors document clear incident- and person-level differences in police shooting outcomes. For example, compared with police shootings involving unarmed persons, those involving persons armed with a knife or other cutting instrument were nearly twice as likely to be fatal (odds ratio = 1.92, see Table 4). Meanwhile, police shootings involving non-Hispanic Black victims were less likely than those involving non-Hispanic White victims to be fatal (though readers should bear in mind that race/ethnicity was unknown for 29% of victims).⁷ Finally, the authors point out that some groups are overrepresented in injurious police shootings given their representation in the US population. As but one example, “[u]nhoused victims comprise nearly 3% of injured people, despite representing just 0.2% of the US population” (p. 394). Without knowing how often police interactions with unhoused persons do not result in the use of deadly force, it is difficult to make sense of this disparity. But, to be sure, this is an understudied population in the police-use-of-force literature, so simply calculating this disparity is an important contribution.

The next step for researchers is to focus on discerning the causal mechanisms at work here. Are officers firing more rounds, on average, when they confront people armed with knives?⁸ Does proximity to trauma care⁹ or agency policies regarding when officers can render aid¹⁰ account for some of the observed victim-level differences in shooting outcomes? Are officers more likely to use deadly force when interacting with unhoused persons? Hopefully this study by Ward et al. will inspire researchers to do the additional work necessary to unpack some of their findings and answer these questions.

INTRODUCING SPOTLITE

It bears mentioning that Ward et al. restricted their focus to injurious shootings by police officers, leaving it unclear how often noninjurious police shootings occur. It seems unlikely that these incidents are as reliably reported by local media and subsequently included in the Gun Violence Archive. However, since Ward et al. completed their study, a new data set has been made available by the Cline Center for Advanced Research at the University of Illinois Urbana-Champaign. Their Systematic Policing Oversight Through Lethal-force Incident Tracking Environment (SPOTLITE) data primarily originate from Gun Violence Archive and Fatal Encounters and include “any discharge of a firearm by law enforcement personnel as well as any other use of force by law enforcement personnel that produces a lethal outcome.”¹¹

Unfortunately, at this time, the SPOTLITE data are incident-level, and users cannot easily determine which incidents resulted in fatalities, nonfatal injuries, or no injuries (the Cline Center is still working on an individual-level data set, which they will release later). However, users can merge SPOTLITE with GVA using the gva_id variable. Doing so reveals that SPOTLITE extracted 14 320 deadly force incidents from GVA between 2015 and 2020—an average of 2387 per year (data downloaded January 4, 2024). Ward et al. extracted 10 308 injurious shootings over the same period, or 1718 each year on average (reported in Table 1). If we assume the difference in each research team’s total is noninjurious shootings (i.e., incidents wherein officers shot and missed), it would mean there were 4012 such shootings from 2015 to 2020, or 669 each year on average.

Relying on the figures Ward et al. reported in Table 1, this would mean that from 2015 to 2020, approximately 31% of all known police shootings resulted in at least one fatality, 41% resulted in nonfatal injuries, and 28% resulted in no injuries. It would also mean that all the research published in the last 10 years drawing on data from The Washington Post, Fatal Encounters, and Mapping Police Violence was relying on a nonrandom sample of only about one third of all incidents involving police use of deadly force. Depending on the research question, complete omission of two thirds of the phenomenon we seek to understand may produce statistically biased estimates of the causes and consequences of said phenomenon, or a misunderstanding of the causal mechanisms altogether. Thus, many of these studies will likely need to be replicated with GVA or SPOTLITE data.

CONCLUSION

Including nonfatal shootings in research on the causes and consequences of police use of deadly force is critical, because, as Ward et al. correctly note, “[U]nderestimating the true scale of injury impact is a further injustice and may obstruct progress toward preventive action and reforms” (p. 394). And, to be clear, we should be trying to prevent unnecessary and avoidable uses of deadly force (i.e., force likely to cause death)—not merely those that do, in fact, result in death. Going forward, I encourage scholars working in this space to take advantage of the data being compiled by GVA and SPOTLITE, which are more comprehensive than any of the other data sets available at this time.