Since 2015, police in the United States have shot and killed an average of 988 people a year. This is based on reporting by the Washington Post, which began maintaining a national database of such incidents after the controversial 2014 shooting of Michael Brown in Ferguson, Missouri. Without more analysis, the fact that 5,000 Americans lost their lives to armed agents of the state in half a decade is undoubtedly concerning. But context is important, and viewing the number of fatal police encounters in its proper context undercuts the claim—central to the argument advanced by Rizer and Mooney in their thoughtful article—that America has a “police violence problem.” This is not to say that police never use excessive force, nor is it to assert that there is no room for improvement in police training practices that might, at least at the margins, reduce the number of fatal police shootings and other uses of force.
However, to the extent the police violence problem is overstated, so too will be the potential impact of any policy levers pulled to address that problem. This is important because the advisability of a particular policy change will depend in part on how much the change will help if adopted.
Rizer and Mooney propose five policy recommendations to reduce police violence:
- Emphasize and Support De-escalation in Use-of-Force Policy and Training
- Require Greater Transparency Around Department Use-of-Force Policies
- Study and Promote Successful Field Training Officer Programs
- Limit Police Acquisition and Use of Military Resources from the 1033 Program
- Ensure Greater Accountability for Misuse of Force
These recommendations are measured and well-intentioned. But Rizer and Mooney overestimate their potential impact. That overestimation—which varies in degree as to each of the recommendations made—becomes more apparent upon a closer examination of what the available data and literature tell us we can expect from acting on them.
This response will begin by placing police use-of-force data in their proper context. Doing so casts doubt on Rizer and Mooney’s assertion that there exists a significant “police violence problem” in America. It will then assess Rizer and Mooney’s policy recommendations, offering specific critiques and highlighting competing analyses of the underlying issues.
I. Use-of-Force Statistics in Their Proper Context
A. The Statistics
Rizer and Mooney write that “Existing guardrails against excessive police use-of-force are far too weak,” which, they argue, is evidenced by America’s “police violence problem.” There is no question that, every year, there are many documented instances of excessive police violence which are individually problematic. However, to establish that there exists a “police violence problem,” one would need to demonstrate that such individual incidents are representative of a larger pattern. The data on police shootings and other uses of force weighs heavily against such a conclusion.
According to the Washington Post’s database, there were 992 fatal police shootings across the country in 2018.The vast majority of these shootings—91.6%—involved visibly armed suspects and were therefore quite likely justified. But fatal police shootings are only a subset of the total instances in which police applied deadly force with their firearms, given that many suspects survive their wounds or are not hit at all. To get a more complete picture of how many times police used deadly force in 2018, I analyzed a dataset of police shootings by officers working in the nation’s 50 largest departments maintained by VICE News.VICE’s dataset documents 3,936 police shootings in those departments since 2010, 32.6% of which were fatal. Based on that breakdown, let us assume that the fatal shootings documented by the Post in 2018 represented 32.6% of total shootings that year. Having so assumed, we can estimate that, in addition to the 992 fatal police shootings documented by the Post in 2018, there were another 2,051 non-fatal police shootings, for a total of 3,043 firearm discharges (including those in which no one was hit). That represents an average of more than eight shootings every day. That sounds like a lot, but it needs to be contextualized in light of the overall volume of police activity.
B. The Context
The Federal Bureau of Investigation’s Uniform Crime Reports document that in 2018 (the most recent available year), there were an estimated 686,665 full-time police officers working in the United States. That year, those officers made 10,310,960 arrests. Those arrests represent a small fraction of the total number of contacts police have with the public. The Bureau of Justice Statistics (BJS) reports that, in 2015, 53.5 million people in the United States had contact with the police; it stands to reason that some of those people had more than one such contact.
This data should frame our analysis. If we attribute each of the 3,043 estimated firearm discharges by police in 2018 to a unique officer, we can infer that, at most, 0.4% of police officers purposely discharged a firearm in 2018. If we assume that every shooting happened during the course of a separate arrest, we can infer that, at most, police applied deadly force with a firearm in 0.003% of arrests.
The case for the existence a national police violence problem doesn’t get much stronger when considering the data on non-deadly force. According to a BJS study covering the 10-year period between 2002-2011, 0.8% of people who had contact with the police reported being subjected to physical force. In 2018, a research team of doctors and a criminologist published a thorough study of police use of force in The Journal of Trauma and Acute Care Surgery, entitled “Injuries associated with police use-of-force.” The study analyzed over a million calls for service to three midsized police departments in Arizona, Louisiana, and North Carolina over a two-year period. Those calls resulted in more than 114,000 arrests. Physical force was used in 1 of every 128 of them, meaning that more than 99% of arrests were effected without any use of force. The study went on to find that, based on expert medical examinations of suspects’ medical records, 98% of suspects on whom police used physical force “sustained no or mild injury,” and 1.8% of suspects sustained moderate or severe injuries (only one suspect was fatally wounded by police gun fire during the study period).
Another important contextual consideration is that not all police uses of force are unjustified. Incidents of excessive force—force that goes beyond what a situation warrants—are a small subset of a small whole. While an exact assessment of what percentage of police uses of force are unjustifiable is hard to come by, there are some instructive scholarly estimates that have been published. One example: A BJS study of 2002 data on citizen complaints filed against officers in departments with more than 100 full-time sworn officers found that just 8% of force complaints for which there was a final disposition were sustained. Some may dismiss this as an illustration of a general unwillingness among police administrators to hold their own officers accountable. However, an even smaller percentage (6%) of complaints were sustained in jurisdictions with a Civilian Complaint Review Board (CCRB), which takes the disposition of those complaints out of the hands of the police themselves. The BJS study arrived at an estimated rate of excessive force of 1 incident per 200 full-time sworn officers. This is in line with the most recent available data out of New York City—home to the nation’s largest police department, which employs approximately 36,000 officers. In 2018, New York’s CCRB received 4,745 total complaints, and just 226 complaints were substantiated against 326 officers. 2,919 of the complaints contained allegations of excessive physical force, and only 73 of those allegations were substantiated. That year, the New York City Police Department (NYPD) made 245,392 arrests—89,685 felony and 155,707 misdemeanor arrests—to say nothing of other enforcement-related encounters such as pedestrian and traffic stops.
Because they are human, even the most competent and well-intentioned police officers will fall short of perfection—a standard to which they are often unfairly held. But there is a great deal of space between imperfection and deficiency. Police use force tens of thousands of times every year, and among those are cases of both carelessness and malevolence. But in the context of almost 700,000 officers making more than 10 million arrests and engaging in millions more stops and other interactions, can we infer an institutional police violence problem? Given the data outlined above, it would be difficult to answer this question in the affirmative.
II. Militarization and Police Violence: A Very Tenuous Relationship
A. There Is No Evidence That Militarization Causes Excessive Force
If the case that America has a police violence problem is weak, the connection between that problem and police militarization is weaker still. Rizer and Mooney argue, in part, that police attitudes have become militaristic. To their credit, they acknowledge that “[t]he hypothesized causal link between police militarization and excessive force is . . . far from conclusively established.” But this acknowledgement is understated: absent from Rizer and Mooney’s article are hard data outlining how and to what degree such attitudes have changed, let alone any evidence linking such changes to actual outcomes in police-citizen encounters.
To be clear, a diligent research effort would uncover some suggestive evidence. For example, one study, published in the Oxford Journal of Public Health last year, found that military veterans were significantly more likely to be involved in officer-involved shootings (OISs). However, that study did not shed light on the mechanisms that explain that relationship. One potential explanation could be that police officers with military experience are placed in units more likely to be involved in dangerous encounters.
On the other hand, there are other data points which cast doubt on the validity of the hypothesis that police militarization begets unjustified uses of force. For example, in making the case that police have become more militarized, Rizer and Mooney point to the creation of Special Weapons and Tactics (SWAT) units and “their increasing integration into everyday policing.” However, at least within the NYPD, Emergency Service Unit (ESU) officers (the department’s SWAT team members) seem to be extremely well-disciplined. The NYPD’s ESU commands did not record a single on-duty shooting in 2019. In 2018, ESU officers were involved in just one shooting. In 2017, the number was just two. In Chicago, SWAT teams filed just 26 (approximately 0.003%) of the department’s 10,068 Tactical Response Reports—paperwork filed whenever officers are involved in reportable uses of force such as a takedowns, punches, and firearm discharges—filed in 2017 and 2018. Windy City SWAT officers were involved in just four firearms discharges during that two-year period—4.5% of the department’s 88 reported discharges.
Rizer and Mooney argue further that “law enforcement has been increasingly militarized and its tactics more confrontational since at least the early 1970s.” The empirical support for this proposition is quite tenuous. Looking again at the NYPD, we see that even as the department has grown in size,OIS numbers have declined precipitously during the time frame identified as marking a rise in police militarization. In 1971, NYPD officers shot and injured 221 people; in 1972 they shot and injured 145. In 1990 that number was down to 72, and in 2016, it was just 23. In Chicago, police shot 523 civilians between 1974 and 1978—approximately 131 per year. Chicago police reported just 43 firearm discharges in 2018.
Among the other reasons to be skeptical of the asserted link between increasing militarization and police violence are a series of BJS reports on citizen perceptions of police use of force in 2005, 2008, and 2015. The percentage of those subjected to actual force who felt the force was excessive held essentially steady over the 13-year period. In fact, the percentage of citizens characterizing the force used against them as excessive declined slightly from 83% in 2005, to 80.6% in 2008, to 78% in 2015.
In 2006, the BJS compared the rate of citizen complaints about police use of force in the United States with the rate of such complaints in England & Wales. With the vast majority of police in the United Kingdom patrolling without a firearm, it would be difficult to argue that they suffer from the same militarization problem alleged to exist here in the U.S. Still, the BJS reported that “the overall rates of complaint per 100 officers in both countries are similar (7.5 force complaints per 100 officers in large U.S. local agencies, versus 7.2 oppressive behavior complaints per 100 officers in England and Wales).”
B. Limiting the 1033 Program Is Unlikely to Reduce Police Use of Force
Perhaps the strongest evidence offered in support of a causal connection between police militarization and violence concerns the federal government’s “1033 Program,” which Rizer and Mooney propose limiting. That program allows tribal, state, and local law enforcement agencies to obtain excess equipment from the U.S. Department of Defense through the Defense Logistics Agency. Rizer and Mooney rightly point out that since the very public display of military equipment by police during the 2014 riots in Ferguson, Missouri, this program has become the object of much scrutiny. Despite that scrutiny, the best attempts to assess the connection between the program and police use of force weigh against the recommendations to limit both an agency’s ability to acquire equipment through the program and the circumstances in which officers can use such equipment.
In prosecuting the case against the 1033 program, Rizer and Mooney rely heavily on a 2017 study which found “a positive and statistically significant relationship between 1033 transfers and fatalities from officer-involved shootings.” However, the authors of that study base their analysis on county-level data in just four states (Connecticut, Maine, Nevada, and New Hampshire) and acknowledge that their analysis is “relatively preliminary.”
On the other hand, three other empirical assessments have found the opposite. The first, cited but not expounded on by Rizer and Mooney, is a study by researchers at the University of Tennessee. That study uses a much larger sample of police agencies, and it finds that certain acquisitions made through the 1033 program “reduce citizen complaints,” as well as “assaults on and deaths of police officers.” Most relevantly, the study finds no effect on fatal police shootings. The same journal concurrently published a second study of the 1033 program’s effects that used slightly different methods and also found that the program had “no effect on the number of offenders killed.” The second study also found that the program showed a cost-effective crime reduction effect through the mechanism of deterrence. The authors make the case for a deterrent effect partly by noting that “two highly visible tools, gears and vehicles, have strong and sizable effects on all the types of crime,” which, they go on to explain, “is consistent with early studies by e.g., Bell (1982), which explore how police wearing military-style uniforms influences citizens’ perception of the police’s authority and legitimacy, and reinforces the notion that a main causal channel could be based on perceptual deterrence.”
The third study—a working paper authored by Olugbenga Ajilore, a senior economist at the liberal Center for American Progress—found “little evidence of a causal link between general military surplus acquisition and documented use-of-force incidents.” “In fact,” Ajilore says, “the acquisition of military vehicles leads to fewer use-of-force incidents.” If there is additional support for the conclusions reached by Rizer and Mooney as to the 1033 program, it can be found in Ajilore’s conclusion that the program does seem to increase use of force by SWAT teams—an increase that was, in the end, too small to change the overall effect on use of force.
III. The Practical Limits of De-escalation Tactics in OIS Contexts
According to a comprehensive field guide of best practices for law enforcement published by the Department of Justice last year, “Research has identified five attributes common to the clinical literature of de-escalation: communication, self-regulation, assessment, actions, and maintaining safety.” The guide defines those attributes as follows:
1. Communication encompasses specific verbal and non-verbal strategies to begin an effective dialogue with an individual and earn that individual’s trust and cooperation.
2. Self-regulation reflects skills and techniques used by individual service providers to manage their emotional or behavioral responses to an individual encounter. This includes techniques that they can use to provide the subject time and space to cool down.
3. Assessment is the task of collecting as much data about the person and situation as possible to make informed decisions about subsequent actions, including understanding when using force becomes imperative.
4. Actions refer to the behaviors and activities a service provider can engage in to reduce the likelihood and the severity of use of force.
5. Maintaining safety describes the paramount need of service providers to ensure their own welfare and public safety. Specific actions can reduce the likelihood that they will be injured if the person becomes violent or coercive methods of control are required.
While the guide seems to favor de-escalation training for law enforcement officers around the country, it acknowledges that “to date, there is still limited empirical literature examining the effects of de-escalation in law enforcement beyond C[risis] I[ntervention] T[raining].” There is indeed a paucity of empirical research on whether de-escalation techniques can be learned, internalized, and effectively deployed by police in the field.
However, there is a relatively substantial literature on the effectiveness of de-escalation in healthcare settings—the context in which comprehensive de-escalation strategies were developed before being marketed to law enforcement agencies. The development of that literature is not encouraging for those holding out hope that a broader commitment to employing de-escalation tactics in the field will, as Rizer and Mooney put it, “further reduce police use of force and violence.” A comprehensive systematic literature review published in the British Journal of Psychiatry offers little cause for optimism. Pertinently, given Rizer and Mooney’s focus on police attitudes, the reviewers concluded that “No study of moderate quality or above provided any evidence of attitudinal change impacting de-escalation performance or rates of violence and aggression,” and that “There was little evidence to suggest that de-escalation skills may be influenced through modification of staff attitudes.” That review did find “evidence that de-escalation trained wards increased staff risk of exposure to being involved in an aggressive incident when compared with control and restraint trained wards,” and it ultimately concluded that while “It is assumed that [de-escalation] training may improve staff’s ability to de-escalate violent and aggressive behavior, [t]here is currently limited evidence to suggest that this form of training has this desirable effect.”
Even less promising is the literature on the application of de-escalation techniques developed in healthcare settings to policing. A 2019 review of the literature on the effectiveness of police crisis intervention training (CIT) programs published in the Journal of the American Academy of Psychiatry and Law concluded that “There is little evidence in the peer-reviewed literature . . . that shows CIT’s benefits on objective measures of arrests, officer injury, citizen injury, or use of force.” Another comprehensive literature review—of empirical research on de-escalation training more broadly, not specifically focused on CIT—concluded that the “conclusions concerning the effectiveness of de‐escalation training” identified by the reviewers—which were both mixed and not particularly weighty—were “limited by the questionable quality of almost all evaluation research designs.”
Just as there is a paucity of evidence in support of de-escalation training, there is of course little evidence that an expansion of de-escalation would hurt. It is quite possible that this particular policy recommendation advanced by Rizer and Mooney could have some moderate, positive effects on at least some relevant outcomes. However, given the extremely short and chaotic time frames within which so many police shootings occur, de-escalation may not even be a viable option for those trained in such tactics in a substantial majority of the situations likely to receive significant media attention and, as a result, impact public perceptions. Moreover, there may also exist a subset of police shootings in which even an ideal deployment of de-escalation tactics fails to avoid the need for deadly force. An empirical assessment of how many police shootings might be avoided by using de-escalation techniques (assuming, for the purposes of argument, their effectiveness) could shed more light on the likely limits of the impact better and more-intense de-escalation training might have on police use-of-force.
IV. Litigation and Police Accountability
While police violence is relatively rare and usually justified, those who violate the public’s trust by abusing or exceeding their power ought to be held accountable, and systematically failing to hold police to account for such violations would understandably strain the relationship between a centuries-old institution and the society to which it has become an essential source of protection. On this, I am in full agreement with Rizer and Mooney.
But on the question of qualified immunity, the controversial doctrine’s actual role in police litigation is far from clear. The legal foundation upon which modern qualified immunity doctrine has been built is controversial, and there are strong arguments on both sides; I will leave it to others to fight it out. Instead, I’ll focus on the claim that qualified immunity has “made it extremely difficult to hold law enforcement officers who use excessive force or otherwise violate people’s constitutional rights accountable.” While one can surely find egregious examples of judicial deference to clear police misconduct on qualified immunity grounds, the aggregate role the doctrine plays in modern police litigation is overstated.
Consider first an empirical assessment of qualified immunity by Joanna C. Schwartz, published in a 2017 issue of the Yale Law Journal. Her study analyzed 1,183 cases filed against state and local law enforcement defendants under 42 U.S.C. § 1983, which creates a right of action for constitutional violations committed by state actors. Qualified immunity could be raised as a defense in 979 (82.8%) of those cases, yet just 38 (3.9%) of those cases resulted in dismissals or grants of summary judgement on qualified immunity grounds.
Consider also the data out of New York City, compiled in a database launched by the Legal Aid Society last year. The database contains data on federal lawsuits filed against New York City police in the Eastern and Southern Districts of New York between January 2015 and June 2018. The database contains 2,387 such lawsuits. Filtering the entries by case disposition reveals just 74 cases resolved in favor of the defendants. Even if all 74 were decided on qualified immunity grounds, that would mean that the doctrine proved to be an effective bar to plaintiffs’ recovery in just 3.1% of cases. Here again, we must temper our expectations of just how much life would change in a world with no qualified immunity.
While it has no bearing on the validity of the criticisms of the federal judiciary’s development of the qualified immunity doctrine, it’s worth mentioning that the Supreme Court has done quite a bit to expand police accountability through its constitutional jurisprudence. The Supreme Court’s 1989 decision in City of Canton, Ohio v. Harris held that “under certain circumstances,” municipal liability could attach under § 1983 when a police department failed to properly train its employees. This particular example is worth noting because Rizer and Mooney note in their wonderful case study of how departmental policies evolved alongside the expansion and professionalization of policing in Miami-Dade that “police use-of-force policy in the 1980s remained largely similar [particularly in its simplicity] to that of the 1960s,” but by the mid-1990s, “the Miami-Dade Police Department manual dedicated ten pages to an in-depth articulation of the policy, rules, and standard operating procedures around use-of-force.” Given the timing of that change in relation to the Court’s decision in City of Canton, it’s possible, and perhaps even probable, that the Court’s decision shaped the improvement of previously inadequate policy surrounding the use of deadly force. The Court’s police training jurisprudence also led the National Institute of Justice, in May of 1990, to urge police departments to develop more robust pursuit policies, specifically citing City of Canton.
The debate about the proper scope of police power and whether the hundreds of thousands of law enforcement officers serving American communities operate within that scope is important and should be ongoing. Police in the United States wield an enormous amount of power, which, as the old adage goes, comes with a great many responsibilities. Those of us operating in the public policy space can play an integral role in helping to ensure that those responsibilities are carried out, and that the ideals they embody are lived up to. Through their work, Rizer and Mooney admirably pursue those noble goals. Nevertheless, their article suffers from one key flaw in that both its diagnosis of a police violence problem, and its prescriptions to address that problem go beyond what the available data support.
Viewed in their proper context, the data on police use of force are incongruous with the characterization of police as having gone off the proverbial rails in the violence department. To the contrary, those data reveal a great deal of both professionalism and restraint. Moreover, there is essentially no empirical support for the asserted connection between police militarization (attitudinal or otherwise) and use of force. There may be some anecdotal evidence for the rise of a more militarized police subculture since the 1970s; but that rise corresponds with a sharp decrease in overall police use of force. And the best empirical assessments show that the acquisition of military equipment has not made police more likely to abuse those with whom they interact.
Though there may not be enough police violence to support the existence of an institutional problem, there is always room for improvement through training. Police departments and professional associations should continue to identify and develop training programs that will improve both enforcement outcomes and community relations. However, there is little evidence that expanding de-escalation training can produce meaningful, if any, reductions in uses of force (which can fray police-community relations). This does not mean police departments should necessarily abandon such training initiatives; but it should temper our expectations of how big an impact such programs can have. The same goes for limiting or eliminating qualified immunity for police officers—the benefits of which seem overstated by critics in light of how rarely it functions as a bar to recovery.
All that said, Rizer and Mooney do identify a real problem in the deterioration of police-community relations and the legitimacy of police in the eyes of the public. Furthermore, they are correct to connect that problem to controversial uses of force by police. Reducing unnecessary and excessive uses of force—which we’ve seen throughout the U.S. for decades—is undoubtedly part of the solution. But because the level of unjustifiable police use of force has been so overstated, even meaningful progress on that front is bound to underwhelm. Perhaps one solution to that problem is to reorient the public debate about policing around a more realistic, properly contextualized assessment of the available data.
This piece originally appeared at the The Federalist Society Review
Rafael A. Mangual is a fellow and deputy director for legal policy at the Manhattan Institute and a contributing editor of City Journal. This piece was adapted from City Journal. He is the author of the recent Manhattan Institute issue brief, Reforming New York’s Bail Reform: A Public Safety-Minded Proposal. Follow him on Twitter here.
Photo by Chalabala/iStock