Jun 16 2020

The Stats on Police Killings

During the current national attention being paid to police practice and inequities of police killing of African Americans it is important to put the data that we have into as much context as possible, in order to understand the phenomenon and make sure that our efforts to improve the situation are properly targeted. Unfortunately, the data are complex, which makes it easy to see what one wants to see. I will try to break down the research as objectively as I can, although it is likely that perception of bias will also depend on perspective.

We can start with the most basic numbers:

Risk is highest for black men, who (at current levels of risk) face about a 1 in 1,000 chance of being killed by police over the life course. The average lifetime odds of being killed by police are about 1 in 2,000 for men and about 1 in 33,000 for women. Risk peaks between the ages of 20 y and 35 y for all groups. For young men of color, police use of force is among the leading causes of death.

This same data, from an August 2019 study, also shows that the overall risk of death at the hands of the police is about 2.5 times greater for black Americans than white Americans. This is an often cited figure, and it is salient, but there are additional layers here. Let’s break down the police use of force into non-lethal force, lethal force against armed citizens, and lethal force against unarmed citizens. Some studies focus on “shootings” rather than all uses of lethal force, mostly because that is how databases are often set up, but that can also miss important cases.

There is strong evidence that police use of non-lethal force is greater against black individuals than white. This holds up across a broad range of activities, such as drawing a weapon, using a baton, handcuffing, and using a taser. This difference is not explained by factors other than race. (The study authors controlled for other causes, and race emerged as an independent variable predicting use of force.)  This is where police education is likely to be most effective, because it does seem to be a factor of police behavior.

But this same data also shows that when you get to police shooting of armed suspects, the racial difference can be explained entirely an an increased risk of being in a violent encounter with the police and factors other than race. Race turns out not to be an independent factor. However, there are extremely important caveats to this fact. First, the data with respect to killing unarmed suspects (which also may not involve shooting) is not clear. This is because the absolute numbers are relatively small and therefore it is difficult to achieve statistical significance. But, we cannot rule out independent racial disparity in this category, and the fact that police use of force is clearly racially biased by other measures legitimately raises concerns that it extends to deadly force used against unarmed suspects.

The researchers are also careful to point out that they are looking at group-level data, and this cannot explain an individual case. They are also focusing on specific cities, and it is not clear how generalizable the results are. So you cannot use this data to argue that any individual case was or was not racially motivated. Also, an important caveat, researchers generally agree that we need more and better data. It is interesting that the more data we have the more clear it is that racial bias exists, which suggests the failure to find a statistically significant racial factor may be an artifact of too little data.

Add to all this another salient fact – in the US there are many more police killings of suspects than in other industrialized nations, by orders of magnitude. Basically, the number of killings in the US per day on average is roughly equal to the number of police killings per year in the UK, for example. So there are at least two factors we must confront – the total magnitude of police use of force (lethal and non-lethal) in the US, and racial disparities in the use of that force.

This is all happening on a backdrop of historical and current systemic racism. High levels of use of police force can magnify racial inequities in socioeconomic status, housing, how neighborhoods are treated, and other examples of racial inequality.  In this context, social inequality transfers to policing inequality, even if the police force itself did not display direct bias. And, to be clear, the data we have clearly shows police use of force is racially biased when it comes to non-lethal force.

But if we take this data at face value, what does this mean for instituting reforms to mitigate racial disparities in policing? Let’s look at some of the more common proposals:

  • Ban choke holds
  • Ban no-knock warrants
  • Rebalance qualified immunity
  • National registry of police offenders (so that offending officer can’t just move to another department)
  • Emphasize deescalation in police training
  • Police use of force as a last resort
  • Police must report every time they draw their weapon or point a weapon at a civilian
  • Mandated use of body cameras
  • Better data collection across the board for all forms of police use of force

You will notice that none of these are directly related to racial bias. They are proposals intended to hold police more accountable and reduce the overall use of police force. But there are also proposals that do directly address the racial angle, to make efforts that the demographics of a police department reflect the demographics of the community they police, for example. And yes, there are proposals for police education regarding the potential effects of implicit bias in policing. We have this in medicine – it seems this should be a standard part of any profession who wields power and authority over the public.

Further we need to note that the problems of police use of force in the US relate to deep problems in society, including socioeconomic disparities, that go beyond policing. We also have to confront the gun violence issue in this country. The vast majority of police shootings are against armed suspects, so this raises the stakes for everyone. The more threatened police are in the course of doing their job, the more likely they are to escalate the use of force to defend themselves.

Finally, there is a big discussion around “defunding” the police, which is an unfortunately misleading phrase. The real discussion is about restructuring the police. Over time cities have generally dumped on the police every social problem that was otherwise underfunded, from mental illness to substance abuse. We don’t necessarily need armed officers to deal with every social disturbance or non-criminal issue. Other professionals may be better trained and prepared to take these issues off the plates of our police officers, so that they can dedicate their resources to actual crime. The current structure wasn’t really planned. It evolved organically, and partly out of neglect of certain social issues. This is a good opportunity to rethink how optimally to address a range of issues.

The good news is, unlike many complex social issues, there does seem to be a clear path forward here. There are many common sense recommendations for specific legislative change, many of which should be able to garner bipartisan support. The reduction in the overall use of force by police, especially when you consider other countries, seems like an obvious goal that is likely to be a net win for everyone. Racial disparities will necessarily shrink in absolute numbers along with the overall use of force. And there are common-sense steps to take to directly reduce those racial disparities further.

Of course correcting the underlying social disparities is a much longer project, but policing should not magnify those differences. If anything it should mitigate them.

I am also speaking as a fellow professional. Certain professions are given a contract with society – granted special privileges and authority in exchange for a transparent and regulated system of ethics, quality control, and professionalism. None of this discussion is meant to imply that the majority of police are not good people with high levels of professionalism who genuinely want to serve their community. The issue is with systemic problem within policing that need to be fixed. (Just as there are identifiable systemic problems within the medical profession – once identified we don’t deny them or decry the implications that doctors are bad people – we look for ways to fix them.)

This will be good for policing. We want to set up our police to succeed, not to fail, by training them properly, and not asking them to do things that are more appropriate for other professionals. This also requires accountability, and effective ways of retraining or removing those who break the contract. I hope the energy of this moment does not peter out before specific legislative improvements are made.


Note to my readers: I know this is a more political issue than I typically write about, but it was my intention to focus on the research and issues of professionalism. But topics like these tend to provoke high levels of emotion and issues of personal identity and ideology. I am asking to make a special effort to keep the comments focused as well, and remember the principle of charity. Thanks.

No responses yet