Sunday, October 18, 2015

Racial bias in police killings

There's a very interesting debate going on between Harvard's Sendhil Mullainathan and Barnard's Rajiv Sethi, on racial bias in police shootings.

Mullainathan looked at the data and found that police likelihood of shooting a black person in a given encounter was no higher (or barely higher) than their likelihood of shooting a white person. He concluded that while police may be racially biased in terms of how often they go after black people, they're not very biased in terms of how likely they are to shoot the people they go after.

Sethi said: Not so fast. Mullainathan just looked at the killing rate conditional on an encounter, but Sethi wants to also condition on the type of encounter, which probably differs on average across races. Police might be more likely to misinterpret nonthreatening encounters with black people as threatening, and therefore be more likely to pull the trigger. Sethi's hypothesis says that police basically give black people a hard time, initiating disproportionately many encounters for innocuous things (racial profiling), and then pull out the gun when they interpret those innocuous encounters as threatening situations.

Sethi might be right. But it seems to imply some odd psychology on the part of cops. It implies that cops routinely go after the people who scare them the most. It implies that cops routinely and preferentially launch themselves into what they expect to be threatening situations.

That would be really weird. It could be true, of course - maybe cops have an incredibly strong sense of courage and duty, forcing them to bravely stand up to the (not actually) threatening, dangerous black people time and time again. Or maybe there is some institutional pressure from higher-ups forcing cops to do racial profiling. These explanations were both suggested by a sociologist friend to whom I posed this question. But they feel like quite a stretch. Cops get a lot of discretion in who they stop. And I doubt they are usually driven by an almost pathological sense of duty and courage.

Let me offer an explanation I see as more likely: Cops often tend to shoot (or otherwise brutalize) people not out of fear, but out of wrath.

My hypothesis goes like this. Cops pull out their guns and their nightsticks when they see suspects as having challenged their authority. They are determined to maintain power and control at all costs (i.e., South Park nailed it). Black people are more commonly seen as challenging cops' authority, probably because a lot of black people grew up in a state of relative anarchy and therefore lack other people's conditioned response of instant meek submission to police.

This seems to be exactly what happened with Eric Garner. He wasn't threatening at all; he's obviously a big teddy bear, he doesn't have any weapon, and he wasn't making any move to attack anyone. But he's an insubordinate teddy bear, who thinks he can reason his way out of an unfair arrest. So the cops grab him and choke him to death.

It also explains why so many suspects get shot in the back. For example, Walter Scott. A man who's running away is not a threat. He is not a source of fear. He is, however, flouting authority. Same with Samuel Dubose. Type "police shoot black man" in Google, and "police shoot black man in the back" is one of the first results that come up.

Here the police shoot a black guy in a wheelchair.

This psychologically plausible hypothesis is also parsimonious, because it allows police racism to explain both racial profiling and excess unjustified brutalization of black people. It also implies that in areas with entrenched racial conflict - say, the South - white police will be more likely to kill black people, because they view blacks as socially subordinate (hence any backtalk or resistance will be seen as more unacceptable if it comes from a black person than if it comes from a white person). So that would be an interesting test.

Anyway, to sum up, I think Sethi is right to say that equal rates of shooting per encounter don't imply that police are unbiased. But I think his explanation - police fear of blacks - isn't as likely as a wrath-based explanation.

Of course, if you drag a cop in front of a judge (or the media) and ask "Why did you shoot that guy?", of course they're going to say they were afraid. Fear generates sympathy. But color me skeptical.


  1. Or those they are most suspicious of, and cops are selected and trained for assertiveness.

  2. Anonymous2:55 PM

    Why are they angry at black people? Or does that angry response happens when whites are insubordinate?
    Are whites less insubordinate than blacks?
    How are they not more likely to shoot blacks and still kill more of them?
    If you include non-deadly encounters what is the percentage of cases where violence escalates versus those in which it deescalates involving white vs those involving blacks?

    1. Yes, my hypothesis is that whites are less insubordinate than blacks, on average. Whites are more likely grow up thinking that the police can be called on for dispute resolution - i.e. that the system is on their side. Also, weak institutions in heavily black areas (the "state of relative anarchy") probably leaves black people less conditioned to immediately submit to authority.

    2. There's been some studies on this where the brain needs different frameworks. To survive in high violence poverty areas one has to have habits showing one is not a victim. While that's successful in some settings in other settings (interactions with politics, interactions at school) that can be counterproductive. While I'm sure you look askance at the podcast Freakanomics had an interesting show about this and attempts to solve it in Chicago and elsewhere with cognitive behavior therapy. (See here and here)

    3. I like this fake discussion about "disproportional" victimization of blacks. It's only possible by not looking at the offending rates.

      Given that young black men (20-24) are roughly 40 times more likely to be involved in homicide (and by extension in violent crime) than corresponding whites, I find it surprising that cops don't shoot them more. That actually speaks to excellent training.

  3. I think you get close but miss it...

    I'd argue cops see themselves as cross between bureaucrat and animal trainers. You say:

    "Black people are more commonly seen as challenging cops' authority, probably because a lot of black people grew up in a state of relative anarchy and therefore lack other people's conditioned response of instant meek submission to police."

    I'm not sure I even agree, but...

    Most cops I know have a very simple take: "do what you are told, so I can get thru with my job fast." (be a good dog, so I can be good bureaucrat).

    Meek submission is not what people unafraid of police do, they say "yes officer, no officer, my apologies, is there anyway sir..."

    Interaction with cops for upper class folks with far more status and power, they never feel like they are being a good dog... isn't submission to the cop, it's support of the system that ensures your big house and happy family pay lower security costs.

    When I warn cop friends that they are about to be in the Customer Service business and must be kinder with lower class because of cameras and tech..., they don't push back on "that reduces my status" they push back with "this is going to make you less safe."

    1. Anonymous7:12 AM

      They should push back with "You're a creepy doofus".

    2. totes.

      training wheels, you're why I support (but discount) anon posters.


      to make good public policy, wonks must deal w reality of hegemony:

      What strategy do cops use when they want more "public" support? Let everybody see what happens when they take a breather. Who are they negotiating with?

      There are Carmen like cops, but the vast bulk of them IMO are bureaucrats who get paid, like all bureaucrats, to make sure the top half's (hegemony) normative reality is enforced.

      If you don't start there, it's very hard to make policy, bc it'll sound like you are making policy and actually you are just complaining about hegemony.

      Which is fine, but good lord, stop calling yourself a wonk.

    3. I agree with a lot of this.

      Also, black people have more erroneous interactions with police. If you get stopped once you probably think there is a good reason. If you're stopped constantly for dubious reasons, or know people who are stopped for dubious reason, you are more likely become belligerent and question subsequent stops.

  4. Anonymous4:29 PM

    Interesting.... I suspect that the post-9/11 excess of worship of authority in general has massively increased the sense of entitlement and expectations of deferenceof the police (and the liltary). As an affluent white male even I do not want to have anything to do the police under just about any circumstances.

    When I was much younger and living in the UK a much older relative said frequently that the police were totally corrupt, but at that time the UK was pretty smug about their wonderful and honest police force. Fast forward to Guildford pub bombings etc, and eyes were open over the following years. Trust lost cannot be easily regained. It will be interesting to see if the erosion of trust and confidence in the police in the US grows and is sustained, especially by the younger generation. I believe this would be a good thing, as excessive deference to authority to deleterious enemy to a robust democracy.

  5. R.U.Batty6:05 PM

    "This psychologically plausible hypothesis is also parsimonious, because it allows police racism to explain both racial profiling and excess unjustified brutalization of black people."

    I tend to think that black people have more violent interactions with police because, for some reason they have significantly more violent lives than americans of other backgrounds.

    For instance, of police killed by civillians, 50% are killed by black civillians. This sounds really disproportionate when you consider that black people are about 12% of the population. In fact it would seem indicative of a longstanding racial conflict.

    But then you also need to consider the bigger picture: black people commit about 50% of murders in the general population, too, so why should killings of police officers be any different? In light of this the fact that 30% of victims of police killings are black should strike us as surprisinly low,not surprisinly high.

  6. Andrew Greschner7:41 PM

    You don't need to believe cops believe black people are more likely to do something threatening, you just need cops to believe black people are more likely to do something illicit. Interpreting innocuous behaviour as sinister behaviour explains both relatively high levels of interactions for innocuous things, and why the innocuous interactions turn violent when an innocuous action is misinterpreted in a sinister way.

  7. It's worth noting that as tragic as the killings of Garner/Scott/DuBose are, very few *unarmed* African Americans are killed by police.

    28 in the last year. But if you look at their stories, the police had reason to fear their lives in a few of these stories as well.

    I believe that police brutality against African Americans is a problem, but I think we're exaggerating the occurrence of these especially tragic incidents.

  8. Wrath is irrelevant. This is just the vector. What drives that behavior is lack of accountability. CoPS know that there's little consequence for killing civilians. That's why they kill so many.

    As to racial prejudice, there is little evidence for it. Considering the extremely high crime rates of young black men, it seems to me that the number of young black men killed by cops is disproportionately low.

    The big problem is cops killing so many people: 10% of all homicide victims. This is horrific.

  9. "Black people are more commonly seen as challenging cops' authority, probably because a lot of black people grew up in a state of relative anarchy and therefore lack other people's conditioned response of instant meek submission to police."

    I'm not quite sure why you make this assumption. It seems to me at least as likely that cops more readily perceive black people to be insubordinate. That is, had a white person acted just as Eric Garner had, police would have been much less likely to perceive it as insubordinate, and thus less likely to respond with violent force.

  10. Have you listened to Heather MacDonald's recent testimony before Congress? Do you read her? She seems way too important, and too sensible, to ignore.

  11. BTW, physics envy I think refers to something Paul Samuelson in particular was guilty of: the wish to be seen to be as smart as top theoretical physicists, which led him to over-mathematize the subject and neglect ordinary language. For proof near the end of his life he compared himself to Euler, Gauss, etc. He had a terrible inferiority complex. And his influence on the profession has been both enormous and damaging, in my opinion.