9

The incidence of people dying due to use of force by the police is a high profile issue at the moment. Figures about the distribution of these deaths, such as found in this paper, can be compared to population statistics to demonstrate how different demographics are affected at different rates to those of the general population.

We can also compare the demographics of the officers involved in these deaths (table 6) to the demographics of the police force as a whole, and see the features that are overrepresented. For example 67% of police are white but 84.3% of officers involved with the deaths are, indicating a hazard ratio of 2. However the largest effect I have found is that of sex, in that males make up 84.9% of police but account for 97.4% of deaths, indicating a hazard ratio of 5.8.

Does this represent the true risk, in that your chance of dying in an encounter with the police is nearly 6 times greater if that police officer is male compared to female, or is there another explanation? For example, a policy of assigning female officers to lower risk areas or roles would explain this observation.

14
  • 4
    This seems like a question that is intended to push a point. Underrepresentation of women could be a dangerous thing, but the question seems framed to suggest that police brutality and police racism are, at most, minor problems in comparison.
    – Obie 2.0
    May 2, 2021 at 18:46
  • 2
    I’m voting to close this question because this question belongs on Skeptics. May 3, 2021 at 14:24
  • 1
    If my question was "are these numbers correct" then sceptics would be better. My question is more "are there policies that cause this, rather than men being more likely to kill people in any given situation", which seems to be political to me.
    – Dave
    May 4, 2021 at 9:51
  • 1
    I'd note that the number of police killings is quite low (in the low hundreds) per year so you'd expect significant variation from averages simply due to random variation - probably not enough to explain the disparity but probably enough to make the true magnitude of the disparity after considering potential random variation very uncertain.
    – ohwilleke
    Jan 30, 2023 at 17:33
  • 1
    I agree with @ohwilleke is that for crime reporting statistics, the number of police related deaths is two low for the numbers to be indicative of a statistical trend, let alone if the numbers in your data set control for the subset of police involve deaths of suspects that were deemed not justified by the agency's "Use of Force" policy and relevant training.
    – hszmv
    Jan 30, 2023 at 17:47

2 Answers 2

9

Does this represent the true risk, in that your chance of dying in an encounter with the police is nearly 6 times greater if that police officer is male compared to female, or is there another explanation?

Female officers are vastly underrepresented in SWAT teams and other special teams designated to go after the very highest risk criminals (I am unable to find any actual numbers for this, but searching 'female SWAT officer' reveals a number of articles describing such an officer as the only female SWAT officer in her department).

Thus, the correlation is true, but the implied causation is backwards. If you are a barricaded suspect, have hostages, have an outstanding warrant for murder, etc. the officers sent to disarm you and bring you in will almost certainly be male. Given that you allowed things to escalate to that point without peacefully surrendering, the odds that when the officers make their move, that you'll represent a mortal threat and thus get shot is high.

The implied causation in your question is not addressed in the paper you linked, as it makes no attempt to account for victim behavior nor circumstances (at least as it relates to its discussion of the demographics of officers). However, there is an interesting tidbit that addresses a different point in your question:

Ridgeway compared shooting and non-shooting officers at the same scene (using data from 106 officer-involved shootings in New York City) and found that black officers...were more likely to shoot

Thus, despite white officers without normalization having a hazard ratio of 2 (per your question), black officers are more likely to shoot when on the same scene. Looking at Ridgeway we can see that they did look at the sex of the officer. While they found an odds ratio of 2.1, with a p-value of 0.29, it is very far from statistically significant. Given the low likelihood of a female officer at a shooting scene, it would almost certainly take a far larger sample than Ridgeway looked at to tease out the actual difference in risk.

Even Ridgeway has the confounding factor of victim behavior. Is there perhaps still a subconscious chivalrous behavior that reduces the likelihood of victims to have attempted to shoot at female officers vs their male colleagues?

Thus, all else being equal, are you six times more likely to be shot by a male officer than a female officer? Almost certainly not. Is there a somewhat higher risk? Very difficult to tell and I would be dubious of any study that claimed it had a definitive answer.

7
  • There are also police officers that aren't normally doing patrols where they would be first responders and more likely to be the ones in a high risk situation than, for example, a detective. We'd need to control for gender representation in patrol units as well. Overall, we'd also need to control for the relative danger of different precincts - if hypothetically police officers are less likely to be women in high risk precincts, that would also skew the statistics. Jan 29, 2023 at 0:46
  • 2
    @IllusiveBrian The clever thing about the design of the Ridgeway study is that it compares who actually took the shot versus who was at the scene of the shooting. Thus, lower risk precincts, detectives, desk officers, etc. shouldn't impact the result. If women happen to be underrepresented in high risk precincts, they'll also be underrepresented in being at the scene of a shooting, thus the ratio is maintained. There are still possible confounding issues as far as victims possibly being more aggressive toward male officers, or female officers being posted to less dangerous parts of the scene.
    – LambdaCalc
    Jan 29, 2023 at 5:26
  • Or indeed male officers being on the scene earlier or for longer? (And of course there's a difference between officers doing crowd control and those storming a building.)
    – Stuart F
    Jan 29, 2023 at 14:27
  • I suspect that the proportion of people killed by police or other specialized officers who are killed by SWAT teams is rather low, judging from reading news reports about officer involved shootings. Traffic stops, being the nearest officer on patrol to respond, and routine arrest warrant execution and/or things like routine evictions make up a pretty big share of all police shootings. Likewise, detectives usually come in once all the excitement is over and try to figure out what happened.
    – ohwilleke
    Jan 30, 2023 at 23:29
  • @LambdaCalc Another confounding question is who was responsible for actually dealing with hostiles. So for example if SWAT is called in to a hostage situation there likely are already cops on the scene. however, the SWAT team is the one expected to deal with the hostage takers, other officers may be 'on the scene' but it's reasonable to suspect they wouldn't be the one making a lethal shot, it's not really why they are there. Thus there could still be a skew in sex of people who would be seen as responsible for using lethal means on scene vs those on scene who aren't expected to take a shot
    – dsollen
    Jan 31, 2023 at 16:07
1

I see no reason to doubt the results of this study, though I'll admit I haven't done due diligence and critically examined their methodology (subtle methodological errors are the main source of misinformation in academic journal papers...). It also conforms to population norms, in which men are far more likely to commit violent crimes than women, by factors ranging from 2:1 to 4:1. See FBI stats and this law review.

I suspect the somewhat larger '6 times' factor in policing is due to:

  • A general tendency among female officers to communicate and de-escalate in tense situations, where male officers first tend to try to establish dominance and control
  • The comparatively low numbers of female officers, and the relative newness of women in policing, which speaks to a more idealism among female officers than male officers
16
  • 4
    “I suspect the somewhat larger ‘6 times’ factor in policing is due to:” What makes you suspect this? What evidence do you have to back this up? Right now those two bullet points look somewhat sexist and also have zero proof. May 2, 2021 at 21:51
  • 1
    Well one could argue that not many women want to become police officers in the first place, and the few that do decide to take on the job are more motivated to act according to what they think an ideal police officer should act.
    – Ray
    May 3, 2021 at 2:08
  • 9
    Or maybe female police officers simply don't deal with extremely dangerous situations as often as male officers. Boys are often taught to treat girls more softly, so it's possible female police officers get "assigned" less dangerous tasks in general. See tandfonline.com/doi/abs/10.1080/15614263.2018.1468258
    – Ray
    May 3, 2021 at 2:41
  • 1
    @TedWrigley I somewhat disagree with 1st point (as of now). Also, you were already required to back up what you say in answers (see this and also this) May 3, 2021 at 12:34
  • 1
    @TedWrigley here’s some evidence that the second point is (kinda? You said “relative newness” which can be interpreted many different ways) false (this site says women have been in the police force for 131-ish years, and this site says the US police force has been around for 230 years, I wouldn’t call that relative newness. As for the first bullet point, as far as I can tell, you’re correct. I still believe you should cite sources, but I’ll stop arguing that for now :) May 3, 2021 at 14:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .