Variation in fatal police shootings

Explaining all the data, not just some of it

Since 2015, the Washington Post has been carefully tracking fatal police shootings in the United States. Their database includes over 8,000 cases. If you ignore those where the race and/or sex of the victim are unknown, the demographic breakdown is as follows:

• White males = 3,106
• Black males = 1,707
• Asian males = 121
• Native American males = 99
• White females = 192
• Black females = 58
• Asian females = 8
• Native American females = 6

There’s obviously a huge disparity between the sexes, which is evident in all four racial groups. But aside from sex, what explains the variation in these data? (I’m ignoring Hispanics for reasons that will become apparent later.)

One possibility is population. Whites are still the largest racial group, and native Americans the smallest, so perhaps that explains why they’re at opposite ends of the distribution (when controlling for sex).

The chart below plots fatal police shootings against population for the eight demographic groups. Population is the value for 2015. Both variables were logged to reduce skewness.

Larger groups get fatally shot more often.

There’s a fairly strong correlation between the two variables (r = .64), though it’s not statistically significant due to the tiny sample size. So while population can explain some of the variation among the eight groups, it leaves much of the variation unexplained.

Unsurprisingly, all four female groups are below the line, confirming that they’re fatally shot less often than you’d expect based on their numbers in the population. Asian females are particularly far below the line.

By contrast, black males are particularly far above the line, indicating that they’re fatally shot by police much more often than you’d expect based on their numbers in the population.

How can we explain these patterns? One possibility – the one favoured by the mainstream media and the “liberal establishment” – is that the police are biased against certain groups. Yet this theory immediately runs into problems. While it makes sense that the police would be biased against black males, why would they be biased in favour of Asian females?

Could there be another explanation – one that explains all the data, not just some of it? Yes: differences in violent criminality.

The chart below plots fatal police shootings against murders for the eight demographic groups. Murders is the total number of murders committed by each group between 2015 and 2020. Which brings me to the reason Hispanics were not included: the FBI treats Hispanic ethnicity as a separate variable that may overlap with race.

Groups that commit more murders get fatally shot more often.

There’s an extremely strong correlation between the two variables (r = .94), and despite the tiny sample size it has a p-value of 0.005. Murders can explain almost all the variation in fatal police shootings.

Now Asian females are exactly on the line, while black males are slightly below it – which means they’re fatally shot marginally less often than you expect based on the number of murders they commit. Interestingly, the group furthest from the line is black females – they’re fatally shot considerably less often than you’d expect.

Black and white males are positioned close to one another on the chart. Don’t black males have a much higher murder rate? Yes, but remember that the chart plots absolute numbers on both axes. The black-white difference in the absolute number of murders is much smaller than the difference in murder rates because there are many more white males in the population. (If you plot rates, the chart looks pretty much the same.)

By what mechanism do murders explain demographic variation in fatal police shootings? It’s not that the police go out and shoot different groups in proportion to how much they murder, as some form of punishment. Rather, murders serves as a proxy for the kind of behaviour that leads people to get shot by police (such as, say, violently resisting arrest).

Asian females don’t commit many murders, and they also don’t get into situations where police officers end up shooting them. For black males, it’s the reverse. Looking at the chart above, there is no need to invoke any kind of police bias – both black males and black females are below the line.

This issue has already been covered extensively by people like Steve Sailer and Heather Mac Donald (though so far as I’m aware, no one’s plotted fatal police shootings against murders at the demographic group-level before). I myself wrote an article pointing out that “systemic police racism” fits the definition of a conspiracy theory.

So why revisit the subject? Because the arguments bear repeating.

In 2020, riots following the death of George Floyd – riots that were encouraged by the media – got 19 people killed, caused $1–2 billion in property damage, and spurred a sustained rise in the murder rate. These riots were based on a lie: the lie that racist police officers go around ruthlessly killing black people.

Image: Chris Yarzab, LAPD officers at the Staples center, 2010

---

Thanks for reading. If you found this newsletter useful, please share it with your friends. And please consider subscribing if you haven’t done so already.