Gosh, here it is again.
Now, I have no idea whether these numbers are in any way accurate, but let's take it as true, and explore why it's stupid (and probably means the opposite of what OP thinks).
Causal inference is hard, but ignorance sure is easy.
Thread.
Now, I have no idea whether these numbers are in any way accurate, but let's take it as true, and explore why it's stupid (and probably means the opposite of what OP thinks).
Causal inference is hard, but ignorance sure is easy.
Thread.
Let's start with a simple one, which is the thing I discussed a few days ago, here:
When you condition on "arrest," you find little differences.
That's important, because the largest portion of the racism part is before/at the point of arrest.
When you condition on "arrest," you find little differences.
That's important, because the largest portion of the racism part is before/at the point of arrest.
(it is right here where I wish I had used a black square instead of red for shielding OP's identity).
If Black folks are disproportionately arrested, but not disproportionately killed after arrest, that wouldn't show up in the OP's stat.
But it's so much crazier than that!
If Black folks are disproportionately arrested, but not disproportionately killed after arrest, that wouldn't show up in the OP's stat.
But it's so much crazier than that!
That, on its own, would be enough to explain how you end up with proportionate post-arrest death rates but vastly disproportionate overall rates of death via police.
But it doesn't explain why the Black part is a little lower even.
And this is the part where things get wild.
But it doesn't explain why the Black part is a little lower even.
And this is the part where things get wild.
Up until this point, in the back of our heads, we may have been making a (wrong) assumption, whether we knew it or note: that the KINDS of arrests are roughly equal between Black and white folks.
AND WOW is that a bad assumption.
AND WOW is that a bad assumption.
Lots of ways to think about this one, but if you're in the medical field, a good way to start is "confounding by indication."
One thing that appears universally true is that Black folks are vastly more likely to get arrested for minor infractions, through all kinds of mechanisms
One thing that appears universally true is that Black folks are vastly more likely to get arrested for minor infractions, through all kinds of mechanisms
You could think of this as a threshold. For me, a white dude, to get arrested, I'd probably have to have done something pretty bad.
There's a good chance that I, being white, got away with at least one thing a Black person would not have (I can think of a few specific instances)
There's a good chance that I, being white, got away with at least one thing a Black person would not have (I can think of a few specific instances)
So, the "case mix" is that Black folks have much much higher proportion of minor infractions, which are less likely to be violent in nature.
And since violent infractions are more likely to produce violent interactions with the police....
And since violent infractions are more likely to produce violent interactions with the police....
The net effect is that you'd end up with a LOWER proportion of deaths post-arrest for people who were MORE policed.
And that's reflected in this (incredibly misinterpreted) statistic.
It's a bit counter intuitive, but I bet it isn't so counterintuitive to everyone.
And that's reflected in this (incredibly misinterpreted) statistic.
It's a bit counter intuitive, but I bet it isn't so counterintuitive to everyone.
Put together, we would expect that this stat would both underplay the severity of racist policing, and would push it in the OPPOSITE direction, without being technically inaccurate.
The worst lies come in the form of being technically true and lazy (and racist) interpretations.
The worst lies come in the form of being technically true and lazy (and racist) interpretations.
