[dljsjr]

That tweet you referenced is heavily cherry-picked: it talks about the results of violent-crime calls, and it doesn’t discuss location.

There are two issues with this:

1. Civilians are also murdered by police who are responding to non-violent offenses. A good example here would be the classic “shot during a routine traffic stop” situation. That won’t show up in the stat from your sources. Your source tries to say that by “controlling for violent crime, the racial disparity goes away” but that doesn’t make any sense.

2. Huge disparity across different LEO organizations. For example, according to this Harvard study, in Chicago the racial bias in the statistics is much stronger than in other areas: https://www.hsph.harvard.edu/news/hsph-in-the-news/blacks-wh...

That same study also concludes that once you account for and amortize things, you are still 3x more likely to die in an encounter with LEO if you are non white.

[beaner]

The NYTimes, which is a left-leaning institution to begin with, released an infamous paper on race-based police killings itself in 2015 by a black researcher and author, and found that "this data does not prove that biased police officers are more likely to shoot blacks in any given encounter." [0] People were angry that taking a look into the data did not say what they wanted to here.

And as another poster said [1], even if there were bias, that does not equate to likely, in absolute terms. The idea that you're essentially sending a black person to death, or rolling the dice on their life, by calling the cops on them for nonviolent crime, is ridiculous.

[0] https://www.nytimes.com/2015/10/18/upshot/police-killings-of...

[1] https://news.ycombinator.com/item?id=29200539

[dljsjr]

You’ve still completely ignored the peer-reviewed Harvard study. Also I don’t know what left or right social politics has to do with any of this.

[beaner]

I looked at it and the study doesn't account for anything. For example, if a certain demographic were 2x more likely to die in an encounter, but 4x more likely to respond to a benign pullover with violence, then they're actually 2x more likely to be treated with less force than their actions require.

This is basic, the study (at least on my glance) appears to not account for anything like this at all.

Also the source isn't everything but its political leanings are somewhat-important as this type of misdirection that I've just showed here is used all the time to craft narratives around statistics, so even when you're crafting leans in a certain direction and you still can't tell the story you'd like, it's a strong indicator that the data more distinctly show certain patterns.

[dljsjr]

The fact that a certain demographic might be more likely to respond to the LEO with violence doesn’t matter because this isn’t a discussion about causation and correlation, it’s just a discussion about the fact that non-white people are more likely to be subject to LEO violence.

You can’t use a stereotype to excuse LEO’s treating people like stereotypes. Especially when there’s a tremendous amount of evidence that shows that those stereotypes are externally enforced (systemic racism, for example). I don’t want to debate whether or not that is a thing, that’s a separate discussion, and it’s totally unrelated to the one we’re having here.

> If a certain demographic were 2x more likely to die in an encounter, but 4x more likely to respond to a benign pullover with violence, then they're actually 2x more likely to be treated with less force than their actions require.

This is some insanely racist bullshit that you’ve just said so I don’t think that there’s going to be a productive end to this debate and this will be my final comment in the discussion. That’s as civil as I’m willing to be about it.

[beaner]

You understand that I am not saying that black skin is the cause of the way people respond in police interactions, right?

When I say "If a certain demographic were ... 4x more likely to respond to a benign pullover with violence", that is a post-fact observation, with the demographic being the specific collection of people pulled over.

Why would you interpret it the way you did? Holy shit. I feel like we're going into the "math is racist" territory.

> it’s just a discussion about the fact that non-white people are more likely to be subject to LEO violence

There are answers, but you're ignoring them.

[OldHand2018]

The study you linked to above? I think it is highly informational, but drawing specific conclusions from it should be done very, very cautiously. The page you linked to has a Methods section and a Statistical Analysis section. They are quite informative.

1. They are using the Fatal Encounters dataset - probably the best, most complete, least biased dataset (GOOD)

2. It turns out that police aren't actually killing very many people (VERY GOOD), so there isn't a lot of data.

3. They are doing their analysis at the MSA level as a simplification to account for the fact that people move around and don't just stay at their house or only in their city. They state that they are doing it and why they are doing it (GOOD).

4. They are using a statistical process that assumes that everything (race/ethnicity, deaths, etc) is normally distributed within the MSA. In real life, this is not remotely true, but they acknowledge this fact (GOOD?)

5. Their analysis finds that many of the higher racial disparities found within MSAs occur in MSAs with the lowest rates of police killings and/or a lot of places with high rates of geographic segregation. In other words, places with the least amount of data and places where #4 holds the least.

I don't think anyone is going to claim that this is a bad study, or wrong, or anything like that. It looks pretty solid to me. It's just that you are trying to draw some conclusions from it that it can't support.

I would also like to point out that "If you are in a police encounter (traffic stop, arrested for theft, etc), what is the likelihood of being killed by the police?" is not a question being answered by this study. "If you live in this metropolitan area, what is the likelihood of being killed by the police?" is what this study provides the answer to. That includes such things as people in nursing homes, toddlers in daycare, etc. Saying "You are 3x more likely to die in a traffic stop if you are black" is, sadly, probably very true - but it is not what this study is about nor what it proves:

> We use population denominators to align with, and allow comparisons to, previous demographic work in this area, and because using race-specific crime or arrest counts—themselves shaped by racial bias and segregation—yields estimates of a different and potentially biased contrast than the rather simple ones we answer here: in which metropolitan areas are people most likely to be killed by police, what is the difference in these rates by race, and how does this vary across MSAs?

[tashi]

Is Sendhil Mullainathan the "black researcher and author" you're talking about? I'm trying to get a sense of how much fact-checking you've put into this.

[beaner]

Hm, no, good correction. I might be thinking about a separate article from the same time period that I've lost track of.

[ryan93]

https://eml.berkeley.edu/~webfac/card/e250a_f16/fryer-use-of...