Hacker News Reader

Comment by bryant | Hacker News Reader

[flagged]

From your source, white/black people shot to death by cops in 2018 breaks down as 399/209. That's about 2x.

Now if I take a look at: https://www.ojjdp.gov/ojstatbb/crime/ucr.asp?table_in=2&selY...

And I focus on arrests made in situations such as incidents of weapons possession, aggravated assault, or the violent crime aggregate, I don't see much of a discrepancy versus that 2x. For example, for the violent crime aggregate it breaks down as 288620/187470. That's about 1.5x. That is, 1.5x as many white people as black people got arrested for violent crimes. Yet 2x as many white people as black people got shot by cops in that year.

I don't see the relevance. Can you help me understand how this is relevant to police shooting and killing 3x more Blacks than Whites per year?

They don’t. They shoot more caucasians per year by a significant margin.

> They don’t. They shoot more caucasians per year by a significant margin.

Not when normalized. See my parent comment.

Or, reposting:

    Ethnicity | EOY-19 killed | July-19 pop est | humans per 10m

    ------------------------------------------------------------
    
    White, non-hispanic | 370 |     198,256,671‬ |          ~18.7
    
    Black               | 235 |      43,984,096 |          ~53.5

You're normalizing to raw population. You want to normalize to "armed conflicts with the police" or similar.

Using the above FBI 2018 homicide statistics, blacks committed 3177 homicides in 2018, while whites committed 3011---in other words, blacks account for ~50%, despite only being 13% of the population.

Taking that proportion as a rough "criminality index," and multiply 53.5 by 13/50, and things look very different.

The whole point of the comment you are questioning is to also normalize it to something besides total percent of population. The new comparison seems to say that if you are a violent criminal your chances of getting shot by the police are higher if you are white.

This population measurement is not useful for determining racism. It does not consider that blacks make up 13% of the population, but commit 52% of the homicides. It doesn’t consider if the shootings are justified. It doesn’t consider black cops shooting black suspects.

There is in general insufficient data to support or reject that there is systemic racism.

Here is a good article on things to consider https://www.citylab.com/equity/2019/08/police-officer-shooti...

"black cops shooting black suspects" ... that's still police violence against black people, still racist.

You think there's no black servants of white supremacy?

Yeah but that’s only one way to calculate it. Other ways might be to calculate based on number of encounters, or number of criminal acts, or to control for location, or control for income. Etc.

I think it’s important to be really careful saying a given statistic, any statistic, says what we think it says.

Please bring one such alternative that effectively captures the impact on a per-capita basis.

Otherwise, this is a false analogy, since the normalized measurement accomplishes precisely this.

It really doesn’t and I sense that you seem to be downplaying the whole concept that statistics can be interpreted in many different ways.

Edit: I can’t reply to your reply, I think it is a new hacker news feature that’s trying to slow down these back and forths. I listed several ways we could control the data, and I’m not proficient enough in statistics to do all that myself.

One talking point is: per police interaction, an unarmed non-Hispanic Caucasian person has a 4 in 10000 chance of being shot while an unarmed African American has a 3 in 10000 ... so the only huge person actually has a higher level of risk per interaction.

Again, this is not my area of expertise.

I’m not suggesting what all these numbers mean. I am suggesting that when you bring numbers, it’s only fair to discuss different ways to view the numbers.

> It really doesn’t and I sense that you seem to be downplaying the whole concept that statistics can be interpreted in many different ways.

Statistics can be presented in any number of ways; data can be used to lie in any number of directions.

But if you're not going to come here with a counterproposal, you're only undermining the movement Black Americans have been struggling with for so long by stating obliquely that these numbers are a lie.

Edit per yours (sorry you've been hit by rate limits. I get the feeling):

> One talking point is: per police interaction, an unarmed non-Hispanic Caucasian person has a 4 in 10000 chance of being shot while an unarmed African American has a 3 in 10000 ... so the only huge person actually has a higher level of risk per interaction.

The unknown variable here is the number of police interactions by race and whether they were the initiator or the intended target (or suspect, if a crime had taken place). That's why it's normally just reduced to shootings by police.

If you have the details and the exact breakdowns, it'd be helpful. But the BJS doesn't seem to discriminate between whether someone initiated the contact or whether they're the target of law enforcement action.

-------

Also, thank you for engaging.

Reply on news.ycombinator.com