I’m afraid this isn’t correct, Steophan. Your analysis was correct assuming all the variables involved were independent. iiiandyiii made two valid criticisms of your analysis. First, he pointed out that some of the basic probabilities you gave were not supported by research. Second, he pointed out that at least one of the pairs of probabilities you used was not independent of each other, though in your analysis you treated them as though they were.
There is another thing to worry about concerning your analysis. Though it’s mathematically correct as far as it goes, such that if its numbers are accurate and if its variables were independent it would show that the disparity between black people and white people killed by police is “as expected,” this does not in itself show that black people and white people who behave similarly when confronted by police are equally likely to survive the encounter. There is plenty of room for confounding factors here which may make your analysis irrelevant. What is needed to establish that is research that measures what happens when black people and white people who behave similarly are confronted by police officers. Harder to carry out to be sure! But no less valuable for establishing what you’re hoping to establish.
I am assuming that the majority of encounters between young black men and the police are not related to violent crime. If so, the effect of those that are will be small.
That’s entirely correct. My argument is based around iiandyiiii claiming that the rate of shootings of young black men was so high that it was strong evidence of racism, and my attempts to refute that.
This seems to be handwaving, especially if you don’t have any better data.
If you have better data, please present it. From the data we have, there’s no indication that black people are any more violent towards the police, statistically, then they are towards non-police.
Frylock, do you think that the data from post #57 accurately addresses/refutes the argument that greater contact between police and black people explain the larger disparity in police shootings of black people as compared to overall statistical violent crime among black people?
Alright, I’ve gone back and read through, and I think this is the cited source for the 14 figure.
I say “source”, but it actually links to a source itself for the claim that doesn’t exist anymore (and, in reading the old thread, didn’t exist at that point, either), and which seems to be a news aggregation site rather than a data source in and of itself (that is to say, we would expect a further cite from them for the figure.
As an aside, may I strongly suggest that anyone not read the comments on that second site? I thought for research purposes I’d better do so, and was quickly proven wrong.
I am not sure what that chart is saying. Is it saying that between 1995 and 2000, there were about 3500 incidents in which police used force against white people? How is “use of force” defined here, I ask, because that seems like an incredibly low number.
I want to make sure I’m understanding that correctly before saying anything about the information.
Also I’m pretty sure I’m misunderstanding something in the chart because if my understanding is correct, then a random black person in America was around 3000 times more likely to have police force used against him than a random white person in America between 1995 and 2000. That’s orders of magnitude higher than any numbers I’ve seen mentioned before!
Though as I typed the above I realized the calculation I used to arrive at 3000 assumed all the incidents of use of force involved distinct civilians, but I’m dubious that this is the one thing that would make the number come out right. More probably I’m just not understanding what the numbers in the chart represent or something.
There were far more total incidents, but from earlier pages, this table only includes incidents of police force in which the race of both the police officer and the subject were known.
Further, the layout of the chart is a bit confusing – there are separate columns for officer and subject race, and one must add up the various numbers from the “inter-racial” and “intra-racial” sections (top left, top right, bottom left, and bottom right sections) to get the total use-of-force incidents for each race.
Look at each corner separately – each corner shows the numbers for the race of the cop and subject – top left is “black cop, black subject”; top right is “black cop, white subject”; bottom left is “white cop, white subject”; and bottom right is “white cop, black subject”. For this discussion, the race of the cop doesn’t matter – we’re discussing incidents in which the “subject” of police use of force (we’re assuming that any police use of force must necessarily have involved violence/resistance by the subject) is black as compared to white.
Okay that explains a lot of what I was confused about. Unfortunately, I think it also makes the chart useless for our purposes. It’s too likely that whatever led to the difference between cases where the races where known and cases where the races weren’t known, would also work to bias the sample in some racial direction or other.
This is a possibility, but I’m not sure if it makes it totally useless – the numbers are pretty close to the national numbers for violent crime statistics in terms of ratio.
If anyone cares about the opinion of this lurker, one who works professionally with statistics, this quoted passage is 100% correct, and Steophan clearly is misusing statistics. The irrelevant mention of expected value in post #50 is enough to prove to me that this person is ignorant of how probability and statistics work. Expected value has literally nothing to do with this discussion.
If we’re trying to figure out whether the amount of black people killed by the police is more than it should be, if the police treat black and white people equally, then the expected value is exactly what we need to determine.
The expected value for discrete events is equal to the sum of the products of the probabilities and values. Please explain how this is relevant. Your ‘explanation’ above is not sufficient.
As an example towards the larger point: the percentage of men in the American popoulation with height greater than 6’ is x%. The percentage of American males in the NBA is y%. Therefore, the percentage of American males taller than 6’ in the NBA is x*y%. True or false?
Black men in the NBA average between 6-9 (or 14, whatever)x as many fouls as white men in the NBA. Black men also average twice as much play time as white men. Your contention, based on the logic you display in this thread, is that black men actually average 12-28x as many fouls per minute as white men. The truth is actually 3-7x. Not only are you wrong, you are wrong in the exact opposite direction!
You’re arguing something entirely different from what I’m arguing.
I’m not talking about how often young black men get shot per minute, or per encounter with the police, but over their lifetime.
To try to use your analogy and the actual figures we’ve been using… The claim is that black basketball players are being penalised for fouling 21 times more often than white ones over their career. But people watching the games independently see that black players are committing between 6 and 14 times more fouls per match, regardless of the official’s decision.
The question is, is the difference because of racism? My claim is no, because black players play at least twice as often as white ones, so whilst they only commit 6-14x more fouls per match, over their career they will commit at least 12-28x more fouls.
To go back to the original issue, there is no claim that young black men are 21x more likely to be shot per encounter with the police, but a claim that they are 21x more likely to be shot by the police overall.
The 14x number has been debunked multiple times – why do you keep using it? If I find a blog in which some guy says that white people are 20 times more dangerous to cops without any references, does that mean I can use it in this discussion?