Interesting graphic
Posted: Thu Sep 22, 2016 8:28 am
Page 1 of 1
Re: Interesting graphic
Posted: Thu Sep 22, 2016 9:27 am
Not really.
Re: Interesting graphic
Posted: Thu Sep 22, 2016 10:05 am
Black Lives Matter is an ideological and political intervention in a world where Black lives are systematically and intentionally targeted for demise. It is an affirmation of Black folks’ contributions to this society, our humanity, and our resilience in the face of deadly oppression.
Re: Interesting graphic
Posted: Thu Sep 22, 2016 10:14 am
Thunderbunny has also skewed the numbers in order to fit his narrative. Namely, why are the numbers per murderer's race? That makes the racial majority look less murderous of other races than they really are just due to their numbers. Imagine a rare race of people that only had 10 surviving members. White people could commit complete genocide against this race, and it wouldn't even show up on the graph.
Re: Interesting graphic
Posted: Thu Sep 22, 2016 8:37 pm
But, you're logic is backward because you're somehow valuing those 10 people more than everyone else out there. Conversely... if your race of 10 members has one murderer while the white race has 1,000, it would be fair to say that your races has more murderous tendencies.Jeff250 wrote:Thunderbunny has also skewed the numbers in order to fit his narrative. Namely, why are the numbers per murderer's race? That makes the racial majority look less murderous of other races than they really are just due to their numbers. Imagine a rare race of people that only had 10 surviving members. White people could commit complete genocide against this race, and it wouldn't even show up on the graph.
Re: Interesting graphic
Posted: Thu Sep 22, 2016 8:41 pm
That image is also immensely misleading in that Black Lives Matter is not protesting blacks killed by whites, but instead blacks killed by cops (some of whom in recent high-profile incidents have been black themselves), and the institutional issues implied by such.
Re: Interesting graphic
Posted: Thu Sep 22, 2016 10:37 pm
Imagine that race X is 10% of the population and race Y is 90% of the population and that they are both equally likely to murder and that they choose their victims at random from the entire population. The "X killed by Y" bar is going to be barely visible compared to the "Y killed by X" bar, even though both X and Y are equally likely to murder and even though they just choose their victims at random from the entire population. This is why including the "blacks killed by whites" and "whites killed by blacks" bars in this graph is highly prejudicial and does nothing to compare anything we would actually care about.snoopy wrote:But, you're logic is backward because you're somehow valuing those 10 people more than everyone else out there. Conversely... if your race of 10 members has one murderer while the white race has 1,000, it would be fair to say that your races has more murderous tendencies.Jeff250 wrote:Thunderbunny has also skewed the numbers in order to fit his narrative. Namely, why are the numbers per murderer's race? That makes the racial majority look less murderous of other races than they really are just due to their numbers. Imagine a rare race of people that only had 10 surviving members. White people could commit complete genocide against this race, and it wouldn't even show up on the graph.
Re: Interesting graphic
Posted: Fri Sep 23, 2016 1:01 am
100% of people who are murdered are killed by other people.
Re: Interesting graphic
Posted: Fri Sep 23, 2016 1:19 am
Are they?Jeff250 wrote:they are both equally likely to murder
Re: Interesting graphic
Posted: Fri Sep 23, 2016 6:08 am
Note that I'm not necessarily saying that the graphic isn't prejudiced... I'm just saying that your argument isn't that strong...Jeff250 wrote:Imagine that race X is 10% of the population and race Y is 90% of the population and that they are both equally likely to murder and that they choose their victims at random from the entire population. The "X killed by Y" bar is going to be barely visible compared to the "Y killed by X" bar, even though both X and Y are equally likely to murder and even though they just choose their victims at random from the entire population. This is why including the "blacks killed by whites" and "whites killed by blacks" bars in this graph is highly prejudicial and does nothing to compare anything we would actually care about.snoopy wrote:But, you're logic is backward because you're somehow valuing those 10 people more than everyone else out there. Conversely... if your race of 10 members has one murderer while the white race has 1,000, it would be fair to say that your races has more murderous tendencies.Jeff250 wrote:Thunderbunny has also skewed the numbers in order to fit his narrative. Namely, why are the numbers per murderer's race? That makes the racial majority look less murderous of other races than they really are just due to their numbers. Imagine a rare race of people that only had 10 surviving members. White people could commit complete genocide against this race, and it wouldn't even show up on the graph.
In your new example: both races equally likely to murder, pick their victims totally randomly - then that plot would match your population statistics - the common victim bars would both match... and the distribution between the common perpetrator bars would match the population distribution.
If you look at the raw numbers: the number of perpetrators are very close to each other for both black and white race - which means that per-capita the homicide rate for African Americans is higher. Also, if you look at the plot between the common "by X" in the plot above it's obvious that intra-racial homicide is much more common than inter-racial homicide. You can't really argue the validity of either of those facts, assuming the plot above isn't falsified.
Now, we can argue all day about the usefulness of the plot. I think we'll all agree that there's a link between poverty and increased homicide rates, but are the wealth (and crime) differences a result of systematic oppression of African Americans or a result of cultural differences? I'd argue that there's a large "systematic oppression" (whether conscious or unconscious) component, but I don't think it's the only component. The point of the plot dredges up the Guliani argument that it feels disingenuous to riot over one person killed while hundreds of others are killed without a blink of the eye, but that's not exactly a fair point. As has been pointed out - it's about law enforcement's systematic oppression of African Americans - and our society's doing the same.
The really hard question which we have to face both as a nation and as individuals is what should we do about it - and I'll tell you now that every single individual has something that they can (and maybe should) do about it - the question is 1. finding a way to all agree to row in the same direction and 2. working up the courage to do our part about it as individuals.
Re: Interesting graphic
Posted: Fri Sep 23, 2016 8:46 am
I don't know which you mean by common victim bars and common perpetrator bars. The only two bars I've spoken about are the "whites killed by blacks" and the "blacks killed by whites" bars. Returning to my example, suppose that someone from X or Y has a 1% chance of murdering a person chosen randomly from the entire population. We would then expect the # of X killed by Y per |X| to besnoopy wrote:Note that I'm not necessarily saying that the graphic isn't prejudiced... I'm just saying that your argument isn't that strong...
In your new example: both races equally likely to murder, pick their victims totally randomly - then that plot would match your population statistics - the common victim bars would both match... and the distribution between the common perpetrator bars would match the population distribution.
(# of X killed by Y) / |X| = |Y| * Pr(Y murdered & the person murdered was an X) / |X| = |Y| * Pr(Y murdered) * Pr(a person is an X) / |X| = 0.9 * 0.01 * 0.1 / 0.1 = 0.009
Moreover, we would expect the # of Y killed by X per |Y| to be
|X| * Pr(X murdered) * Pr(a person is an Y) / |Y| = 0.1 * 0.01 * 0.9 / 0.9 = 0.001,
which is much lower number, despite the fact that both X and Y are equally likely to murder and each murders randomly from the entire population. Note that without misleadingly normalizing by murderer's race and instead normalizing by the entire population, then you end up with the same # in each case: 0.0009.
Note that I'm not saying that the conclusion that the OP came to doesn't happen to be correct, which is what you seemed to be concerned about. I'm saying that his argument included deceptive statistics that served only to prejudice because your typical reader can't be expected to recognize the deception and perform the math to fix and properly interpret the numbers. The problem is compounded because the bulk of his argument is a red arrow pointing to how low one of these bars is.
Yes, because that's the way I defined it. I suspect you're talking about black and white people though instead of X and Y. If you really have a case to make, I'm sure you can do it without resorting to using deceptive figures.Thunderbunny wrote:Are they?
Re: Interesting graphic
Posted: Fri Sep 23, 2016 8:32 pm
(In your math you normalized by the victim's race.)Jeff250 wrote:Returning to my example, suppose that someone from X or Y has a 1% chance of murdering a person chosen randomly from the entire population. We would then expect the # of X killed by Y per |X| to be
(# of X killed by Y) / |X| = |Y| * Pr(Y murdered & the person murdered was an X) / |X| = |Y| * Pr(Y murdered) * Pr(a person is an X) / |X| = 0.9 * 0.01 * 0.1 / 0.1 = 0.009
Moreover, we would expect the # of Y killed by X per |Y| to be
|X| * Pr(X murdered) * Pr(a person is an Y) / |Y| = 0.1 * 0.01 * 0.9 / 0.9 = 0.001,
which is much lower number, despite the fact that both X and Y are equally likely to murder and each murders randomly from the entire population. Note that without misleadingly normalizing by murderer's race and instead normalizing by the entire population, then you end up with the same # in each case: 0.0009.
Given normalizing by the perp's race:
X killed by Y per Y: .9*.01*.1/.9=.001 (1% of the X population density)
Y killed by X per X: .1*.01*.9/.1=.009 (1% of the Y population density)
If you furthermore looked at the other two bars:
X killed by X per X: .1*.01*.1/.1=.001 (1% of the X population density)
Y killed by Y per Y: .9*.01*.9/.9=.009 (1% of the Y population density)
That's what I was trying to say with the common victim bars matching.
The question is what's the story that you're trying to tell. If you're trying to say "okay, I've got a victim, what are the odds on this perp's race" - then non-normalized gives you the answer. If you want to say "okay, I have a victim, which race is statistically more likely to have produced the perp" then normalized gives you the answer.
Re: Interesting graphic
Posted: Sun Sep 25, 2016 11:41 am
Whops, you're right--I had accidentally swapped the divisors, and I didn't notice in my post that the numbers came out the opposite of what I was saying they should be (X the smaller race looked 9x better). But when you divide by the right one, they prejudice the smaller race as I was saying (X the smaller race looks 9x worse).snoopy wrote:(In your math you normalized by the victim's race.)
Given normalizing by the perp's race:
X killed by Y per Y: .9*.01*.1/.9=.001 (1% of the X population density)
Y killed by X per X: .1*.01*.9/.1=.009 (1% of the Y population density)
I don't have an objection to these bars, but these weren't what the red arrow was pointing to either.snoopy wrote:If you furthermore looked at the other two bars:
X killed by X per X: .1*.01*.1/.1=.001 (1% of the X population density)
Y killed by Y per Y: .9*.01*.9/.9=.009 (1% of the Y population density)
If you have bars on a graph, there's an expectation that the bars should be comparable. For instance, you would expect X and Y in my example to be the same because they are both equally likely to murder and have no prejudice in who they murder, and yet X looks 9x worse. This issue can resolved by instead of normalizing by the murderer's race you normalize by the entire population. Otherwise, the bars on the graph are, while mathematically correct, deceptive and prejudicial.snoopy wrote:The question is what's the story that you're trying to tell. If you're trying to say "okay, I've got a victim, what are the odds on this perp's race" - then non-normalized gives you the answer. If you want to say "okay, I have a victim, which race is statistically more likely to have produced the perp" then normalized gives you the answer.
Re: Interesting graphic
Posted: Sun Sep 25, 2016 8:34 pm
Yeah, I think ultimately the motive is to make the one with the big red arrow look as small as possible.Jeff250 wrote:This issue can resolved by instead of normalizing by the murderer's race you normalize by the entire population. Otherwise, the bars on the graph are, while mathematically correct, deceptive and prejudicial.