Why not?
Just to elaborate on my response, you seem to be basically saying that my calculation is incorrect because background statistics don’t matter.
But the calculation is an example that shows how background statistics DO matter.
So basically you are saying my calculation is wrong because you think it’s wrong.
Unless I misunderstood you.
Yeah, but if I were an android, I like to think I’d go out like HAL. If I’m going down, other people are going to go down with me. 'Cause that’s the kind of person I am.
No. That’s faulty thinking.
For example, give me a deck of cards and allow me to pick one and look at it and direct me to tell you something about it.
I say:
“I picked a heart.”
Now give the pack to somebody else. Ask them to pick a card and tell you something about it:
They say:
“I picked a seven.”
Which is more credible?
See the problem? The problem is not about the probabilty of a given event occuring. I’m much more likely to pick a heart than a seven. The problem is about somebody lying about a given event.
Because the accusations for analyses are selected and not randomly generated. Since they are selected, their probability for occuring is moot as long as they do, in fact, occur.
Let me reframe the question differently, so you can see where your mistake is:
We examine a pool of 100,000 reported rapes.
99,000 of these rapes are white women accusing white men.
1,000 are black women accusing white men.
If we randomly select 1 white woman and 1 black woman from this sample, which is more credible an accusation?
You are saying, black women are only 1/99th as likely to report rape by a white man than a white woman, therefore the accusation by the black woman is less credible.
Or:
From that same sample of 100,000 you could select 10 accusations from black women and one accusation from a white woman and ask which are more credible.
As long as you are selecting from a sample the probability distribution of what you are selecting is moot so long as it actually exists for you to select.
Not true.
Not true.
Not true.
In my calculation, I assumed that all rapes are reported.
Anyway, my conclusion follows from my assumptions.
Not at all. If you can’t understand why it’s not a contradiction to assume women of both races are equally likely to fabricate a rape claim and, after considering the backround statistics, conclude that a black woman’s claim is more likely to be false, then it is you who have failed to think critically.
Background statistics matter. And yes, it’s about math.
Why not point to the line post 207 that you disagree with?
Actually, now that I look over this, that 1% should be 50% if we are to assume randomness. If a white man rapes at random, then 50% of the whites convicted of rape should have black victims. In order to conclude that black women makes disproportionately more false rape accusations, you need to assume that black women report disproportionately more accusations of rape. So assumption D should be modified to " You’d also have to assume that whites and blacks are equally likely to report rape when it truly happens as well."
In your givens, you’d need to be told that white women are responsible for a smaller percent of accusations than black women.
I ALREADY DID. First, you said that X = the total number of rapes (we’ll make an assumption to correct your first mistake; you haven’t accounted for rapes within the same ethnicity, so we’ll assume that X is the total number of interracial rapes and set aside the other cases).
If the total number of interracial rapes is X, and the numbers of false reports of such rapes by both African Americans and Caucasians is X/6 (given a rate of false reporting of 1/6 for each ethnicity), that means that not only are the numbers of rapes for each the same, but they are both equal to the total number of rapes. I already pointed out your fuck up, and said that the numbers should be Y/6 and Z/6.
Thus, not only do you introduce the frequency of rapes incorrectly, you suggest that they are equal, but then introduce different prevalence rates (99 versus 1) later on. Which is it, equal or vastly different?
Then, in your equations, X suddenly becomes equal to 1. Why even bother with all the definitions of X, which are fucked up anyways, when you set X equal to 1?
But all the equations are not even necessary. If you say that the rate of lying is 1/6 for both races, you can make Y and Z whatever you like. Say 1 and 99. If they both lie 1/6 of the time, the total numbers of lies will be 1/6 and 99/6. There will be more total lies, but the probability of a lie will remain 1 out of 6. Given a Caucasian and African American woman, their credibility will remain the same because they are equally likely to tell the truth (1 out of 6 will lie).
Think hard about my penny and nickel example. The raw number of lies or the raw number of heads is determined by the number of occurances, but the rate of the process will be the same. What you’ve done is talked yourself around in a circle from the raw number of rapes being greater for one race to the raw number of lies being greater for one race, but the credibility of reports will remain the same, 1/6.
But if we take any information about the propensity for lying completely out of the equation, we can still make logical choices. Working with your deck of cards, I will pick a card and make three claims about what the card might be. Only one is true. The claims are:
I have a seven.
I have a jack.
I have a heart.
You pick the correct statement and you win a 50% stake in The Bellagio.
(Obvioulsy these three claims could be made by three different people, as well. I was just trying to restrict as many variables as possible.)
Maybe if Bricker would stop abandoning these threads whenever somebody proves him wrong, we could settle this issue once and for all.
You didn’t read the question I was answering, which specifically asked me to assume that 99% of rapes were black on white and 1% were white on black.
Idiot.
Nowhere did I assume a rate of false reporting of 1/6 for each ethnicity.
Absolutely not. The number of rapes is different. The original question specified that 99% of rapes are black on white and 1% are white on black. Try reading a little before you post, moron.
I never suggest the prevalences are equal. I suggest that the rate of false reporting is equal.
What a moron you are. The X’s cancel out.
Anyway, you admit that you aren’t familiar with Bayesian reasoning. Why don’t you go educate yourself a little before posting about things you don’t understand?
No I’m not. In plain English, I’m saying that if the number of false reports is the same for two groups; while at the same time the actual incidence is a lot less for one group; then a report by a member of that group is more likely to be false.
Then explain what you mean by this:
X/6 does equal X/6, right?
Then how does the number of false reports for both black and white women equal the total number of rapes divided by six?
Then the credibility for both is equal, and only the raw number changes.
Your errors are at a level far below that required of any advanced reasoning of any kind. I’m starting to feel a little like I’m picking on you, so if you can’t demonstrate that you are not challenged in some way, I’ll have to back off out of guilt.
Let me take you by the hand and guide you through this step by step.
A. Your assumption: black and white women are equally honest.
B. Given #1: 25% of rape accusations about white men are false.
C. Given #2: 1% of rapes are white on black.
D. If black and white women are equally likely to lie, then the false accusation rate for black women is 12.5% of the total accusations made. The same for white women.
SCREEECH! Stop the presses!
That’s a bad assumption. There’s no reason to assume that black women are reporting the same number of complaints as white women. But that’s the only way you could assume that black women contribute to half of the false accusations made. If there are more white women being raped than black women, then you’d expect white women to contribute more false accusations than black women. Assuming that both groups are equally honest, this contribution would be proportionate to the number of white women being raped, but in an absolute sense there would more false claims assoicated with white women.
And that’s where the rest of your problem falls apart.
The number of false reports is different than the incidence of false reports? Yeah, I’m going to have to excuse myself now. Sorry for your troubles.
That only works if 50% of women raped are white and 50% black. You derive x/6 from this but then apply it to .01x in your next sentence.
No. No. No. This is just embarassing. Lets make it simple.
We start with 100,000 reported rapes. According to your data 99,000 of these are white women reporting white men and 1,000 are black women reporting white men. 25% of each group is a false accusation. 250 black women are lying out of a 100,000
Again, embarassing. If you start with 25% of accusations are false, than you regardless of the distribution of other factors that 25% must hold since it is a given condition.
No. Again, we can do this with simple math rather than bad algebra. Let’s start with a pool of 100,000 and apply your steps. I’ll put you in quotes and apply your steps outside of quotes.
- “99% of rapes are black on white and 1% of rapes are white on black;”
99,000 black on white rapes. 1,000 white on black.
- “black and white women are roughly equally likely to fabricate a rape claim;”
Ok. We apply a 1 in four fabrication rate to both groups giving:
24,750 false black on white rapes. 250 false white on black rapes.
- "all other things are equal, then
If a black woman makes a rape accusation, there is a 94% chance the claim is false, and if a white woman makes a rape accusation, there is a 14% chance that her claim is false."
No. If a blackwoman makes an accusation the chances of it being false are 1 in four. Since we have selected a black woman, we apply the math to her.
You start with that assumption. You do not follow it in your math.
In a later post you ask what step or line is wrong. Your first mistake is when you generate x/6 in your third line as the number of false accusations made by black women. Since black women making accusations represent 1 percent of your population the actual number should be .0025x. You mistakenly assume that being equally likely to make a false accusation means that they represent half the population of false accusations. This directly contradicts the later statement that only 1% of the population is black woman making accusations. They cannot be both 1% of the total population yet have one half of that number equal 12.5% of the total population at the same time as you state.
This is an inherent contradiction in violation of the givens you accepted as well as being simultaneously a logical impossibility.
You make a whole lot of other mistakes too. In fact, I can’t find very much that you did right. The whole thing is screwed from the beginning.
Your math sucks dick.
It means that if there are 10,000 rape reports, 7,500 represent real rapes and 2,500 are false reports.
Half the false reports are made by black women. That’s 1,250. 1,250 = 7,500 / 6.
The other half are by white women. Simple.
I have no idea what this means.
GMAB. Did you or did you not accuse me of making an error by not counting intraracial rapes?
LOLOLOL.
BYEEEEE.
but consensually, hence, no rape.
How’s the Mustang? I actually thought of you while getting out of mine a couple weeks back. Odd, isn’t that?
Mine’s actually pretty dirty because my driveway is full of deck-building material. It’s a little embarrassing, but I don’t want to take it to some carwash and end up with some kind of scratches.