That’s a fair point, actually. I’ll see if there’s another way to put what I’m trying to say.
But as a preliminary to that, and this may be something of a concession, I am starting to think that what’s happening is that people (including me) are treating extremely low confidence levels as tantamount to having no basis at all for computing a probability–which is of course strictly false, though works pretty well as a heuristic.
If someone goes to Las Vegas and plays a slot machine, and they pull the handle 100 times, the majority of the time they will pull no matches or very low matches. So if you pick randomly from amongst the pulls, the likelihood is that you will hit on a pull that did not win much or any money.
But if they say they had a pull that won ten dollars, you would not disbelieve them because of all of those other pulls. Winning ten dollars happens. Winning ten dollars is not some incredible outlier that should be disbelieved on its face. Winning 100 dollars is also not some incredible outlier that should be disbelieved on its face. Winning 10 million dollars on a slot machine is an outlier.
The only thing I can glean from threads like this is that some people appear to think most rape or sexual assualt accusations are 10 million dollar claims. I think they are 10 dollar claims or, at most, 100 dollar claims.
And if people are saying they all were playing the same slot machine when it happened, I think the 10 or 100 dollar claim seems even less of an outlier.
This is a fundamental misrepresentation of the dynamic here.
No one is claiming that the likelihood of consensual sex in and of itself renders a rape any less likely. It’s only when the likelihood of one claim is being measured against the other in a mutually exclusive scenario that the a priori likelihood of each becomes relevant.
In the typical case where it’s relevant, the rape claim is being bolstered by the perceived unlikelihood of the alternative consensual claim. To the extent that it can be shown that the alternative explanation of the facts is not actually unlikely, then this bolstering is legitimately removed.
As a an actual real-life illustration I posted earlier (I think in this thread) that in the DSK case his claim that he had had a consensual sexual encounter in the span of a few minutes with a woman he had never met before was thought to be more implausible due to these factors, and this bolstered her claim that she was raped. If it could be shown that she had a history of this type of sexual encounter, then her claim would not be bolstered.
I think it is a possibility.
Let me ask you - what are the chances of anyone raping 25 women and having the good fortune of not a single victim pressing charges at the time? Does that happen often in your opinion?
How about some citations to studies on the matter that you have now gone so far out on a limb that your ego is at stake? You clearly have no fucking clue about the causation of rape or the victims of rape, much less how to determine the truth in any case using probability. Hint: you can’t.
You pulled this probability about rape victim stuff out of your damn ass.
The probability of a rapist being proved a rapist is determined by evidence that the man (usually a man, but not always) has raped one or more people. The probability of a victim of rape lying about the incident is determined in each case by a thorough investigation. And sometimes alleged victims do make it up. But that isn’t determined by probability. Ever. Weather and quantum mechanics are informed or determined by probability.
I remember the first time I encountered this line of reasoning, and I found it as humorous back then as I do now.
It was back when Walter Mondale was running for president against Ronald Reagan. In the closing days of the campaign, with all the polls showing him behind by big double-digit margins, Mondale took up a new slogan. “Polls don’t vote, people vote”.
Some people may have been skeptical of this line, but when the results came in the doubters had to eat their words. Not one poll had voted. Only people. Mondale was prescient after all.
For what it’s worth, Cosby himself and his team of high priced lawyers seem to have decided there’s merit in questioning the character of his accusers. According to this story in the NY Post Cosby has hired a team of investigators to look into the pasts of his accusers looking for evidence he can use in his defense.
A couple of things mentioned already are claims by Beverly Johnson’s boyfriend that she only had good things to say about Cosby and never told him any of the claims she’s making now (there’s also been speculation not mentioned in the article that she made up her claims to support her former model friend Janice Dickinson). And that Katherine McKee (Sammy Davis, Jr.'s…uh, mistress), who claimed Cosby raped her (while not uttering a word of protest, mind you, because it “happened too fast”), had made internet postings speaking glowingly of Cosby’s stand-up act and praising him, and admitting in a separate published interview that she is “used to lying”.
Quite a few of these women are full of it and I’ve felt all along that Cosby was making a mistake by keeping quiet and letting the accusations pile up unanswered. It only adds weight to the claims and encourages others to get in on the cash grab that I suspect is at the heart of much of this. It’ll be interesting to see what Cosby’s investigators uncover and how all this will shake out in the end.
So polling data in a political contest is a true analogy to not having any polling data and just pulling postulates and fabricated data, not from studies, but out of your ass and based on your preconceived idea that prostitutes like to lie about rape? Shit, I’ve heard better statistical arguments from drunken relatives at family gatherings about racist notions.
It is clear you don’t know what a study is, what a statistic is, what probability is and what an analogy is. That is a The Grand Slam of Stupid. Congratulations?
I don’t think this tells us much. This is what lawyers do, and they do this for innocent and guilty clients. All this tells us is that Cosby is rich and hired lawyers.
I’m amused to see conservatives rush to claim that defense attorneys are the way and the truth and the light.
Regarding the prostitute/virgin rape accusation scenario, is this, in your view, the same idea put a different way?
Assume two rape accusations, A1 and A2. For each accusation, there is a body of evidence, respectively E(A1) and E(A2). One of the elements of E(A1) is i=“When presented with the opportunity, the accuser frequently consents to sex.” One of the elements of E(A2) is j=“When presented with the opportunity, the accuser frequently does not consent to sex.”
Consider F(A1), which includes only and all the elements of E(A1) except for i. And consider F(A2), which includes only and all the elements of E(A2) except for j.
Now compute a probability for the statement L=“the accuser is lying” for all four of E(A1), E(A2), F(A1) and F(A2).
Is the following basically the same as what you’re saying?
P(L(E(A1))) - P(L(F(A1))) > P(L(E(A2))) - P(L(F(A2)))
(I’m trying to avoid the original stipulation, which I think was granted simply to make the argument simpler, that the two bodies of evidence are ‘otherwise identical’ since, for reasons I’ve explained earlier in this thread, it is very difficult to know exactly what that amounts to, without just resorting to saying there simply is no other evidence, which has been tried in the thread and led to no progress.)
Unless we have data to put into those equations, all you have are equations. Until Galileo gathered actual data, two thousand years of the top thinkers were safe in assuming that a canon ball fell faster than a bullet. What if real data weren’t just different than the assumptions, but the opposite? Assuming is garbage when you have scientific method at your disposal.
Sounds OK, subject to 3 provisos.
[ol]
[li]i & j are not mutually exclusive as stated. You need to change “frequently” to “usually”, at a minimum. (The further apart i & j are, the more the impact on probability.)[/li][li]You’ve not specified that the crux of the issue here is consent versus coerced. (E.g. if the dispute is over whether the accused was present altogether or similar, then this wouldn’t hold.)[/li][li]You’ve not specified that P(L(F(A1))) = P(L(F(A2))). If you’re assuming this, then the equation should hold.[/li][/ol]
I should clarify that if applied as a general rule the last clause is not really necessary. But if you’re trying to compare it to the situation we’re discussing, then it’s appropriate to add that - in this case the F(A1) & F(A2) are in fact the exact same.
No. Simple algebra can be used to prove that F(A1) and F(A2) are not equal. First divide both by F. Then divide both by A. You are left with 1 and 2. While they are close, and in comparison to an infinite number of real numbers insignificantly different, they are not equal.
A physicist, and engineer and a statistician all go polar bear hunting. The physicist, aims too low and misses the bear. The engineer over-corrects and his shot goes too high. The statistician yells: “You guys nailed it!” and while they are running the statistician turns into tomorrow’s bear shit.
Nice and neat.
It’s like I remember from back in seventh grade or so, we had this writing assignment where we had to complete a story which had broken off at the point where a boy and his dog were adrift on a raft in middle of the ocean. And a classmate of mine wrote “First the dog fell off the raft and a whale are him up. Then the boy fell off the raft and a whale ate him up too. The end.”
Case you ever need to use this again, it goes "… The statistician yells “on average you guys nailed it”.
Just for the record, this doesn’t work because the values you refer to aren’t being multiplied together in the expression you’re talking about. “F(A1)” doesn’t mean (in what I wrote) “F multiplied by A multiplied by 1,” but rather, “The function F when applied to set A1.”
Yeah, I get that. It’s a lot of nonsense without values for any/all your variables. By using A1 and A2, you are implying there is some relation between the two, being the same type of data. They aren’t related at all. Thus the mocking of the equation by the “algebra” showing you were trying to prove 1=2. You don’t get to make up the equations until you have your data sets. First you get your data sets, then you examine them. You do not make up the relationship to fit the drunken family gathering discussion and pretend the data supports it due to preconceived notions, which is what you are doing.
Statistics and probability are appropriately applied to useful samples, not a prejudice from an insufficient sample size, which in this case appears to be zero.
What I am doing is trying to make it clear what F-P and possibly what jtgain are saying. The expression I typed out isn’t part of an argument, it’s just an act of clarification.
As to your larger point, I don’t agree that you have to have exact numbers established by statistical methods in order to do good probabilistic reasoning. You can, for everyday purposes, get by with rough rules of thumb like “I don’t know means 50/100” and “I’m almost certain means 90/100” etc.
Time to find a thread where the actual subject matter is being discussed. What a boring fucking hijack.