Alsimon2, I know you are making the exact same point. I used the examples to tell you I understand. I dealt with theses issues for years. I hated dealing with these issues.
I understand you can’t always apply macro formulas to individual cases. Sometimes there are too many variables. Sometimes one variable has too many possibilities to apply a formula. I did do that. Sorry.
I am going to check out your link. I never said my thinking was original. There are thousands and thousands in just one field that understand what I wrote in my prior post.
I guess my argument is we can’t get exact numbers to figure out who is guilty and who isn’t on an individual basis. We also know we are falling far short of convicting or punishing people for crimes of sexual assault and rape. How do we reconcile this? I was also a bank reconciler when I was a lot younger. Easy stuff. Easier than this.
I read the Stanford link. Great link. I used to coach kids. I like to listen. The kids know so much more than we know about so many things.
The Stanford student comments are upsetting. Reading the comments, it is pretty clear to me we are not capturing a few rapists and assaulters. Quite a few.
I think 2 percent lying is probably too low. The FBI uses 8 percent. And there are subsets of populations in these numbers.
Maybe college girls lie more than the rest of the female population. 7 times more? I doubt it. I don’t think there had been a study on this.
The 98 is greater than 50 argument…we might get more accurate results. I think we do need proof though. Multiple independent cases where the same guy is accused…I think the 98 percent rule could come in handy.
bearpanther - I don’t know if the attorneys on this thread are litigators or not. Maybe they can chime in. I’m not an attorney, but I’ve been involved with a dozen or so civil suits at my company over my career. Here’s my guess - when an attorney is writing up grounds for filing a civil suit to appeal the verdict of a college hearing, they throw in as many things as they can. They’ve basically got a checklist of things that they’ve seen in their prior experience. Item #22 on their checklist is investigatory irregularities, item #23 is sloppy note taking, item #24 is “you didn’t mention this, didn’t mention that in your report”, etc. The more the merrier.
All we’re looking at is a kitchen sink approach. I wouldn’t read too much into it unless claims are backed by tangible specifics and actually are of material relevance to the outcome of the case. The cases I’ve seen - let’s say 50 issues are raised. 47 don’t matter or can’t be proved. It comes down to 2 or 3 that are the key ones. Not necessarily issues of fact - could be issues of law.
That’s not to say that Sessions-Stackhouse did or didn’t do a bad job (or that colleges do or don’t do a good job). But she’s going to be accused of it no matter what. After all, the loser is always going to be unhappy. For that matter, the winner is always going to be unhappy too, regardless of what the law shows on TV portray.
Never said that. In fact I am the last person to say that. But that doesn’t mean the findings don’t need to be put in clear context. Sometimes it isn’t enough to just state the facts, or to not note that these “facts” might still be in flux. The system might not be in its final state of equilibrium yet. You might be measuring data halfway through the reaction cycle.
Yes and as a non-stemmy person Google’s best and brightest will tell you that all the algorithms in the world are shaped by social science. I can type a sentence and 3 different people will interpret that sentence differently based on their unique filter. It’s no more different than tribunals hearing allegations. All we can require of the colleges that they have the most buttoned up processes and procedures availing people of their civil rights and applying those processes and procedures fairly. In a less than slam, dunk case you can perhaps apply some probability but in the absence of hard data I’m not so sure just how accurate that probability can really be. I always thought it would be very interesting to be a jury-picker consultant .
[quote]
The other problem with the Emma math probability is that she and the other two accusers talked which I would think adds a bias factor - they aren’t entirely separate or random. There are two common points - knowing Emma and knowing him.
[quote]
I agree with you Momto3boys. Emma befriended the second woman who had a long-term relationship with Paul, and who had slept with him repeatedly over a period of several months. The second woman said there was a pattern of abuse (when she thought of it - after the fact). The second backed her up but didn’t bring rape charges against Paul. The third was the woman he kissed. Emma found the other two victims and from what I read convinced them that he was a “rapist.”
Alsimon 2, I reread your post 467 and your link. I love the link.
You mentioned you don’t necessarily believe those three scenarios in post 467. I love the post by the way.
Do you think decisions should be made based on extreme scenarios?
Your scenarios are based on Paul assaulting both parties in post 467.
That doesn’t matter to me. I am saying that if Paul is accused by two indendent people, the odds that he is guilty of assaulting one of the two people is dramatically higher.
I’ll be interested in the answer also. But how do you use statistical analysis to filter out the probability that the accusers influenced each other in their decision making and how they shaped their accusations?
This entire sentence was not directed at Emma, but was addressing all three cases I mentioned. For clarity sake, I should have written this:
“Bottom line is no improvement can come from cases where there is no there there either in fact (Duke), in substance (UVA), or in strength of accusation (Columbia - Emma). Lots of noise, but nothing useful, and in actuality may even set back the issue.”
I am not doubting that Emma had some sort of sex with Paul, but clearly I am not the only one does not automatically jump to rape.
You are arguing with me, but it still remains the fact that even adults who looked at the case twice found her claim not substantiated, i.e., they found no evil and no rape or no sexual misconduct of any kind, like you see. And they are much closer and actually looked Emma in the eyes, listened to her words and syntax, and heard her claim. You are living off of third party news. OK, I get your position, just do not think it holds much water given the two tribunal hearings.
dstark - YES!!! - But now you’re switching what we’re aiming at.
Before, you were saying the fact that Adam is accusing Paul meant the chances that Emma was telling the truth are dramatically higher. That not necessarily right (that’s the point of my examples). The logic about independent accusers aiming at the same guy was woefully incomplete, too incomplete to generate any sort of useful number.
What you’re now saying is that Adam accusing Paul means the odds that Paul is guilty of raping somebody, not necessarily Emma, are higher. That’s right! I’m sure that was your intuition behind the 1/2 thing. But the bearing it has on whether Emma is telling the truth is limited. And that’s (mostly) what the specific claim you made was. (To be honest, you only stated a specific claim once. But your supporting statements mostly only made sense if that was what you were thinking, and that’s how I read your posts).
But that’s not the way the legal system works. We don’t say “Well, I don’t know what the guy did exactly, but I’m sure he did something. Fry him”. We just don’t do that.
All the stuff about independent accusers aiming for the same guy was incomplete … at the least, you have to believe in some version of the serial rapist theory (let’s ignore the fact that one accuser was a woman and one was a man please). And since people are saying that the odds went up by factors of 100 or so, they have to have a pretty different view of the world from what their prior posts would indicate they have (don’t ask me to explain this please).
You may think I was harping on a technical issue. You’re partially right. But I’ve told you that I agree with your conclusion in almost every post I wrote to you, and I’ve been trying to beg off from talking statistics for a while now.
@momofthreeboys You’re right about the independent thing. But I think dstark was already fully aware of this. The issue is what are the odds that in the Columbia case we have some scheming people who would coordinate stories. Here’s what I wrote about it:
FYI - five orders of magnitude means a factor of 100,000 times too small.
In ordinary english, what I’m saying is the odds that any given woman you know, say your daughter, would be such a scheming person are probably less than 1 in 10,000,000. But the odds that there might be someone like this who has gotten national attention are not nearly so small. And the fact that accused was found not guilty of all charges means the odds are even a little bit higher, and the fact that the accuser stands to benefit from the attention means the odds are higher still. But no one knows these numbers well enough to estimate them.
You can’t use statistical analysis. We were using temporal analysis. We were considering the (I now believe fictitious) Adam, who allegedly made his accusation before Emma made hers. He couldn’t have known about her accusation because it hadn’t happened yet, and if she had known about his accusation we would have heard about it before now.
Ok…I am sorry if I misled you, alsimon2. I thought I said Paul was guilty. I don’t remember saying Paul is guilty of assaulting Emma. I know there is a difference. I am not playing games here. I don’t remember saying Paul was guilty of assaulting Emma. If I did, oops.
The formula I used had it as a 50/50 chance Paul assaulted Emma.
Anyway, I think we are fairly close in our ideas. And if we aren’t, it doesn’t matter. I stll like your posts.
Ok. I guess i’ve never considered Adam as a real claim other than the idea as presented that he,too, was a friend of Emma through the Greek literary organization.
Two independent accusers. 50/50 cases.
Accuser 1, loses to the accused in a hearing. Accused is not guilty.
Accuser 2, loses to the accused. Accused is not guilty…
The accused can not be proven in guilty in either case.
However, if two independent accusations increase the chances the accused is guilty, in this scenario I presented, the accused is guilty if we use an above 50% standard.
This is why I want multiple accusations considered in a hearing.
I want a school to be able to punish somebody even if that somebody is not found guilty in individual cases.
Because he probably is guilty of at least 1 assault.
Now I talked about theories not being facts. Two real 50/50 cases occuring may be to precise in the real world. I want schools to be able to look at multiple accusations during hearings. The schools can still say the accused is not guilty.
(I don’t think you need two 50/50 cases either. The odds can be favored more towards the accused and he can still be more likely guilty in at least one case. That is another argument).
The second accusation is a big deal if the second accusation has nothing to do with the first accusation. If. There is an if here.
I cam’t believe I said Paul is guilty of assaulting Emma. @@@@.
I’m now thinking that “Adam” pranked Jezebel. But no matter. Math is eternal.
So let’s return to it. I don’t think you’re going to find a model where a second accusation doesn’t lift the probability considerably of at least one of the accusations being true.
First let’s consider a simple model, where we have a population of men, of whom r fraction are rapists, all serial rapists. al2simon, you didn’t include an important parameter in your model, but we need it: f, the fraction of accusations that are false accusations. It is then a simple matter, using the definition of conditional probability, to compute the chance that two accusations of the same person are accusations of a rapist. It’s P(second accusation is accusation of the same guy and he’s a rapist)/P(second accusation is accusation of the same guy). If we let g= 1-r be the fraction of good non-rapists, and t=1-f be the fraction of true accusations, then the chance that the doubly accused guy is guilty would be i/(2f + t(1/r + 1))*
If at least half of accusations are true accusations under that model (t > .5) then the probability that if two people both accuse, the guy’s a rapist can never be less than 75% for any value of r– and that’s if every single guy is a potential rapist. If half of accusations are false, and 1 percent of guys are potential rapists, then the doubly accused guy is guilty with probability 99%.
This is routine conditional probability. And it doesn’t get much less routine if we expand the model to include serial rapists, mono rapists and non-rapists. I’m not going to type out the equations, but the spreadsheet tells the tale: for any plausible numbers, the second accusation increases the probability of the guy’s guilt considerably.
