Look, you had a perfectly simple mathematical question based on an assumed ability to tell red from maroon 95% of the time. You floored the question, embarrassingly, and now you try to tell us it’s because you were too clever for the question, because, like, the question didn’t say whether it meant “picking out the red from among the maroon 95% of the time” or “picking out the maroon from the red 95% of the time”, as though it weren’t obvious that in a simple, artificial, mathematical problem, the implied ability ran both ways unless it were specifically stated else. Instead you have to throw in some technobabble because another poster seems to have given you an out.
And you produce this rant:
in which you toss out the 95% value you were given and instead make up some numbers that support your own case. Feeble, Doctor. Feeble. If I’d meant to say “The witness is shown to be 99.5% able to distinguish red from maroon” I’d have said so, Doctor.
Almost as laughable as
“I haven’t an answer to the rest of what you wrote so I’ll say it doesn’t matter.”
I repeat what I earlier said about being deeply glad that your professional expertise is unlikely to affect my life, Doctor.
Big deal. We shouldn’t be worried about how accurately the witness can spot a red car. Simply due to their prevalence, the witness could make a guess about what car is red and be more right than wrong. As I already pointed out, it’s their accuracy with respect to maroon cars that matters the most, since A) that is what the witness says they saw on the day in question and B) maroon cars are less prevalent than red. And as I also pointed out, you can’t assume that information from the numbers presented in your hypothetical.
And instead of going through my post and tackling my arguments, and pointing out where my reasoning falls astray (as I did for you), all of this reads like a whiny, hysterical screed.
That’s a rant to you? Wow.
Your 95% statistic allows free room to assume any and everything about the witnesses’ reliability with respect to maroon car identification. That’s why it wasn’t clear to me how you reached the 75% false ID rate. The assumptions you would have had to make to reach that conclusion were not evident in or deduceable from the information in your hypothetical.
Dragon Ash, you’ve been conducting yourself like a fair player as of late. Does what I’ve written strike a cord of reason within you? Do you see where Malacandra’s logic falls apart?
But there has always been other dispositive proof, so that caveat is meaningless. The first time I read about this story I learned more than the fact that the stripper was Black and the men were White. So why should we ignore everything except the race of those individuals? That’s the sticking point. It’s not that people don’t understand your caveat, it’s that it is not applicable to this specific case.
It does appear that interracial rape is much less common than same-race rape. But that is probably due to the fact that most people move in social circles composed of same-race people, and most rape victims are known by their attackers. Since the guys in this instance definitely knew the alleged victim, that most likely made the racial statistics worthless. You can’t just look at raw data like those particular DoJ statistics and make meaningful statements about any particular case.
No, because there was always more information so that the DoJ data was insignificant.
That “challenge” was meaningless. I was slightly doubtful about the accusation when I first read about it, but not because of the race of the inidividuals invovled. The more I read about it, the less likely it seemed that the woman was telling the truth. However, I never thought the race of the individuals involved was significant because there were so many other overriding factors.
I’m not a stats guy, but it looks like in my not a stats guy opinion, your question isn’t an either or answer, despite your desire to make it one.
I suspect if we can come to an understanding why it’s not an either or answer or why it really is, we’ll also understand why using the race as a data point isn’t correct either…although one can certainly have an opinion.
But in order to that, everyone will have to put their privates back in their pants…
Fair enough, Dragon Ash. In the context of a simple mathematical model, do you view it as reasonable to assume that
means [ul]“Can correctly spot red cars from a mixed sample of red and maroon 95% of the time” or
[li] “Can correctly spot maroon cars from a mixed sample of red and maroon 95% of the time” or[/li][li] Both of the above[/ul].[/li]Can you also (or any other disinterested poster) tell us how
can be construed as anything but a leading question? :dubious:
Since the hypothetical stated that the witness is 95% accurate, doesn’t that pretty much nail down the sensitivity and specificity? If we treat maroon as positive and red as negative, the formulas for sensitivity and specificity would then look like this, I believe:
[QUOTE=Malacandra]
Fair enough, Dragon Ash. In the context of a simple mathematical model, do you view it as reasonable to assume that it means [ul]“Can correctly spot red cars from a mixed sample of red and maroon 95% of the time” or
[li] “Can correctly spot maroon cars from a mixed sample of red and maroon 95% of the time” or[/li][li] Both of the above[/ul][/li][/QUOTE]
In my not-a-stats-expert opinion…I don’t know. Because of the way the original case was presented, I certainly was thinking along the lines of the ‘pick maroon cars out of a sample of red and maroon cars’; my question was whether picking maroon cars out of the red/maroon sample with 95% accuracy was the same as picking red cars out of the same sample with 95% accuracy. I don’t know, which is why I asked.
I guess I would assume that accuracy for the two cases could indeed be different - if the answer is ‘both’ - is the 95% an average? Don’t flame me, I don’t know, and I’m honestly asking.
Again, I’m not a stats, guy, but it looks to me that the real question is how many times (what percentage) our witness can’t tell the difference? If the witness can tell the difference between red and maroon 95% of time, then clearly she should be able to tell the difference between maroon and red the same percentage. Yes?
Oh, I’m not throwing in the towl I’m always a fair player; I didn’t like how you were playing, and from the 17-odd pages of this thread, it wasn’t clear to me (my fault, perhaps) of why you were against one particular data point out of many that someone was using in arriving at an opinion.
I’ve been thinking over your comments:
And honestly, I’m struggling to come up with a reaosn why race wouldn’t be one factor that determined whether someone thought another person was sexually attractive - why do white guys rock your boat compared to black guys? Race, sex, age, hair color, along with fashion sense, education, etc - a host of factors are also at work, but race is a pretty major determinant of physical appearence, right? Why wouldn’t it play a factor in whether I find someone attractive (sexually or otherwise)?
Dude, you’ve done nothing flame-worthy - but the honest thing to do if you don’t understand the question is to ask, as you did, and as you with the face singularly failed to until she’d shown a complete misunderstanding of the answer as given. The spoilered text from the original question provided context, as can be seen from the figures.
I’m not a statistician either, which is why I’m sticking to really simple probability theory and not buggering about with bell-curves and normal distribuitions t-distributions and variance and standard deviations and lord knows what…
you with the face has asserted that the 95% reliability figure can be applied only one way, and implies nothing at all about the reliability the other way. She must have had any amount of fun with mathematics.
But it’s interesting to note, I find, that as well as lodging the above objection, you with the face has to come up with some other figures that purport to show that the witness can identify a maroon car as such 100% of the time, and a red car mistakenly as maroon about 99% of the time. If the data don’t support the contention, the data must be revised, I guess. That 95% imaginary reliability just wasn’t good enough.
Anyone want to run at the “leading question” supplemental I posed about the good Doctor?
(“Last ditch”, forsooth! This is fun. And we haven’t even gone over that whole calling me a retard thing, which was truly classy.)
That’s not what the hypo says. This is what is says:
Which is the same thing as saying from a population of maroon and red cars, the witness was able to correctly ID a red car 95% of the time.
This is NOT the same thing as concluding that 95% of the time they are able to correctly ID a maroon car.
Have you guys ever heard of false positives and false negatives? A medical screening test that has a high false positive rate means that there’s a good possibility that you could be healthy but still be informed that you have a disease. A test with a high false negative rate means that you could be sick as a dog and be told that you are healthy.
Understanding this principle should tell you why Malacandra’s assumptions are wrong.
OK - but in this case, aren’t we back to a 95% probability of the witness being correct? For the one case in question, can’t we say we’re 95% confident in the person’s testimony? Wouldn’t that negate the ‘3 out of 4 cars identified as maroon were incorrectly identified’ point?
Man, this thread is making me want to take a stats class online or something.
It’s clear to me that Malacandra and Weirddave have demonstrated a lack of understanding of what information can be correctly drawn from statistics.
If they can’t understand hypothetical stats, they really have no business using real stats to gauge the merits of an allegation.
If someone can show a cite that demonstrates skin color supercedes a factor like genitalia in determining the likelihood of one individual raping another individual, all this discussion about race is, like you with the face said, a red herring. It’s not just the fixation on stats that’s driving me bonkers. It’s the emphasis on race-based statistics that’s nuts.
Someone brought up an individual’s credibility as having predictive power. I agree with this to degree. It’s a fallacy to conclude that because someone has lied in the past, they are lying now, but if they are making the same accusation that they levied in the past and it was found to have been false, then I don’t think it’s wrong to doubt the accuser. However, a statistic is not guiding this judgement. A statistic is a figure–usually an average–that represents a population of observations of behavior. The detective in Weirddave’s case didn’t sit on the stand and say, “Weirddave has a honest/lie ratio of 100%, so therefore I believe him when he says he didn’t do it.” Rarely does this kind of information exist for an individual, but when it does I could see it being relevant in judging the merits of an allegation like the Duke case.
But stats representing the behavior of populations? No, not appropriate.
No. We’ve taken 1000 cars, 20 of which are maroon and 980 of which are red, and asked the witness to ID the colors. If reds are 49 times more prevalent than maroons, the witness could close their eyes and simply guess “red” and be right more than wrong. That alone should tell you the answer to your question. A simple guess would not work with the maroons as it would the reds.
The ability to find a needle in a haystack can not be assumed from stats that can only tell you the ability to NOT identify a needle from a haystack.
Only if my crediting you with 100% ability to tell shit from sherbet means that you can identify shit from a mixed sample 100% of the time, but you could be really really wobbly on the sherbet. Most people would perhaps not parse the statement that way.
That’s a real shame, 'cos correctly identifying a maroon car was what we were hoping the witness could do. :dubious:
And understanding why you with the face finds it necessary to introduce additional variables into a mathematical problem she failed either to understand or produce reasonable objections to until she got some prompts from the floor should tell you a lot about the good Doctor’s grasp of statistics.
Boy, I have never seen a self proclaimed “expert” in a given subject implode so completely so quickly. Usually people who claim to be experts have at least some knowledge of the subject instead of booting 101 level questions on it. I think we’ve moves beyond “insane” into “what else is there to do but shake one’s head and chuckle sadly”.
I’m always a fair player; I didn’t like how you were playing, and from the 17-odd pages of this thread, it wasn’t clear to me (my fault, perhaps) of why you were against one particular data point out of many that someone was using in arriving at an opinion.