How are you able to assign this “not much more” credibility estimate? What is the logical basis for it being “not much more” as opposed to it simply being in direct proportion to the rape disparity? If your thought process can be reduced to math–you were the one extolling the virtues of such a while back, right?–then you should be able to show the work that led to this estimate. Otherwise, it just looks like you are flailing your arms around, hoping to create the illusion that what you speak is grounded in something more than nonsense.
The problem with using math to drive the entire decision making process is that the default equal credibility that you are assigning to the claimants is 1) not a mathematical quantity, and 2) I don’t know what it is, either in the level of confidence you’re assuming, or the rigidity to which you’ll adhere to this assumption.
Since I can’t translate your assumed confidence in an unknown claimants credibility into numbers, I can’t do numerical comparisons against it.
Hold on a sec; the little lego people around here are talking to me. They say…that you’re going to claim I’m waffling and am a disingenous punk. Well, I never doubt a lego person, so I had better answer the inevitable question: “So if you can’t run numbers about it, then how can you be so sure that the population/race/rape frequency stats are going to be overwhelmed by that unknown level of initial credibility?”
Some numbers I was given a while ago indicate we have:
3836000 total females
1,398 actual white rape victims
790 actual black rape victims
From that, we can calculate:
1/1753 chance or any w-o-rape
1/2744 chance of wow rape
1/4856 chance of wob rape
64% of rapes were wow
Now, if this equal credibility you’re granting these women has to be enough to overwhelm the four-digit odds that a rape even occured, or you’re not going to be able to withhold judgment at all, correct? So I can assume that you’re assigning at least that much credibility to the claimints. Compared to those odds, the ‘two red side, four blue side’ difference of odds we have when we compare the wow rapes with the wob rapes really is insignificant.
No that is not what we are doing. This is what you think we are doing, but it is not what we are doing.
We are making best-guesses about the REPORT of an event.
We are not guessing about the event itself.
The minute you switch over to guessing about the report of an event the model changes and you can no longer use the frequency of the event in the past to determine credibility.
Now you must use the rate of inaccurate reports of that event.
You are stating that using your method will produce results better then a random choice.
But I keep asking for and have yet to see any kind of logical proof or real-world data that indicates your method works.
In the absence of that logical proof or real-world data you can’t just say “trust me, my method works”, show us with symbols and logic and sets, etc. why we must agree with your logic.
I don’t even know what a zombie is, how can I assign a probability.
I understand your point. There are lots of claims that we apply all sorts of logic to and arrive at our best guess. Unfortunately, we are not allowed to include other information in the OP.
If the OP was:
Person A claims to be Osama Bin Laden
Person B claims to be the POTUS
Then the exact same logic would apply that I have been arguing all along.
My final thought is to emphasize that I see your point and that, as humans, we use exactly the method your are talking about. Humans brains are very effective, but prone to “sloppy” thinking and incorrect results also.
But this discussion is not about how do humans do it. This discussion is about what information do we have that is provably true.
So why are you treating it like is a quantity, if you admit that it is not? If two people are equally credible, it doesn’t matter if they are 100% credible or if they are less than 100% credible. They are equally credible. Which means that their stories are equally credible. Real simple, if you ask me. No need to bring out any equations.
Assume that we hold that both claimants are 85% credible. What changes?
Are you talking to me when you say “you”? Because my assignment of credibility has fuckall to do with the “odds” of being raped (I still can not believe this language is being used to describe a crime that is the antithesis of randomness and predictability, but whatever, it’s a surreal thread, and apparently lions will lie down with lambs before everybody understands that prevalence doesn’t denote probability either).
Let’s all be on the same page here.
begbert2 Do not mock the risen and hungry spectre of the Great Emancipator, lest he send his hordes of freed colored people against you.
Of course, as with any other area, you can use math to make any point you’d like to make, if you so choose. Were you hoping for a specific result before you started, and if so, why?
For example, in the state of North Carolina in the year 2004, 1 out of 2120 white women and 1 out of 1104 black women were raped. Therefore, any given black woman is twice as likely to have been raped as any given white woman. Do we therefore reverse your conclusion, and become twice as likely to believe a black claimant?
Hell, tell me what result you want, and I’ll use actual numbers to reach it. Want the black woman to have zero credibility using your methods? We’ll just calculate the frequency in which black women are gang raped by white members of the Duke Lacrosse team. It’s never happened before, so apparently it’s not possible.
As an aside, you keep forgetting the word plausible when formulating some of your more extreme hypotheticals. Let me repeat, those of us who believe both should be considered equally credible, without any evidence showing a problem with their individual credibility, do so only when both parties are making plausible claims. You also asked where the cutoff is, and I think it’s likely arbitrary, as is everyone’s opinion of what constitutes plausibility. I think you’ll find it difficult to demonstrate to anyone that a black women being raped by a white man in the state of North Carolina is anything but plausible, but knock yourself out if you’d like. Zombie Abe Lincoln, on the other hand, is quite a different story.
Let me repeat that. The key word is plausible. P. l. a. u. s. i. b. l. e.
Nothing outside of math and formalized logic is “provably true”, including any math or logic that is based on real-world premises. All else is estimates, probability, error approximations and the like which translate into (drumroll please) best guesses. (Which you yourself seem to admit a little earlier in that very same post; see below.)
Regardless; if you demand the impossible, I promise not to deliver.
Here is what you’re telling me:
Step 1: Events may or may not have happened, in the real world, about which statistics may be used to draw vague conclusions.
Step 2: A person or people makes claims about these events.
Step 3: As a result of the claims people made, step 1 ceases to be true.
I refuse to accept this because I feel it to be silly, not the least because step three makes no sense whatsoever.
The fact that step 2 would probably give us better information than the information that is actually available to us is NO REASON to discard the information one already has.
And didn’t I already cover this in post #559? Why yes, yes I did. Stop pretending that due to the counteracting force of imaginary, unavaiable information, I can’t use real, available information. To assert such is Strawman City.
Here’s some logic for you:
P(x) = the credibility of claim x based on the person claiming it alone, assuming we know nothing about the person claiming it.
O(x) = the credibility of claim x based on the odds.
C(x) = the credibility of the claim based on both P(x) and O(x)
a and b are the claims.
Premises:
- P(a) = P(b)
- O(a) > O(b)
- all values in 1 and 2 are assumed to be > 0.
- C(x) = P(x) + O(x) :we assume credibilities are additive
- P(x) = P(y) & O(x) > O(y) = C(x) > C(y) :We know this from basic math, 3, and 4; in logic it must be stated explicitly.
Argument:
6) P(a) = P(b) & O(a) > O(b) :conjunction 1,2
7) C(a) > C(b) :modus ponens 5,6
And 7 is our conclusion. Note that you could alter 4 to use multiplication instead, or any combination of multiplication and addition and non-negative constant values, and 5, and therefore the argument proper, would be unaffected. It is, after all, a really, really, really, simple argument.
Now stop pretending that english text is an inherently ineffective way to convey ideas. Translating the english into symbolic logic is a pain in the ass and doesn’t accompish anything other than confusing the untrained.
If you want to get pedantic, I need the odds, the confidence you have in those odds being relevent to the truth; your chosen default credibility, and the confidence you have in that default credibility being relevent to the truth. Then I would invert the confidences, put them as the deniminators under the odds/credibilites, and add the two fractions together (over a common denominator, of course), to find the credibilities for each claimant, which can then be compared. Go ahead, have at. (Note that this procedure can be expanded to include any and all evidence you have available, that you can stick numbers to. Nifty, huh?)
And of course it matters wether or not they’re 100% credible. If they’re 100% credible you’re done. No guesses, no trials, nothing; their word is fact and neither statistics nor sperm samples nor notarized confessions nor videos of the event can possibly make any difference. (If you say much more stuff like this I may be forced to conclude that you’re not thinking clearly about this.)
Generally I use “you” in the general case, unless it’s painfully clear I’m addressing somebody in particular. I’m not making this argument just for your benefit, after all. That would be wasting my time.
And yes, I know you ignore the knowledge available from the statistics when you assess credibility. Whoop-dee-do. That doesn’t mean you should.
Actually actually what I’m using is the fact that probability denotes prevalence. (You had it the other way 'round). Denoting prevalence is, of course, what probabilities are for. Though, prevalence also does denote probability in the case when the probabilies are derived from sampled population data, which ours are. So that’s correct too. So lets hope those lions don’t lie down anytime soon.
Oh no!! :eek: In my flimsy-and-oh-so-inadequate defense, I wasn’t the first person to invoke the venerable Undead Emancipator; I was merely responding to the complaints of how people were rolling new hypotheticals for each new example; I picked one and stuck with it. (Sorry, Abe! I beg for forgiveness at your rotting feet!)
I cannot prevent people from doing bad statistics. The conclusion presumes you use the best of your ability to honestly determine what the real-world probilities are. If the person using this argument shares your disturbing willingness to juggle the numbers, then their conclusions can be attacked via attacking the stat manipuation in their premises, bypassing the inpenetrable perfection ( 
) of the argument proper entirely.
And as an aside, an arbitrary cutoff that varies via the observer is no cutoff at all. Therefore there is no meaningful distinction between a plausible and an implausible claim; the words mean nothing more than ‘decent odds’ and ‘poor odds’, with no division between the two; just one big continuous range of probabilities over which the same rules can be expected to apply. So, your ‘key word’ is meaningless, and all the related objections are void.
I think this covers everybody. Later, all.
Sorry about that, face; what I actually meant to say was “Then I would multiplicitavely invert the confidences, put them as the deniminators under the odds/credibilites, and add both the numerators and the denomerators of the two fractions together, to find the credibilities for each claimant, which can then be compared.” I’m sure you deduced all that already, but I wanted to be explicitly clear regardless. (Adding the fractions directly would result in an unfounded positive bias, just like multiplying them would result in an unfounded negative bias, as you surely realized.) Again, my bad and I apologise; I’m tired. G’night.
When Bricker posts that one claim is more credible than another, he/she is making a statement that is supposed to be “provably true.” If it’s not “provably true” then it is just merely an opinion. Opinions can be valuable and accurate quite often, but they are based on a mechanism in our brains that attempts to match the current set of data with past experience as well as some logical inferences. This mechanism is a very valuable part of our existence, but often it produces answers that are completely and utterly wrong.
If your position is that Bricker is really saying “some (possibly most) people will come to the conclusion that claim A is more credible than claim B”, then I can’t argue with that, it’s just a matter of opinion for all involved.
The alternative is that Bricker is making a statement about the mathematical probability of the 2 claims and making a comparison between them. In this case, it is not opinion, it is a conclusion based on math and logic. This implies that everytime a situation with the same statistics arises, the same credibility formula can be applied. Even if the statistics are different, the basic model can be used, according to Bricker. To be able to do this, Bricker or someone must show the relationship between the statistics provided and the claim credibility.
That still has not been done.
If someone asked you to calculate the probability that an event happened this year, you would take yearly statistics and use that to come up with your answer. This is what you have been doing in your calculations for the OP. That works fine for problems stated as I just worded it.
If someone asked you to calculate the probability that event 2 (a claim) regarding event 1 (rape) is accurate, you can’t ignore the fact that this is now a different problem. You can’t just say “well, lets look at data regarding event 1 only just like we did in the paragraph above where we used yearly statistics.” This is now a different problem. This problem is now concerned with the relationship between event 2 and event 1. Any analysis that ignores that fact has not modeled the problem properly.
The setup of the problem does not really model the problem we are trying to solve.
O(x) is the credibility of a claim based on odds - here you have just assumed the answer, that you can determine the credibility of a claim based on the odds of the event happening per year per US population.
If O(x) should really have been worded “the probability that event x will occur in 1 year in the US”, then you need to re-do your problem because that is a very different starting point (and I see other problems in the logic if that is where you start).
I’m not now and I haven’t ever pretended anything in this conversation. In my logic courses (symbolic and math), we spent several sections translating english text into logic. It’s tricky. Language is littered with ambiguity and our minds make assumptions about what is being said and we don’t make the same assumptions.
I did this type of translation into math in a post many pages ago. If you can find fault with it, I’m happy to reconsider my position. My problem is I don’t see how to get around that math.
I don’t see how the problem can be collapsed back to the type of problem you want to make it. It can only be the simple model if the question asked is different. Again, if the question is “what is the probability that this event will happen this year in the US?” then you can use your logic.
[/QUOTE]
My willingness to juggle the numbers? Those are actual numbers. I just chose to use a different set of actual numbers than you did, as I wanted a different outcome. Both sets are perfectly valid numbers, but the result varies drastically depending on how you want to break it down. If we use likelihood of a black woman to be a rape victim, you’re logic would tell us that she is far more credible than a white woman claiming the same thing. On the other hand, if we use the likelihood of a black woman being raped by a white guy, your logic tells us that she is far less credible than a white woman claiming the same thing. So, does your attempt to solve this mathematically not work, or do black women who claim white attackers lie at a vastly greater frequency than black women who claim black attackers?
Sure it is. Zombie Abe Lincoln, for instance? Not plausible. Black woman raped in North Carolina? Perfectly plausible.
So you’re saying that “decent odds” and “poor odds” are indistinct from one another? Hell, do you think there is a single person who thinks a black women being raped by a white man is implausible? Anyone? I didn’t think so.
Nope, but as shown above, your math is in this particular situation.
Unfortunately, I have to be very quick.
Raftpeople, I thought you were saying that I was required to show that a statistical argument could be used to prove guilt. Sorry about that. I think we’re on the same page about the probable part, except this:
Says you. Asserting this does not make it true.
It is highly, highly common for statistics to be applied to a different, later time frame than they were collected in, since it’s impossible to use them before they’re collected. In fact, other than toy problems whose only purpose is to teach statistics, making predictions is the only purpose of such statistics. If you want to deny that using statistics in a predictive manner is reasonable, then you’re the one making the outrageous claim, and so you’re the one who has to defend it. Good luck with that.
I have no problem requiring the statistics to have been gathered in a relatively recent year, the same span of time, and the same country. Since we’re compairing two claims that are almost certian to have occured in the same year and country anyway, I would expect the exactly the same data to be used with your restriction, in exactly the same way, as we would have been doing it without the restriction.
Sorry, but I can’t see how your restriction makes any difference to the results. Could you explain that again?
Which we? I’m fielding at least three opposing positions at once here. I just picked one that seemed feasable to do.
I actually tried to address your question first, but your position requires actual math, and after an hour and a half it became clear that I was trying to rewrite math up through the third grade level in terms of formal logic, which is possible buy would take many hours and turn out to be hundreds of lines long. For a forum post. Hell no. Among other problems, you aren’t paying me enough.
And that’s the problem with writing all one’s debates in formal logic. It’s not that it’s impossible; it just takes forever, produces volumes of text, is hard to read, and doesn’t inform most of the audience. Ergo: not a good idea.
Sorry; gotta run. I’ll try to answer you sometime later this evening, DMC. Bye!
Okay, again with emphasis mine: “Those are actual numbers. I just chose to use a different set of actual numbers than you did, as I wanted a different outcome.” I believe that the word ‘as’ in this context means ‘because’, correct? You chose the numbers you did because you wanted a specific outcome. When a person chooses their outcome before they choose their numbers, then they are not using statistics; they are abusing statistics. And, as I said, persons who are juggling the numbers in this manner are subject to having their conclusions attacked via attacks on their statistical method, without there being any problem at all with my argument.
The nifty thing about this, you see, is that I don’t tell you where to get the (rapes/population) odds. We’re assumed to have them by premise. So I don’t care how much you attack the methods of calculating the odds; that’s not my problem. You can get them off the back of a cereal box for all I care. My only point is, that if you have some odds, you can use them as information, and they are not overwhelmed by the mere presence of a default assumption that, the two claims are presumed to be equally credible, absent other information.
The important thing to notice here is the condition, “absent other information”. Clearly, if the odds are information, then they go straight through the loophole and may modify the claimants credibility, in just the same manner that a videotape of the event in question is allowed to modify the credibilty of the claim, initial assumptions of equality notwithstanding.
Of course, the videotape is certainly more influential as evidence than the odds are, unless of course the odds are much, much, much longer than we can possibly expect. And this leads neatly into the second half of your post: is there some magic level of credibility that divides the plausible from the implausible, which allows us to have a different set of rules for the lesser and the more probable ranges of probability?
Okay. As you surely know, probabilities are noted as different points on a number line; a number line is a continuous realm, where the numbers gradually increase at a constant rate, which is the same at every point, be it long odds or short. The numbers itself have no break in them.
For an analogy, consider a long chart showing a fade from white at one end, through all the shades of grey, to black at the other end. The fade is continous; there are no clear breaks or sharp changes in color.
To me, your delaration that ‘plausible’ and ‘implausible’ are two separate and distinct things sounds exactly like if you were to declare "All the shades of grey are either ‘white’ or ‘black’. And then you point at the one end of the chart, where all she shades are white, and say “This is white”, and the other end of the chart, and say “This is black.” This all sounds very reasonable, since all the shades at the one end are all very white, and all the shades at the other are very black. If you look at one part and then the other, there is a very clear and obvious difference.
But then I step back and look at the entire chart all at once; the whites fade very slowly and gradually into the blacks, so that I can see no place on the chart where one shade is discernibly different than the one right next to it. There are no breaks. If one shade is white, then so is the next; they can’t be told apart. And since that shade is white, so is the one after it; they also are so similar as not to be different. I can keep sliding down the chart in this manner, until I run out of chart, and am in the area you call ‘black’, having never found an obvious place where the color changed from being white.
I ask you about this. “Where is the magic point, where white becomes black?” You respond to me, “There is no such point. It just is black at one end and white at the other.” This is hard to argue with, because the ends are indeed clearly different colors; but it entirely misses the point. Black and white are different in the extremes, but in reality, they are all part of one continuous shading of grey. There is no break.
Probabilities are the same way. You’re pointing to probabilities at either end and labeling them different things, and claiming (are at least strongly implying) that there’s some fundamental difference between the two and that they should therefore be treated differently. But I see that there is no distinction between the two; all probabilities are just points on one single number line. You have already said that there is no fixed point on this number line that divides probable from improbable; ergo; there is no difference.
As I have laid out: yes, “decent odds” and “poor odds” are not distinct from one another, they are continuous with one another, with the only difference being wether there is a lot of improbability, or only a teeny weeny bit. And that teeny weeny bit is treated the same as when there is a lot, just with less impact because there is less of it.
If severely implausible claims may be dismissed out of hand, there must be a reason this is allowed. That reason is because there is a relationship between the amount of improbability and the amount of crediblity inherent in the claim; the more improbable the claim, the less credible it is.
I will freely concede that I, at least, think that there is relatively little inherent implausibility in the claim that a particular black woman was raped. But there is some; after all, the claim that a particular black woman was at one time an infant is even more plausible yet. So, the claim of having been raped is not perfectly plausible. It has at least a few grains of implausiblity in it. A tenth of an iota, perhaps. And that tenth of an iota reduces the inherent credibility of the claim proportionally.
When you have two different claims with differing amounts of inherent implausiblity, then the one with better odds is, slightly, at least in terms of iotas, more credible. Perhaps not enough to notice and, in the case of various rape claims, almost certainly not enough to sway the decision, based on the difference in the odds alone. But the difference doesn’t magically vanish into the ether just because it’s small; it’s still there. The credibility of the claims are therefore not, quite, equal.
Oh, and face? On reflection, it looks like that equation I laid out for you at 2am last night was a bit wonky. I sort of knew it at the time, and attempted to post a correction, which helped a little but was itself afflicted with “2:30am disease”. If I ever claimed to be perfect, you are all free to point at that and call me a liar now.
Just do a weighted average on all the sources of credibility info, with the weights being the confidences you have in each source (naturally enough). That should get you a reasonable result, I think.
This is the only thing I’m going to respond to, because the rest of what you wrote quite frankly basically amounts to a whole lot of nothing.
Prevalence stats are only predictive if you have enough numbers to determine trends over a period of time. One set of figures from one year have no predictive value because in the absence of a reference point (such as an average), you have no way of knowing how well those number represent what is normal for the population. This is another reason why your constant talk about odds is grating to my ears. The “odds” of being raped apparently must vary from year to year, since prevalence figures vary from year to year. But if a rape allegedly occurs in 2005, it makes no sense to apply prevalence data collected in 2003, because for all we know the incidence of rape in 2005 may be twice that of 2003. There is no scientifically sound reason to assume that the “odds” of being raped in 2003 is anything like the “odds” of being raped in 2005, unless your goal is to introduce garbage into your analysis just for the sake of producing garbage.
The more you post, the more I realize that you have no experience working with biostatistical data. I’m not saying this to be mean, either.
We both chose a set of numbers to use, both of which are accurate. You chose a subset of rape claimants that gives less credibility (in your opinion) to the black claimant. I picked a subset of rape claimants that showed the exact opposite when using your logic. Both sets of numbers are accurate. Neither tell us diddly squat about the credibility of the claimant. If they do, then you can answer the question in my post that you spent several hundred words not answering. It should be noted that using your method to determine credibility makes it quite easy to know what the result will be upfront, well before any calculations are performed. Is that why you chose the subset you happened to choose, and if not, what method did you use to pick the subset to work with?
What’s even niftier is that I can take two equally accurate pieces of information and come up with exactly the opposite results from one another. My only point is that this proves that your method of determining credibility is quite bogus.
Not important at all. We know both the rate of white on black rape, and black rape victims in general. Choosing one over the other to get your result is no different than flipping a coin.
There don’t need to be clear breaks. Some of the points on the line would be black, some would be white, and some would be grey. There are plenty of points on the line in which we’d actually all agree that we’re looking at black, white, or grey respectively. The points we all agree are black don’t become white just because you’d like for them to.
Either way, that’s a pretty shitty analogy, unless you can find someone who thinks that a white man raping a black woman is implausible. I asked you that earlier, but that also didn’t get answered. If not, then that line goes from black to very dark grey, in which case you’re going to find it hard to find a spot that anyone would consider white.
No offense taken, because I don’t have any experience working with biostatistical data. I don’t even feel all that unfulfilled for the lack of it.
You want to know something, though? I don’t see how pointing out the flimisiness of the data damages my position at all. After all, I already told you where to put the flimsiness of the data: in the denominator of that weighted average. You can give the limited, unstatisfying single statistical sample a nice low weight, as compared to a nice high weight for the ‘assumed equality’ you’re crediting the claimiants with. If you make the difference nice and big, like, oh, say, a factor of two-thirds of a million, then those unimpressive statistics will be almost entirely overswamped by the assumed equality, and certainly overswamped by the first persuasive piece of evidence that comes along.
Why did I pick the ratio 1:666666? Why, because those are the weights that I casually used in my statement of the equation back in post #548*.
You remember that; that’s when I came back from my brief hiatus, with the sole intent of supporting why the presence of a statistical claim that presents a difference in odds that has some nonzero credibility, when added to an assumed equality of odds, does present a difference in resulting credibility, even though it’s not enough of a difference to cause anybody to treat them differently. And I only bothered with doing that because it seemed so sad to see two or three of you going at the lone magellan01 like a pack of starving crazed weasels.
Regardless, I’ll tell you right now, I’m not convinced that a nearly 2:1 difference in prevalence in one sample means nothing. That’s a 16 percent deviation from even odds. For that to be inside your standard error from even odds, your range of error would have to be nearly a third as wide as the total possible range! That seems extreme. Therefore, what that nearly 2:1 difference tells me is not that it was twice as likely that the wow rape occured; it tells me that there is probably a trend towards more wow rapes than wob rapes. Dunno how big a trend, but ‘equally prevalent’ seems unlikely. This is not as informative as if we knew there actually was going to be a 2:1 difference in the number of wow and wob rapes this year. But it’s not nothing.
And, of course, as I mentioned before, I have already provided you a way to account for facts with a low persuasion factor: you give them a lower weight. In the weighted average. Not hard.
(I repeat myself solely to give you the chance to say that my posts are ‘content light’ again. I find it funny every time you say that.)
If there is any weight at all to the data indicating a difference in the probability of the event, you take that + the fact that increased odds -> increased credibility (I’ll get to you in a moment, DMC), and then you can pour the rest into that logic equation that DMC is ignoring:
if a < b and c = d, then a+c < b+d; which is to say, unequal credibilities.
This is of course true regardless of how huge c and d are, and how infintessimally tiny a and b are.
So, it’s not good enough for you to point out that a 2:1 in one year does not mean a 2:1 a few years later. You have to prove that it is no indication of a trend to one side at all. Otherwise it doesn’t make a lick of difference to my position.
Good luck!
(* = that’s assumming that your level of assumed credibility is 50%, which seems the be best we can sensibly assume, given that there is almost certainly another person that we know nothing about except race and gender, who would be making the exact opposite claim: the defendant.)
The method I used to choose my data was as follows: In every single example and discussion that I have presented about this topic, I have not assumed that I had the relative population data available to me. Which as you know, being the seasoned logician you are, is exactly the same as assuming that that data is not available to me.
So, when you handed me this data that I was axiomatically assumed not to have, I was in a bit of a bind. Like you or anyone else, I like to use all the data I have available to me. But I have been axiomatically assuming that that data is not available to me. I won’t let you bring in data that we are assumed not to have, and I can’t either (see, it’s fair that way) so I used only the information I had been assuming I had all along: the rate of rape occurences relative to the total population.
And that’s the sordid tale of how it happened. Any assumptions you had about about me cherry picking my data are, of course, wrong.
You do realize that I don’t give a tinker’s damn which probability turns out to be greater. You’d have to be entirely composed of cheese to think that the teeny little impact that these odds would have of the claimants credibility would stand up for one second against any direct evidence. You’d have to be genetically spawned from tomatoes to think that anybody intelligent enough to actually understand my argument (which clearly isn’t everybody) would use it to do anything but downplay the weight of these flimsy stats as compared to conventional (court-admissible) evidence. And you’d have to have a substantial number of similarities with three-week old mustard and be dishonest to try and use this as a reason to discredit somebody of any race.
So, since I obviously think myself unrelated to sandwhich fixings, why would I be arguing that a piece of information that makes a very, very, very, very weak statement about a difference in credibility between two claims, makes a difference in the credibility of the two claims (albeit a very, very, very, very weak difference)?
Because it’s effing true. No matter how many people shout otherwise.
I’m not sure what you mean by equally accurate; I would be inclined to give more weight to the per-racial-population statistics, myself, if they were available. Not that your conclusion would follows from your point even if it were perfectly applicable to the situation at hand.
So anyway. So this is your “only” point, huh? That "[the fact “that [a person] can take two equally accurate pieces of information and come up with exactly the opposite results from one another] proves that [my] method of determining credibility [taking into account all available information, keeping in mind the possibly low weight of relevence that credibility has] is quite bogus.”?
If that’s your only point, you’re done. Having two pieces of information from which you can draw opposing conclusions happens all the time; if it didn’t every court case would be open-and-shut. In fact, every court case has two equally credible pieces of opposing information: the first two things you have are claim one (“He did it!”) and claim two (“I didn’t!”). According to your assertion, the mere presence of those two claims renders the entire subsequent process of truth finding “bogus”.
And do I need to explain that when two peices of information with different weights oppose one another you get something like having one train engine pulling one way on one side of a box car and two engines pulling the other way on the other side: a result of there being a pull in the direction of the more persuasive evidence but not as much as there would have been without the opposing argument? Do I need to explain that to you? No? Good. It is obvious, after all.
Two points:
- In this argument we don’t know the rate of “black victims in general”; we are assumed not to have any information on bob rape at all. What the hell are you talking about?
and more importantly 2) You do realise that your last sentence there has nothing to do with, and is not supported by, the sentenses before it. Even if they weren’t 1) a baseless assertion, and 2) wrong, respectively. You do realise that, right?
I write you seven hundred eighty five words, slowly and carefully pointing out what I’m talking about as I would to a child, and you still manage to completely miss the point. (Hey face, guess why I find it funny when you attempt to ad hominem me by saying that my long, carefully crafted posts amount to “a whole lot of nothing”. 
)
Okay, here’s the point, O Proud, Noble Dweller of Midbuckheadtown. Let’s see if I can make this really obvious. The reason you can’t draw a big 'ole line between where the probable stops and the improbable ends, is because everything that you can think of as being ‘probable’ has a little bit of improbability in it. Grey is nothing more than black with a little white in it; ‘probable’ is nothing more than ‘certain’ with a little improbability in it.
And improbability always acts the same way.
The more of it there is, the more non-credible the assertion is.
Even if there’s only a little.
Let’s see, did you just admit a difference as great as between black and dark grey? Yes, yes you did. Ergo you concede a little difference in the amount of white (aka the amount of inherent claim credibility) in the two claims, ergo my point is made. At least to people with the wherewithal to extract what I was talking about from my, quote, “shitty analogy”, unquote. 
All of that and you still didn’t answer the question. I’ll ask it again.
I went googling for FBI stats and found a number of major cities reports of crime reports, unfounded claims and “real” claims. I found out a couple interesting things:
-
Crimes reported more often (e.g. theft) tended to have a higher rate of unfounded claims than crimes reported less often. Note: unfounded is police terminology for “false”, not false as in there was no conviction but false as in the claim itself is contradicted or not supported by other evidence.
-
Statistics reported to the FBI are handled so differently by each city, despite a standardized approach they are supposed to follow, that any kind of inference based on these stats would be pretty suspect (pun intended).
Clearly logic, reason and statistics all elude begbert2. He is adept at the prickish attack, on the other hand. It’s nice to see that he is willing to twist all logic into a pretzel in order to defend the honor of poor magellan01 from the pack of crazed weasels. Crazed weasels who are also incapable of comprehending begbert2’s intellect. I do like how he has stumbled into this idea of applying fantasy weighting strategies to make up for misapplying subpopulation probability estimates.
It’s interesting how this issue seems to pull in folks like this clown, bored mathematician, and psychloan, all with only a little knowledge, but a lot of “I’m so smart you can’t comprehend me!”
Where did magellan01 go off to, by the way? Was my post really so scalding that my apology was insufficient? He was the only one arguing from the other side who was not trying to call upon the authority of his own brilliance to augment his argument.
I mentioned Condi and that messed his shit all up.