Re:Smear campaign against Wikileaks begins
by cabalamat on Saturday August 21, 2010 @07:43AM
(#1622634)
Oops my analysis was wrong -- there are two rape cases.
Probability of two unrelated rape cases coming to light just at a time convenient for the US authorities -- about as likely as me getting pocket aces in every hand during a poker evening.
The probabilities of a man committing each of two individual rapes are not independent. This is the worst statistical fallacy you can commit. Please see the following for further tragic details:
http://en.wikipedia.org/wiki/Roy_Meadow#Cot_death_trial_cont...
Re:Smear campaign against Wikileaks begins
by cabalamat on Sunday August 22, 2010 @11:15AM
(#1624718)
> The probabilities of a man committing each of two individual rapes are not independent.
Indeed not, and I never said they were. Indeed if a rapist rapes one woman, then he's more likely to do it again than a man who has never raped.
My point was that it's less likely that a man will be accused of two unrelated rapes at the same time, than just be accused of one rape.
(1) an average man is accused of rape (say in a year)
(2) a person who a government would very much like to bring down is the subject of made-up allegations
Regarding (1), i don't have figures for Sweden but in the UK there are c. 14k reported rapes a year versus 25M adult males, so P = 14k/25M = 0.00056
Regarding (2), I don't have figures but suggest off the top of my head that 0.01 is reasonable.
0.00056/(0.00056+0.01) = 0.053
So Bayes' theorem suggests a roughly 5% probability that the accusations are real and 95% that they are concocted by the US or some other government.