Bayesian Probabilities

hans ze beeman at LGF provides some mathematical calculations using Bayes' theorem.

He shows, using rough prior probabilities, that it's sixteen hundred times more likely that you are a terrorist if you are a muslim than if you are a Christian!

That's the math of rational profiling (I've corrected a factor of ten in the quotation below from the original text):
Screen all persons that appear suspect. And yes, unfortunately, most terrorists these days are Muslims.

/Starting statistical reasoning module (using Bayes' theorem)

Priors (rounded):
P(Muslim) = .20 (Probability that you are a Muslim, regarding world population)
p(Christian) = .40
P(Buddhist) = .10
P(terrorist) = .0001

Conditional probabilities:
P(Buddhist ¦ terrorist) = .001 (Probability that you are a Buddhist given you are a terrorist, regarding world population)
P(Christian ¦ terrorist) = .001
P(Muslim ¦ terrorist) = .80

--

P(terrorist ¦ religion) = [P(religion ¦ terrorist) x P(terrorist)] / P(religion)

(Probability you have a certain religion, given you are a terrorist)

--

P(terrorist ¦ Muslim) = .80 x .0001 / .20 = .0004
P(terrorist ¦ Christian) = .001 x .0001 / .40 = .00000025
P(terrorist ¦ Buddhist) = .001 x .0001 / .10 = .000001

--

P(terrorist ¦ Muslim) / P(terrorist ¦ Christian) = 1600

/end statistical reasoning module


Using the mentioned priors, the probability that you are a terrorist if you are a Muslim is 1600 times higher than if you are a Christian. The priors are assumptions in this case (e.g., 80% of terrorists are Muslim); it could be easily researched how many terrorists were Muslim, Christian or Buddhist (same goes for religious affiliation). If someone has the data, it could be recalculated.

Just think about it. And profiling.