Here's a great post from William Easterly in which he points out a fundamental statistical mistake that a lot of opponents of Park51--aka "Ground Zero Mosque"--and similar projects are making.

In short, we're getting our conditional probabilities wrong. A lot of terrorists are Muslims. This does not mean that a lot of Muslims are terrorists. Consider a similar example. Rhodes is now in possession of an interesting collection of papers from a former KKK leader in Tennessee. The papers are mostly from the late 1960s and early 1970s. If I remember correctly, the gentleman in charge was also a prominent member of his church (director of the Sunday School program, I believe). Their local chapter paperwork included a line item for the number of ministers enrolled. If you took a census of KKK members at virtually any time in the organization's history, it's a fair guess that very high percentage would identify themselves as Christians. The probability that one identifies as a Christian given that he or she is a member of the KKK is very high. It would absurd to infer from this that the probability that one is a member of the KKK given that he or she identifies as a Christian is also very high.

I've been following the mosque controversy and participating in the discussion because I view it as a teachable moment (I have a Forbes piece or two or three in the pipeline). Not only does it teach us about institutions, history, theology, and economics, it's a very useful lesson in statistics. I suspect that p(terrorist|Muslim), just like p(Klansman|Christian), is pretty low.