...apart from all the bits of Superfreakonomics that have been debunked, that is. I believe the proportion of error is somewhere in the region of "every damned thing they wrote".