Iain, you point out that “cluster sampling” of the sort done in the survey purporting to demonstrate 650,000 deaths in Iraq is a recognized part of medical research. I am not a demographer nor a statistician nor a medical person, but isn’t “cluster sampling” a means of extrapolating phenomena occurring in nature like disease patterns and the like? Surely deaths as a result of war and terror are not “natural” phenomena that can be measured through population extrapolation.