As far as I can see (and I have had a surprisingly productive argument about this at the Harry's Place blog, once people stopped chucking the "fradulent" accusations around), people who want to criticise this study are going to have much better luck trying to attack the data itself than the methodology. The methodology is sound and you're not really going to find anyone to attack it.
The survey itself, however, was a lot more difficult to carry out than in 2004 (because Iraq is a much more dangerous place than it was in 2004). The survey team refused point blank to carry GPS units, so the method of geographically selecting clusters is a new one that hasn't been field-tested. I would repeat, without much hope, my plea that people should not obsess on the point estimate of 650k excess deaths, but more on the very high degree of confidence with which we can say a) things have got worse, not better and b) they have got a lot worse, not a little bit worse.
I'd just emphasise that I'm not actually questioning the 655k figure or its confidence interval, which are the best estimates you can get from the data - as far as I can see, the map-based method is equally as likely to result in an underestimate as an overestimate compared to GPS-grid or pen-toss methods - but I'd put less "weight" (in the Keynesian sense) on this number than on the qualitative conclusion.
[the central point that everyone needs to focus on is the question "if the true number of violent deaths in Iraq was 60,000, how likely is it that there would be 300 violent deaths in a sample of 12,000?" and the answer "incredibly, laughably unlikely"]
I've written a bit about the current two favourite arguments bouncing round the nuttersphere - that the pre-invasion death rate is "too low", and that the data ought to agree with (nonexistent) centrally collated death statistics - which are just wretched in quality.
http://crookedtimber.org/2006/10/12/death-rates-and-death-certificates/
best dd
___________________________________________________________ Try the all-new Yahoo! Mail. "The New Version is radically easier to use" – The Wall Street Journal http://uk.docs.yahoo.com/nowyoucan.html