The Lancet study, as Kaplan correctly observes, basically stands for the proposition that the number civilian causalties during the survey period is unlikely to be less than 8000, or more than 194,000. Kaplan characterises this finding as “meaningless”, which observation he bases purely on the width of the confidence interval.
He then goes on to refer to the Iraq Body Count, saying
The IBC estimates that between 14,181 and 16,312 Iraqi civilians have died as a result of the war—about half of them since the battlefield phase of the war ended last May. The group also notes that these figures are probably on the low side, since some deaths must have taken place outside the media’s purview.
The IBC’s finding then, is that the number of civilian casualties is unlikely to be less than 14,000 and quite probably much more. It gives no upper limit. Kaplan is full of praise for the IBC and considers its results to be a sound basis upon which to estimate the casualties. He suggests a figure in the range of 20,000 to 30,000.
I agree entirely with Kaplan’s fullsome praise of the IBC. I also agree that a lower bound of 8,000 is less informative than one of 14,000, but I fail to see why the range [14,000-infinity] should be any more or less “meaningful” than [8,000-194,000].
I also agree with him that the Lancet’s central figure of 98,000 should be taken with a pinch of salt, but it does at least have the virtue of being a statistical calculation. Kaplan would have us eschew this figure in favour of a range which has no basis that I can see other than that he personally find it plausible.
But the true drawdropping irony in all of this is that Adam, apparently with a straight face, should cite this claptrap in a post exhorting the rest of us to higher standards of evidence!