LANCET AND FACE VALIDITY. In a conversation yesterday, a friend told me that he just couldn't accept the Lancet numbers on Iraqi deaths because they seem too high. The study, he felt, had to have cooked the books in some way, even if the method wasn't immediately apparent. My friend's comments echoed those of Fred Kaplan and Michael O'Hanlon, sensible people who have chosen to reject the study for not terribly compelling reasons. It doesn't really matter, the argument goes, how many people have been killed in Iraq as long as we know that it's a lot and that it's getting worse, but it still can't be that many. I have a little bit of empathy for my friend's position. My gut tells me that the number is too high, in an eerily similar way to which my gut tells me that Kenny Rogers' post-season ERA is too low. If someone had forced me to make a guess prior to the release of the study, I probably would have picked a number between 150,000 and 200,000, or maybe a little higher. My guess would be a little bit better than just tossing a dart at the board, as the things I know about the history of military conflict and the nature of the state would guide my estimate. And that's it; my expertise would result in my picking of a round number that I kinda felt was right, without having all that much reason for confidence in my estimate. Herein lies the problem, because despite all estimates, gut feelings, and reasonable expectations, Kenny Rogers really hasn't given up a run in 23 post-season innings. Major League Baseball has a better way of calculating earned run average than my gut, and MLB says that his ERA is 0.00. Similarly, the Lancet study uses considerably more sophisticated statistical and survey methodology than my gut, and is consequently more likely to be accurate. Because methodology is complicated and only professionals really understand it, we have to rely to some extent on arguments between experts to assess the validity of the approach. I'm not an ideal judge, but the defenses of the study seem considerably stronger than the attacks upon it. Kenny Rogers' success is unlikely, but not implausible. Weird things happen in the post-season, and such events can't be dismissed on their face. Awful things happen in war, and Iraq is currently enduring a truly awful war. 655,000 is high, but it's a plausible number that can't be dismissed on its face. More importantly, it's a number that has resulted from the most sophisticated study of Iraqi casualties available. Consequently, it's really not enough to just kind of feel that the number is too high; the Lancet study has shifted the burden of evidence. If scholars and analysts don't like the number, then they need to either soundly debunk the methodology (which hasn't happened) or do a better study.
--Robert Farley
