ParaPundit : November 2004 Archives

The Hundred Thousand:

I never thought the Lancet survey of civilian casualties in Iraq would take off. For one thing, Lancet is a peer-reviewed academic journal, which publications are only rarely read outside the community they intend to serve. For another, the survey was so badly and obviously flawed that nobody could take it seriously. Surely, in the next issue the editors will correct what was a tremendously bad job.

However, perhaps because there are so many people eager to believe what it purports to say, I've started to see "the US has killed 100,000 civilians in Iraq" popping up on discussion boards everywhere.

ParaPundit ably examined the problems with the survey when it came out earlier this month. ParaPundit is very much anti-war: in fact, he is far and away the most convincing anti-war voice in the blogosphere, because he argues out of an understanding of the military science. The issues he tends to raise are the real and severe problems we have to overcome if there will be success in Iraq. I don't share his pessimism about our prospects, but I respect his knowledge and recognize that he points to real concerns.

Still, being of a serious mind, he hates bad argumentation from his own side even more than from his opponents. This is often true of people who are at the top of their game: they recognize that hackery and bad-faith argumentation from their side weaken their argument. Even though they agree with the conclusion, they'll still dispute the method.

As ParaPundit demonstrates, there are two serious problems with the Lancet piece:

1) The survey's "Confidence Interval" runs from 8,000 to 194,000 dead. The "100,000 dead" estimate is merely the middle of that range. However, the confidence interval means that 8,000 is just as likely as 100,000 to be the real number; and that 160,000 or 194,000 are also just as likely. That makes this, as one person sneered, "not an estimate but a dartboard."

2) The estimates that push this out of the 8,000 range and into the 100,000+ range come from hospitals in areas that were insurgent-controlled during the survey. The estimates out of other parts of Iraq, even those which have seen heavy fighting, are much, much lower.

It was standard practice in Ba'athist Iraq to inflate vastly the numbers of starvations resulting from the UN sanctions, and to invent out of whole cloth mass disease and other ills supposedly the result of those same sanctions. This is just business as usual.