Thursday, October 12, 2006

My Name is Bush, Devourer of Worlds

The Lancet has published a mortality study of excess deaths in Iraq since the invasion (lead author Gilbert Burnham). The excess violent death toll is around 600,000. This in a country smaller than Canada.

Any humour here will be grim, but Lindsay Beyerstein has a rundown of innumerate right-wing commentary here. Bush himself claims to be an expert epidemiologist.

The takehome points:

*The fact that you do not want to believe something is not an argument against it.

*It is equally likely that the sample size led to an understatement of the actual number of deaths in the population as an overstatement.

*In particular, there is a 2.5% chance that the actual number of deaths exceeds the confidence interval, just as there is a 2.5% chance that it is beneath the confidence interval.

There is no such thing as the perfect study, and all empirical work is subject to reasonable criticism. So far, the right-wing blogosphere has yet to exceed the level of the Holocaust deniers.

Update: More important than the right-blogosphere is the American national media. Tim Lambert has been covering its reaction (or lack thereof). Basically, the New York Times and the Washington Post have been burying it, while Associate Press claims the study is "controversial" on the basis of the word of a "national security analyst" with no apparent expertise in statistics or epidemiology.

Update 2: Steve Sailer points out that most of the methodological difficulties with a study like this derive from the fact that Iraq is "hideously dangerous", which is hardly a point in favour of the Iraq optimists. He also has an interesting calculation of 1 million American bullets fired in anger per day in Iraq. A lot of those bullets had somebody's name on them.

Update 3: Daniel Davies (D-squared) has the definitive take here.

Update 4Seyad at Healing Iraq provides an Iraqi perspective on the Lancet study here.

No comments: