Thursday, May 05, 2005

Dr. Les Roberts Interview

The eminent Dr. Les Roberts, one of the authors on the Lancet Study of mortality in Iraq (which I have previously discussed here) has apparently given an interview to a propaganda site called Uruknet. I guess the nature of the site was insufficiently outrageous to put him off.

You may remember that the study concluded that the coalition invasion of Iraq led to 100,000 excess deaths. This has been widely and incorrectly cited as "civilian deaths" or "women and children". Roberts repeats the claim that most of the victims were women and children even though, based on his own numbers, this is a very misleading conclusion, and in no way validated by the statistical strength of his study. The study is not, in general, deceitful, although it has numerous problems, but I find the evidence of bias and political motivation so palpable in this interview that Roberts should be embarrassed by it.

I'll try not to fisk it, but I'll just pick out the assertions that are personally annoying to me.

... I thought it would serve the interests of Iraqis if I described what they were really dying of. So, if we found they were dying of diarrhoea we could do something about that.

If they were dying at home in childbirth because they were too scared to go to hospital, we could do something about that. Much to our surprise we found that these things weren’t what they were dying of. Most were dying violent deaths....

Isn't this exactly what we hoped for? The invasion has not substantially increased background deaths. No disease. No starvation, no general health issues. Why does it sound so negative in this context?

...I was smuggled across the border into Iraq. I went with just a suitcase and $20,000 in my pocket. All it took was six Iraqis brave enough to do the survey...

The study was not thusly characterized in the Lancet article. It sounds like he's trying to make himself seem a hero, which he may well be, but it certainly leads to more questions about the methodology.

... Most of the people killed by the coalition were women and children, which implies the use of a lot of force, and perhaps too much...

This really annoys me because there were very few women listed as killed in the study. It is true that when you combine women and children in his study sample, the total within the sample is slightly greater than half. But why would you combine women and children except for the polemical punch of the phrase. This phrasing implies that the dead were civilians. Dr. Roberts will no doubt know more than I, perhaps more than he reports, but I see no evidence that the children killed were not serving as combatants. His group definition includes children up to age 15.

Please note in particular that it is statistically verboten to dissect the sample results into groups without a group significance test. This test was not provided in the study. There have, moreover, been reports of adult combatants using children as cover. I have not heard the same regarding grown women. We can state from the Lancet numbers, in fact, that by far most of the people killed in the sample were "men and children", and we don't know whether these children were boys or girls, nor do we know the age distribution of the children killed.

...Americans are so hated that I couldn’t go around talking to people. We would pick a random point in each "cluster" — each village or town we surveyed...

... I only went out with the interviewers for the first eight days. On the eighth day the police picked up our interviewers while I was in the car watching and that was a pretty bad experience.

After that we were convinced that interviewers knew what they were doing, and they didn’t want me there. For about 15 days I just stayed in a hotel roomand didn’t go out.

These descriptions make it pretty hard to believe that the interview process was properly controlled.


Roberts does give some very interesting quotes about other studies that I'll have to look into. In particular, he talks about a New England Journal of Medicine study from July, 2004, that surveys returning soldiers. He implies that the survey was statistically reliable for the purpose of estimating civilian casualties, and states that the numbers were comparable to his.

5/5/2005 10:06 PM

Links to this post

Links to this post:

Create a Link


Post a Comment

<< Home