Creative Commons License
This weblog is licensed under a Creative Commons License.
Powered by
Movable Type 4.25

« Web hosts get in on the Russia-Georgia fight | Main | Paranoid conspiracy theories: not an American monopoly »

More big numbers in Iraq

Update: The report is now available online

How credible is the study about to appear in the Lancet, estimating 655,000 excess deaths in Iraq as a result of the war?

All this is based on the media coverage I've seen (Wall Street Journal, WaPo, NYT). I haven't seen the report, so I can't say it is trustworthy. All we can say for now is that it is consistent with other figures, and using an appropriate methodology. First, the plausibility. Yes, 600,000 is a very big number. It is about 2.5% of the population of Iraq. But remember that this isn't anywhere close to saying that 600,000 people were directly killed by American soldiers. It is just that the overall death rate has increased massively - that might include inadequate healthcare or nutrition, more traffic accidents, whatever. It certainly includes the violent crime, which we know there is a huge amount of. Granted, it is at the high end of the scale, and I'll want to look at the methodology in detail before I say that I believe it. This is not inconsistent with other accounts. In particular, it isn't disproved by the fact that Iraq Body Count, give a much lower death toll, between 43,850 and 48,693 deaths. It's because they are counting different things. Iraq Body Count simply totals up the civilians reported in the media as having been killed. By their own admission this is will always be an undercount:

We have always recognised and made explicit that our media-derived database cannot be a complete record of civilians killed in violence, and have called forproperly supported counts since the beginning of our own project. What IBC continues to provide is an irrefutable baseline of certain and undeniable deaths based on the solidity of our sources and the conservativeness of our methodology.

The figures are higher than the death counts based on bodies in morgues. These generally relate only to violent deaths (narrower than this study), and count about 100 a day. Juan Cole doesn't find this discrepancy too large to deal with:

First of all, Iraqi Muslims don't believe in embalming or open casket funerals days later. They believe that the body should be buried by sunset the day of death, in a plain wooden box. So there is no reason to expect them to take the body to the morgue. Although there are benefits to registering with the government for a death certificate, there are also disadvantages. Many families who have had someone killed believe that the government or the Americans were involved, and will have wanted to avoid drawing further attention to themselves by filling out state forms and giving their address. Personally, I believe very large numbers of Iraqi families quietly bury their dead without telling the government of all people anything about it. Another large number of those killed is dumped in the Tigris river by their killers. A fisherman on the Tigris looking for lunch recently caught the corpse of a woman. The only remarkable thing about it is that he let it be known to the newspapers. I'm sure the Tigris fishermen throw back unwanted corpses every day.

I'm not entirely convinced by Juan Cole's line of argument here, simply because people generally were able to produce death certificates:

When people reported deaths, researchers asked them about the cause and obtained death certificates in 92 percent of cases (Baltimore Sun)

And at the Washington Monthly blog, Kevin Drum adds:

This time around, the figures from their new study buttress the previous one, and also match up with other data, which suggests their methodology is on target.

How reliable is their methodology? Not perfect, but better than anything else available. As far as I can see, the methodology is the same as what they used back in 2004 - see a collection of defences of it here. Probably the biggest criticism of the 2004 report was the small sample size. But now, as Rubicon says:

For statistical purposes, the sample size is very large, much larger (for example) than typical national voter polls in the US, which sample about 1,000 to 1,200 individual respondents. If we presume 4 persons per Iraqi household, the sample size is over 7,000 persons—in a country one-twelfth the population of the US. The data-gathering and estimation techniques are quite reliable; according to one of the lead researchers, Gilbert Burnham of Johns Hopkins, "This is a standard methodology that the U.S. government and others have encouraged groups to use in developing countries."

One likely methodological problem is 'recall bias' - that is, the possibility people will have forgotten deaths that happened in the past. This would decrease the figures for pre-war mortality compared to post-war mortality, and so give an inflated count of excess deaths. The issue was raised with the 2004 study, and the longer timescale of the latest report makes it an even bigger issue.

Note: I am updating and amending this entry as I find out more about the study. I haven't yet made up my mind on how much I believe it - and in any case, I still haven't seen the report.

Blogs defending the study: Amptoons, mahablog, Barista, Deltoid (not much yet, but likely will have more in time)

Blogs arguing against it (only the ones I think have halfway-decent arguments): Jay Reding. No doubt there are more decent arguments against this, but I've not been bumping into them much.


That's pretty terrible (even if numbers are exaggerated) - *makes angry noises*. Glad you're posting about this stuff, though!


Glad to see recordings of this.