The Lancet study claiming that since the US-led invasion some 100,000 Iraqi civilians died has of course been a gift from heaven for the anti-war crowd and whenever I try to explain quietly and carefully the rationale for the war someone will slap the 100,000 number in my face: end of discussion. I never had the material to debunk that number so I was glad a reader sent me an article he had written about it for a European paper. He wishes to remain anonymous, for business reasons, but I googled him and he checks out well. So here are his comments.

The first being that the Lancet – a notable opponent of the war in Iraq – released its numbers five days before the Presidential election, a calculated political move rather than presenting some scientific results. The real problems are however with the statistical methods used by the Lancet. The study used a sample of 7,500 Iraqis divided over 33 clusters, before and after the invasion. Such a sample is hardly sufficient to measure war casualties as violence in war is not randomly distributed over a population.

Looking at violent deaths it turns out that 14 months prior the invasion in the total sample only one violent death was recorded. The Lancet concludes that as a result in Iraq - for the same period - there could not have been more than some 3,000 violent deaths (out of 25 million Iraqis). An odd number bearing in mind that according to Human Rights Watch a total of 300,000 Iraqis perished under Saddam’s rule, averaging about 12,000 per year and this excludes the approximate 1 million that died as a direct result of the wars Iraq fought. The Lancet may have a point here in that domestic repression and killings were probaly higher at the outset of the Saddam regime then at the very end, but 3,000 appears to be a tenuous number.

Then there’s child mortality. The Lancet finds that the number of children dying during the first 12 months after birth has doubled from 29 per 1,000 births before the invasion to 57 after. This is interesting, Unicef in 2002 put that number at 102 per 1,000 on the basis of a far larger sample then the Lancet now uses. So the Lancet would have you believe that from 2002 to 2003 child mortality dropped by 70% while at the same time arguing that it doubled after the invasion.

There are other inconsistencies as well. One of them is that mortality for the entire population prior to the invasion was lower than that recorded by Unicef, that since the invasion the number of children dying from heart attacks increased from 0 to 6,500, for men the number went up by 15,000 and for women it went up by 6,500.

In summary, my reader argues the Lancet's statistical approach is flawed as it pictures too rosy a picture shortly before the invasion. The sample used leads to many conclusions about the period before the invasion in a way that contradicts pre-war results from more authoritative sources. Conclusions for the period after the invasion based on the same methodology are therefore questionable at best.

My Conclusion
It seems a reasonable but not quite comprehensive critique of the methods used. There is a fairly detailed argument that supports the Lancet's findings and it was put together by Crooked Timber who warned that you would have to take a course in statistics to follow some of their argumentation. Although I did well in statistics at school it is outside the scope of this post to render a final judgment of the methods used. The 100,000 number always looked suspiciously high and the obstacles in data gathering will likely make it impossible to come up with an accurate number. Civilian deaths are awful and I was struck by Robert McNamara’s assessment in the Fog of War that if the allies had lost World War II the numerous allied bombings on Japan and Germany would from today’s perspective have made Americans, Brits and Canadians guilty of war crimes and only because the Axis were defeated, the allies never ended up in court. If the Lancet and the anti-war movement can convince us that we should have appeased Nazi Germany and Imperial Japan than they may have a point about unacceptable civilian casualties, but I bet you they can’t.