Lancet redux

”That these two surveys were carried out in different locations and two years apart from each other yet yielded results that were very similar to each other, is strong validation of both surveys.”

This quote is from the companion document to the 2006 Hopkins study into excess mortality in Iraq. The authors are referring to the correlation with the excess mortality uncovered in the earlier, 2004 study, for the first 18 months of the war (Mar 03 – Sep 04).

The first study estimated 98,000 excess deaths as the confidence interval mid-point for the period March 2003 – September 2004. This is comprised of roughly 57,000 violent deaths and 41,000 non-violent deaths. The second study projects a figure of 112,000 for the same period. In fact, as the second study projects a decrease in non-violent death for the first 18 month period of the war from the pre-invasion baseline, the extrapolated excess mortality for the first 18 months of the war is conservatively above 130,000 (it’s impossible to say exactly, as the breakdown data isn’t available for the first 4 months of the second period in the 2006 study). The fact remains, every single one of the estimated (112,000) or extrapolated (130,000) deaths in Lancet 2 are due to violence. Non-violent deaths have gone negative to the tune of approximately 20,000, down from 41,000 in Lancet 1, whilst deaths from violence have doubled.

So whilst the headline figure for Lancet 1 and the Lancet 2 equivalent are relatively close, the composition of each could scarcely be more different. If you’re sent to the shop to buy an apple pie and come back with steak and kidney, do you get a pass because it’s still a pie?

And is it credible, given everything we were told and assume about the impact of the war on civilian infrastructure within in Iraq during those first post-invasion months that non-violent deaths actually decreased? Even allowing for this, how can the two studies be described as mutually corroborating when the composition of deaths for the first 18 month period are entirely contradictory?

Notwithstanding the above, it is clear the many tens of thousands of Iraqis have died since the start of this conflict. Those of use who supported the war should question our decision to do so every day. To support a war whilst remaining ignorant of the direct cost paid by others is an uncomfortable position in which to find oneself, yet until our governments undertake to conduct their own investigation into post-invasion mortality rates in Iraq, this is precisely where we are. An inability to quantify the level of sacrifice made by Iraqis undermines our moral legitimacy as supporters of the war. An unwillingness by the coalition to undertake such an exercise displays a contempt for the dead that shames us all.

“Hat tip” wouldn’t do it justice. All information gleaned from the comments thread at Tim Lambert’s blog where commenter “Mike H” is asking all the relevant questions. Check it out, starting with his comment on October 13th at 09:15am. He later sums up the Lancet 1 and 2 differences thus:

2004 study – approximately 278,000 total deaths (180,000 baseline + 98,000 excess) for the first 18 months of the war, 218,000 from non-violent causes and 60,000 from violence (3,000 being pre-invasion).

2006 study – approximately 310,000 total deaths (198,000 baseline + 112,000 excess) for the first 18 months of the war, approximately 178,000 from non-violent causes and approximately 132,000 from violence.

Discussion also ongoing at Crooked Timber, where the latest explanation for the violent-non-violent discrepancy is that the study respondents were lying in Lancet 1. To be fair, this is not in the least an irrational argument, but what it does for credibility of the figures is another matter.