Notes on Iraqi death estimates (part one)

Any visitors to this blog will no doubt have noticed, at the top left of their screen, the estimate from Just Foreign Policy indicating, as of this writing, that there have been 1,118,846 excess deaths of Iraqis due to the US invasion and occupation of that currently benighted land. Just Foreign Policy opposes the occupation and, I suppose, could therefore be accused of bias towards inflating the numbers, but let me explain why I think this horrifying statistic is indeed the closest we have to a true accounting of Iraqi fatalities, albeit still an inadequate one.

The numbers derive their base from the Lancet Study of 2006 of excess deaths in Iraq and then extrapolates further in time based on the rise or fall of fatality rates derived from the reporting of Iraq Body Count. In order to judge the validity of Just Foreign Policy's numbers, then, we need to look at the methodologies of both the Lancet Study and of the Iraq Body Count numbers.

The Lancet Study has been subject to a number of criticisms, mostly by highly interested parties, including George W. Bush, who in response to a question about the study dismissed it's methodology as "pretty well discredited" and who had a year earlier cited the far lower number of 30,000 deaths, a number I believe was at the time commensurate with the low end estimate of Iraq Body Count (current low end for IBC being 77,327). I am neither a statistician nor an epidemiologist, but it is notable that when the same methodology was used to estimate death tolls from the Rwanda genocide and the 90's wars in the Congo, cases that didn't (directly) implicate American imperialism, the numbers were largely accepted at face value and included in headlines around the world. It is also notable that Roy Anderson, chief scientific adviser to Tony Blair's Defense Ministry, called the Lancet Study's methodology "robust" and "close to best practice." In a nutshell, what the researchers did is called cluster sampling. They generated random geographic spots across Iraq and where those spots landed on population centers, from villages to cities, generated random neighborhood maps and, from those, random household selections. Interviewers were then sent out to those households to inquire about household births and deaths prior to and following the March 2003 invasion. Following the interviews, and this is important, families were asked to produce death certificates for the deceased and in over 90% of cases these were provided, a remarkable number to my document-challenged self and indicating high confidence in deaths reported.

Iraq Body Count collates only reported and cross-corroborated violent deaths of civilians from the sources listed here (which still reaches the rate of 63.5 a day for 2006 and 50 a day for 2007 so far).

Even if one were to dismiss the quite obvious fact that Iraq Body Count is likely missing a large number of deaths because Iraq has become, as is almost universally acknowledged, the most dangerous place to attempt to be a reporter in the world, it also represents an undercount of the human fatality costs of the war in a number of highly significant ways.

I will discuss these and the reasons that the Lancet Study also represents an undercount of the results of US policy towards Iraq (in addition to some unavoidable problems with the Just Foreign Policy estimate) in the part II of this post.