| |
|
I thought you might be interested in a little detail about these numbers and the methodology used in the Lancet study, since Bush et al have rubbished them. The information is
from Eric Herring, an acquaintance of mine who is a well respected writer on arms control, WMD and some related aspects of Iraq (though not all issues of Iraq). He posted about this on the UK-Dance forum.
Burnham et al in the Lancet say 601,000 violent deaths with 95% confidence of 426,369 to 793,663 up to July 2006. While Bush et al have asserted that the methodology is bad, it's actually rigourous: multi-stage sampling. They identified 50 "clusters" across the country within the 18 Governorates, chosen on the basis of a ratio to population (e.g. Baghdad had 12 of the 50 samples because this is in proportion to its population size as a Governorate within the country). This "systematic sampling" process was then repeated within each Governorate, so that certain localised
administrative units were chosen as the researchable clusters, again on the basis of being the correct ratio to overall population (this time within the Governorate). These 50 clusters were then subjected to a third stage of sampling where each main street in each cluster unit was identified - as was each street that intersected it - and a random sample of the intersecting streets was chosen. Finally, at a fourth level of reduction, each street was randomly sampled by household.
This extrapolation is founded on solid field research. In terms of sample size we're talking about interviews with
12,801 individuals, visits to 1,849 households and - this is crucial - verficiation of actual deaths provided by 302 death certificates. This sounds - and is - very, very rigorous. They have geographically widespread, on-the-ground sampling of real, war-produced violent deaths with good a basis for extrapolating from the samples to their estimates of deaths.
However, there is another criticism of the survey - that the pre-war baseline deaths per population in Iraq are lower than the UK. According to The Lancet article, "pre-invasion mortality rates in Iraq were 5·5 per 1000 people". For comparison, in 2000, the mortality rate in the US was 8.8 per 1,000 (source: US Centre for Disease Control and Prevention), and in North West England, the mortality rate was 10.7 (source: NSO). However, apparently, it is reasonable to expect Iraqi mortality to have been lower than in the US because of the US' older population and higher incidence of obesity, heart failure etc.
Finally, Bush and co refuse to take on the point that the uncertainties of extrapolation could mean that the figures are HIGHER - and the 95% confidence interval of this study goes to 793,000, far beyond the existing estimate of 600,000. Taking the lowest figure in their confidence interval of 426,369 deaths over three and a half years, that's about 334 deaths a day nationwide every day since the invasion (say 500 a day for their most likely estimate of 600,000 violent deaths).
Hope this provides some useful ammunition... |
|
|
