That figure is a statistical extrapolation based on interviewing 1,849 Iraqi families.
Now, if the stratification of the sampling was not properly done, and I doubt it was, then the estimate is worthless. Disraeli is oft quoted as saying "there are lies, damn lies, and statistics." The reason he said that is that even in his days there were so many way to make the statistics say what the researcher wanted that they couldn't be relied on for the truth when politics were the objective. And make no mistake, politics are the objective of the, so called, study which arrived at that number.
That study was done by some folks at John Hopkins and the sample was apparently based on data collected in areas which eight Iraqi Phsycians had access to. These folks have done the same kind of study before, one in which they concluded that over 100,000 people had been killed in Iraq during the first 18 months of the war alone. That study admitted to a mide margin of error and I have no doubt that the same applies to this study as well (the lancet article reports a confidence level of 95% of 393,000 to 943,000 excess Iraqi deaths in the current study, thats a pretty wide margin of error and a fairly low confidence level as medical research goes). The study is conducted by "cluster sampling" in areas which are accessible. That is, the country was divided into zones of equal population size and accessible points where picked within those zones then, "clusters" of families near those points were interviewed (the study says the points picked were random, but lets get real here, access to the point was more important than randomness for obvious reasons). Thus, the methodology of the sampling does not lend itself to a very accurate statistical estimate.
The timing of the publication of the study, as well as the journal in which it was published, both make the purpose of the study and its conclusions suspect as well. I find it interesting that the UN and everybody else over there put the numbers from a low reported by the military of around 50,000 to a high of between 60-70,000. Nobody on the ground over there, including those who'd like to discredit the US, report numbers anywhere appraoching even the low end of the Hopkins study. Since the first contraversial study by this group was also published in 2004, during the presidential campain, I think the motivation behind this work is clear, it is nothing more than a political hatchet job.
Forscher
