In a month of big international news stories, Iraq has again dominated the headlines. Perhaps the most distressing report came from a British medical journal. The Lancet published the findings of a mortality survey conducted in Iraq by Professor Burnham and researchers from the Johns Hopkins Bloomberg School of Public Health in Baltimore and Al Mustansiriya University School of Medicine in Baghdad.
The study estimated that about 601,000 Iraqis have died from violent causes since the coalition invasion of Iraq on 18 March 2003 until July of this year. They found that the pace of violent death has accelerated every year since 2003.
President Bush immediately claimed the methodology “had been pretty much discredited” and the findings were “not credible”. Prime Minister Howard quickly said the results were “not plausible … not based on anything other than a house-to-house survey”. These were personal assessments unsupported by evidence.
It is important then, to understand how the survey was done. The researchers used a sample representative of the entire country. Forty-seven residential streets from 16 of Iraq’s 18 Governorates were randomly selected. In each street, residents from 40 homes were asked about deaths of family members since January 2002. 1,849 households were visited and 12,801 people were surveyed.
This is a large sample. The death rate recorded in these households during the period before the invasion was subtracted from the rate after the invasion. This represents the excess mortality attributable to the conflict. This rate was then applied to the populations of the Governorates to estimate the total excess death-toll for the country.
Because the researchers did not visit every household in Iraq, the exact number of deaths since the invasion remains unknown. Any method less than a census—asking every household in the country—introduces a margin of error. We are all familiar with television ratings; most of us intuitively appreciate that we do not need to talk to everybody in the population to accurately estimate the true number of TV viewers.
For the unfortunate Iraqis, sadly we can be confident that the war has caused at least 426,000 excess deaths from violence and possibly as many as 793,000. The true toll, however, is far more likely to be in the middle of this range than at the extremes. Sampling, when conducted according to strict statistical criteria as in the Iraq survey, is the only tested, valid and practical way of gathering data in insecure settings.
Because surveyors