COUNTING exactly how many people have died as a consequence of war is something of a dark art. This is particularly so in Iraq, where various approaches—including hospital death data, mortuary tallies and media reports—produce different results. A study based on statistical techniques by researchers at Johns Hopkins Bloomberg School of Public Health, in Baltimore, has concluded that many more people have died than revealed by these other methods. It claims that an extra 2.5% of the Iraqi population has died since the country was invaded in March 2003, mostly as a result of violence.
The study, published in the Lancet, is based on random sampling. Selecting small numbers of people at random allows statisticians to say something about the whole population. As there is no reliable census data in Iraq, truly random selection is impossible because the list from which this would take place is incomplete. So the researchers used a technique that is called clustering.
Clustering works by picking out neighbourhoods at random and then surveying all the people living in them. Gilbert Burnham and his colleagues gathered information on deaths from 47 neighbourhoods, each containing almost 40 households.
The survey teams—two male and two female medical doctors fluent in Arabic and English—asked people living in these households whether anyone had died since January 1st 2002 and, if so, what had caused the death. They also asked to see death certificates—and 92% of the reported deaths were certified.
The researchers then compared the death rate after the invasion with that before the fall of Saddam Hussein. They found that it went from 5.5 per 1,000 people a year to 13.3 per 1,000 people a year. They estimate that 650,000 more people have died since the start of the war than would have been expected had the war never begun.
Data from individuals within a cluster are highly correlated. As a result, that 650,000 lies near the middle of a wide range of possibilities. The researchers say that between 390,000 and 943,000 more people have died than would otherwise have been expected. Previous counts have suggested a much lower toll (although earlier work by the same researchers found a high death rate). This, with the wide range, raises suspicions: George Bush quickly dismissed the findings. But there is no mathematical reason why the true figure should be closer to one extreme than the other.
There are other reasons to be cautious. The violence in parts of the country make it perilous for researchers to function. Armed militias may also interfere. But concern that the number of deaths were exaggerated should be partly allayed by the death certificates. At the very least, the study represents a statistically valid attempt to calculate the dreadful things that have happened, and continue to happen.
442