Can I Play With Madness?: Update on Iraqi Death Toll

A New Zealand academic has challenged the widely quoted estimate of a 655,000 death toll in Iraq, which was a US-led study's estimate, published in the Lancet, a British medical journal.

They say the study, by researchers from Johns Hopkins University in Baltimore and Al Mustansiriya University in Baghdad, overestimated Iraqi deaths by a factor of three.

About 218,000 would be "closer to the truth," said Dr Gourley.

The Lancet study - attacked by President George W. Bush and others in his Administration - said between 392,979 and 942,636 Iraqi deaths had occurred because of the war, which began in March 2003.

The research was based on a survey of 1849 Iraqi households in 47 randomly selected clusters around the country, asking about death numbers in each of the homes before and after the war started.

These figures were multiplied to produce a number for the whole of Iraq.

Dr Gourley is a physicist who has been studying statistical patterns in financial markets and civil conflicts.

He said the flaw of the Lancet study was that it polled too many families who lived on main streets or their feeders, where bombings and shootings were more common than in back alleys.

Families living near main streets would have a higher death rate than those in more isolated areas, distorting the national estimate.

The authors of the Lancet study have defended their research methods, saying that they sampled deeper into residential areas than their written report suggested, and that areas other than obvious sites of violence had been included.