The 655,000 fraud

An op-ed in today’s Wall Street Journal (subscription required) somewhat incredulously questions the credulity of the reporters and editors disseminating without question last week’s Washington Post article about the John Hopkins study (published in the Lancet) that attributes 655,000 excess deaths in Iraq as a result of the war (emphasis mine):

“We have no reason to question the findings,” the Post quoted a Human Rights Watch official as saying. The article was fairly typical of reporting on the Lancet study, which has also been all over television and radio, as well as Internet sites such as Google and Yahoo! news.

All of which leaves us wondering if reporters and editors have enough sense anymore to ask basic questions about such enormous numbers, or whether they are simply too biased against the Bush Administration and its Iraq policy to do so. The 655,000 figure is more than 10 times higher than previous estimates of violent deaths in Iraq since the U.S. invasion, and it is larger than the number of Germans killed by allied bombing during all of World War II and larger than the number of Americans who died during our own Civil War.

While it’s obvious that Iraq has a terrible problem with sectarian violence at the moment, we find it hard to believe killing on the scale of Antietam or Gettysburg has been going on without anybody having noticed until the statistical wizards from Johns Hopkins showed up.

The 655,000 figure turns out to be an extrapolation based on a very inadequate sampling process. Pollster Steven E. Moore, who has worked extensively in Iraq, pointed out in an op-ed on this page yesterday that the Lancet study is based on information from a mere 47 “cluster points” around Iraq and 1,849 total interviews.

By contrast, a 2004 U.N. survey of Iraq used 2,200 cluster points for more than 21,000 interviews. The Johns Hopkins researchers also appear to have collected no demographic data on their subjects, so the group cannot be compared to census data to check if it is representative. “I wouldn’t survey a junior high school, no less an entire country, using only 47 cluster points,” Mr. Moore wrote.

Iraq Body Count — a nonpartisan outfit that keeps track of Iraqi mortality figures — has also issued a devastating critique of the Lancet/Johns Hopkins survey. It points out that the study implies that a thousand Iraqis died violently every day in the first half of 2006, with fewer than a tenth of them being noticed by “public surveillance mechanisms” and the press, as well as “incompetence and/or fraud on a truly massive scale by Iraqi officials in hospitals and ministries.”

Wow. Extrapolation like this makes even the people behind the Minnesota Poll look like pikers.

Leave a Reply