Wall Street Journal on Lancet study

I missed this when it first came out, but Carl Bialik has written excellent summary of the issues in the Wall Street Journal.

Researchers concluded that about 100,000 more Iraqis had died outside Fallujah since the invasion than would have died had the prewar death rate continued. Yet the study, published in the British medical journal Lancet, was roundly criticized for discarding the Fallujah data from calculations. Others questioned the study for extrapolating from only 89 death reports outside Fallujah, including reports of 21 violent deaths.

The biggest concern with the Lancet study may be its enormous error range: The study said the number of dead from the war (again, excluding Fallujah) could be as low as 8,000, or as high as 194,000.

Les Roberts, epidemiologist at the Johns Hopkins School of Public Health and lead author of the Lancet paper, said that more than 80% of the time, the first two reported deaths in each neighborhood were confirmed by death certificate, and neighbors often independently corroborated the reports.

The Lancet study was dismissed by many critics, especially supporters of the war, because of two quotes that appeared in early press coverage of the report. Dr. Roberts told the Associated Press that he wanted the study to be published before the election (it was published online by the British medical journal Lancet in late October, just a month after the research was completed), so that "both candidates would be forced to pledge to protect civilian lives in Iraq." He also said he was opposed to the war. Meanwhile, Marc E. Garlasco, senior military analyst for Human Rights Watch, told the Washington Post, "These numbers seem to be inflated." Many commentators used that quote to argue that even Human Rights Watch, a group that has criticized U.S. forces' treatment of civilians in the war, didn't believe the study.

But neither quote provided just cause to dismiss the study. Dr. Roberts and co-author Richard Garfield, a professor of nursing at Columbia University who has extensively studied mass killings, say they had never discussed their views on the war before the AP interview appeared. Both also say that when the five researchers did discuss their views, they found that they were split on the war's merits.

I've written before about the importance of researcher bias, but in this case there's no indication that the research methods were chosen in a way that would intentionally inflate the numbers. Other researchers have told me and other reporters that the methods were sound; Dr. Roberts has experience measuring death tolls, including in the Congo, and his numbers have been widely accepted in less politicized conflicts. Furthermore, there's a difference between researchers letting their opinions affect their results and letting their results affect their opinions.

As for Mr. Garlasco's quote, he has since said that he hadn't yet read the study when he was interviewed. He told me last week, "I think the Lancet study is very valuable in some ways," and added, "I'm not a statistician. I don't really understand statistics. I try to stay away from numbers as much as possible."

Tags

More like this

Les Roberts comments on the shoddy reporting of his study: I thought the press saw their job as reporting information. Most of the pieces discussing our report were written to control or influence society, not to relay what our report had documented. For example, the day after the article came out…
Anjana Ahuja has written an extraordinarily one-sided article attacking the Lancet study. She drags out the same criticisms that were covered in the Nature story, but even though she cites the Nature piece, she carefully avoids mentioning the Lancet authors' replies, or the opinions of the…
The latest pundit to have a go at the Lancet study is Andrew Bolt. Like most of the critics, Bolt just does not have the statistical background to produce a competent critique. In Bolt's case this is even less excusable, since he had the benefit of the Economist's excellent…
John Allen Paulos writes about Iraqi war deaths: Another figure in the news recently has been the number of Iraqis killed in the war. President Bush mentioned last month that in addition to the more than 2,100 American soldiers killed so far in Iraq, that there were approximately 30,000 Iraqis…

As for Mr. Garlasco's statement: "I think the Lancet study is very valuable in some ways," and added, "I'm not a statistician. I don't really understand statistics. I try to stay away from numbers as much as possible." He reasons like many of the troll posters. I can't wait for certain innumerate ones to weigh in on this.

What's to weigh in on? There's nothing new here. Well, except for I'd like to comment on this:

Several researchers told me that the best bet for a complete account would be an effort funded by an international body like the United Nations or the World Health Organization, modeled after the Lancet study but with a far greater sample size and better accounting for high-violence areas like Fallujah.

I'd say they should model it after the UNDP's ILCS, just that they do a mortality study instead of a living conditions survey.

I also found this curious:

[Roberts] recommends a larger sample, and also dividing the country into three areas -- the Kurdish north, where his study found that mortality had declined since the war;...

His study found that mortality declined in the "Kurdish north"? I must say, didn't Roberts exclude most of that region from his survey? Mosul, which was surveyed, is made up of Christians, Kurds, Sunnis, and Turkmen. Oh well.

I'll repeat what I've said before, ad nauseam. The Lancet study was a good first attempt at estimating mortality and gave good reason to suspect the violent death toll was much higher than the Iraq Body Count estimate for that time period (as the IBC researchers may know, but as many people in the mainstream press do not seem to realize). But if we really want to have an accurate estimate, with much smaller error bars, the study needs to be repeated on a much larger scale, as Seixon suggests.

So we're all in agreement on the most important points (he hopefully suggests).

By Donald Johnson (not verified) on 18 Jan 2006 #permalink

Toll in Iraq's Deadly Surge: 1,300
Morgue Count Eclipses Other Tallies Since Shrine Attack
By Ellen Knickmeyer and Bassam Sebti
Washington Post Foreign Service
Tuesday, February 28, 2006; Page A01
BAGHDAD, Feb. 27 -- Grisly attacks and other sectarian violence unleashed by last week's bombing of a Shiite Muslim shrine have killed more than 1,300 Iraqis, making the past few days the deadliest of the war outside of major U.S. offensives, according to Baghdad's main morgue. **The toll was more than three times higher than the figure previously reported by the U.S. military and the news media.**

et tu, WashPost?