Estimate of deaths low
Observer Viewpoint | Wednesday, November 9, 2005
Ryan Crane unfortunately misrepresents statistics in his Nov. 7 letter to the editor (ironically) entitled “Misleading Statistics.” Crane insists that The Lancet article, which estimates that between March 2003 and September 2004 the U.S.-lead Coalition caused 98,000 excess Iraqi deaths, is “easily debunked” due to “a confidence interval so large as to be almost statistically meaningless.” In fact, The Lancet estimate is probably quite low because it excluded data from hot spots like Fallujah (not to mention the number of deaths since September 2004).
Crane misleads the reader about how a “confidence interval” (CI) works in statistics. Crane makes the false presumption that a 95 percent CI of 8000-194,000 means that the number of deaths are just as likely to be any number between 8000 and 194,000. If this false presumption about CI were true, then The Lancet estimate would indeed be meaningless (as well as some of Crane’s own stats about Kurds killed by Saddam, not to mention the deaths from the Dec. 2004 tsunami, etc.).
However, the numbers in the middle of the CI are statistically more likely to be more accurate than the numbers closer to either extreme. The stats are something like this:
(i) There is a 2.5 percent chance that the number is lower than 8000, and a 2.5 percent chance it’s higher than 194,000 (2.5 percent + 2.5 percent = 5 percent, thus the 95 percent chance the number is between 8000 and 194,000).
(ii) There is a 10 percent chance that the number is lower than 45,000, and a 10 percent chance it’s higher than 167,000 (thus a 80 percent chance the number is between 45,000 and 167,000).
(iii) There is a 20 percent chance that the number is lower than 65,000, and a 20 percent chance it’s higher than 147,000 (thus a 60 percent chance the number is between 65,000 and 147,000).
Many refer to the lower number of innocent Iraqi dead (currently around 30,000, but around 15,000 at the time The Lancet article came out a year ago) at The Iraq Body Count website, which is based on the work of Marc Herold, an economist from the University of New Hampshire. Herold says his number is probably very low because it comes only from deaths directly caused by the U.S. Coalition that are reported by at least two different news organizations. So indirectly caused deaths (e.g., deaths due to the anarchy following the war) and deaths not reported by at least two media sources are not counted. According to The Lancet, the risk of death from violence in the period after the invasion was 58 times higher (95 percent CI 8.1ˆ419) than in the period before the war. Note that there is only a 2.5 percent chance that the risk of death from violence is less than 8.1 times higher than it was before.
Americans need to face the fact that our military kills a large number of innocent people. Many more have died since The Lancet article was published a year ago. We need to take an unbiased look at all the numbers in context. We need to admit to the reality of a war that we had overwhelmingly supported. Only then can we even hope to have the knowledge and self-knowledge to make mature, informed decisions about when and how to use the most powerful military in the world.
Sean Walshgraduate studentphilosophyNov. 7