100,000 DEAD IRAQIS

Ever since the story broke about the study that said 100,000 Iraqis have died since the U.S.-led invasion, I've been looking for an analysis of the study to give me more information. Right when I heard the story it seemed completely implausible and downright political. Gerard Alexander in The Weekly Standard shoots holes through the study conducted by the British medical magazine Lancet.

Here's how the study worked. A team headed by a public health professor at Johns Hopkins asked a small sample of Iraqis about deaths in their households, and then extrapolated a national mortality rate from there. The first grounds for skepticism is a very unpersuasive methodology. Slate's Fred Kaplan points out that the study's method only supports a 95 percent confidence in the conclusion that the war caused somewhere between 8,000 and 194,000 deaths, an extremely wide range. In other words, even taken at face value, the study says that the mortality rate may have gone up quite a bit but also may hardly have risen at all, or may have landed anywhere in between. But the study doesn't explain this, its authors haven't highlighted it in press statements, and the Times never mentions it at all. Now, conditions in Iraq probably permitted at most a very small sample, which compromises the validity of findings. But researchers know that going in, and this team chose to press ahead anyway with a study whose results would almost inevitably be so imprecise as to practically invite being overblown and overhyped.

Read the whole thing, as Alexander points out other problems with the study.