Lancet: 100,000 civilians dead in Iraq

2 thoughts on “Lancet: 100,000 civilians dead in Iraq”

1. Did you check the confidence intervals associated with the estimates in the study? They completely invalidate the results.

2. Uuh, no. A wide CI does not “invalidate” a study. CIs tell you something of the spread around the estimate, and its stability. A whole lot of non-expert rightwingers have been bashing this study in remarkably uninformed ways, and this is one example of them.

   The weakest part of the study is the size of the sample. Although they interviewed 7868 people (which is a huge figure in terms of a scientifc study), that is not the number that was randomized, which is the key. The randomized unit was households, which numbered 33, and the analyses were made using that unit (as they should have). That’s the reason for the wide CI.

   The study tells us with virtual certainty that there have been thousands of excess deaths in Iraq. It tells us with less certainty that most of these deaths were among women and children, most were violent, and most were the result of air strikes.

   It estimates with much less certainty exactly how many thousands of excess deaths occured. The best figure they can put on that is 98,000, but it could be much higher or much lower.

   It is not true that it is as likely to be just 8,000 as 98,000, as the uninformed will tell you. The farther the figure moves away from 98,000, the less likely it is that it will be “true”. However, the extent of the range within which the true figure may lie indicates that the estimate is not stable.

   We already know that this war has resulted in a huge number of civilian deaths from American firepower. A bigger study needs to be done to determine how much.

Comments are closed.