Shostack + Friends Blog Archive



There are a bunch of ways to estimate how many people have died in the Iraq war.  One is to keep track of news stories and official reports of combatant and civilian deaths, and add them up.

Another is to employ the tools of epidemiology and demography.  Until now, we’ve had essentially only the former to rely on.  That has changed with the release of a report [pdf] from the Bloomberg School of Public Health at Johns Hopkins and the School of Medicine at Baghdad’s Al Mustansiriya University, in cooperation with the Center for International Studies at MIT.

The headline-making conclusion is that the excess death toll from violent causes is estimated to be approximately 600,000 since the start of the war.  This is ten times higher than other estimates, such as those by

I predict that the authors of this work, their motives, and especially their conclusions, will be the subject of much uninformed debate, and more than a little derision.  It has happened before, when a demographer had the temerity to contradict the US government over the death toll in the Gulf War.  Her job was saved thanks to the efforts of the ACLU and the American Statistical Association.  Those inclined to criticize this latest work would do well to remember their history.

Update:  I neglected to provide a link to the methodological appendix (“full report” means something different to CNN than it does to me!)

Update: The Social Science Stats Blog provides further reading on the methodology.

5 comments on "Measurement"

  • Iang says:

    Interesting … what in brief are the tools “epidemiology and demography” or more usefully, how did they calculate their figures?
    (To be clear, I was hoping someone else would read the paper for me and let us know 🙂

  • Chris says:

    @Ian: My bad. A link to a detailed discussion of the methodology has been added to the post. It is a very quick read, and is better than my “keep it to a paragraph” summary would be.

  • Iang says:

    OK, I read that link. They used a technique called clustering. They constructed a random sample of locations called clusters, and then sent teams into each of those locations. Within these clusters they interviewed a group of residents. From that, they constructed estimates, corrected for the clustering effect.

    “Once the cluster is selected, additional sampling stages are required to locate neighborhoods and eventually a single house where to start. For each of these selection stages, a random process is used so there will be no bias to select one location over another. Once the “start house??? or location is selected, then the survey team moves to the next nearest (or sometimes the second or third nearest) house until the specified number of houses are selected (often from 10-50) to be interviewed in that cluster. The same is done for the other clusters.”

  • Iang says:

    For the easy summary, see Calculating casualties in The Economist.

  • tramadol side effects

Comments are closed.