Study: More Than 600,000 Dead in Iraq

The new mortality survey of Iraq that estimates 600,000 deaths by violence is startling and should alter the way America thinks about this war.

The John Hopkins University researchers were meticulous about the methods used to randomly choose the survey sites and analyze the data. It is state-of-the-art work, and its accuracy is not an issue. The survey is the only scientific account of the war dead. There is no other, and those who publicly dismiss the findings must offer an alternative. There is none. Every other account is deeply flawed in method, and this one is not. It is standard in epidemiology and disaster response.

The survey, which my Center helped organize, is available here.

Just two weeks ago, the Washington Post published a survey of Iraqi attitudes toward the United States and the war. The survey, conducted by the State Department, revealed that enormous majorities blamed the United States for the violence and wanted us to leave Iraq. Another poll from the University of Maryland published the next day confirmed that sentiment and also reported that 60 percent of Iraqis support attacks on U.S. troops. The Johns Hopkins mortality survey and these polls go hand in hand. The Iraqi attitudes are difficult to grasp unless the violence people suffer is an enormous daily threat to them.

The implications of this level of mayhem are profound. Most obviously, the United States is not providing security. It is not viewed by the Iraqi people as doing so, and the death rate confirms why these attitudes are so firmly held. The "mission" is not being accomplished, and if trend lines are an indication, the mission is deteriorating rapidly. The debate about withdrawing must be waged in this context.

It is conceivable that the application of force by the U.S. military is making things worse. Again, this is what Iraqis believe. A number of explanations for the violence see insurgent action in particular as "defensive" -- that is, the insurgents believe they are defending their communities. Because the United States went in with a relatively small number of troops, more force was applied to compensate for those inadequate numbers. (That does not mean, however, that larger numbers would have changed the course of the war.) This strategy has perhaps stirred the insurgency as much as any other plausible factor, and the growing violence then generates itself in a giant feedback loop: the United States attacks a village where they think insurgents are harbored, and this produces more insurgents who then act violently, exacting a new U.S. military response, and so on and so on.

Many of the journalistic accounts of the war, such as Thomas Ricks' "Fiasco," suggest that this may be what is occurring. At the same time, journalists are only seeing a tiny fraction of what goes on in Baghdad, what Dexter Filkins of the New York Times describes as 2 percent of the entire country, and thus their scope is very limited in seeing the violence, accounting for the dead, or drawing out the broader meaning. As a result, we have very little understanding of how the violence affects everything -- politics, ethnic and sectarian divisions, the hundreds of thousands displaced (another invisible statistic), the many thousands leaving Iraq in droves, the deterioration of the public health care system, and every other dimension of life and death in Iraq.

This is what we need to concentrate on as the discussion of the mortality survey unfolds. Even if there were a large sampling error in the survey -- which there does not seem to be -- the numbers would be colossal in scale. And it is the meaning of these colossal numbers that we must debate. We now have empirical evidence of the scale of this human disaster. In that light, what is best for Iraq? How can such violence be ended? How can the United States carve out a constructive role from the ruins of its intervention?

Let's honor the dead of Iraq by grappling realistically with their tragedy and forge a way to ensure that this horrific human cost does not continue to mount.