Wednesday, October 11, 2006
War over the Numbers
Let me say that I am not trying to be flip with the notion of the loss of life--death is never a laughing matter.
But we have estimates all over the place. The Iraq Body Count project estimates 40,000 killed. The Iraqi government seems to be comfortable with a figure of around 100,000. There is a slight difference between those numbers and a much higher figure.
And what we are seeing now, like death counts in previous wars as well as the ever popular clash between ballots counted and opinion poll data in disputed elections, is that the first rule of thumb is never to seek the objective truth but to find the numbers most useful to your cause. Remember the 100,000 killed in Kosovo in 1999 line? Or how in recent years the high numbers cited for the Bosnian civil war have time and time again been revised downward. Opinion poll data that was said to be more reliable in Ukraine in 2004 than the actual ballot counts was held to be flawed when there were major discrepancies in Azerbaijan in 2005.
What we are going to have over the next several days is fighting over numbers and ignoring the trends--because all the various surveys do agree that the death rate is going up and more and more are being killed in recent months.
"And I applaud the Iraqis for their courage in the face of violence. I am amazed that this is a society which so wants to be free that they're willing to -- that there's a level of violence that they tolerate."
By a similar logic, Iraqis must have loved Saddam so much that they tolerated his secret police.
Iraqi casualty figures get inflated in the same way. You interview some families and then you decide that the sample represents overall reality.
Did they even control for the notion of extended families--that family A might say, sure, we had a cousin killed--and that family B might say the same thing--but be referring to the same person because of kinship ties that in the Middle East go far beyond our nuclear family.
We saw in Bosnia how casualty reports quickly inflated; what are the controls here to prevent the same effect?