Wednesday, October 11, 2006

Believe the numbers?

The most thorough discussion of the Iraqi civilian death estimate report I've seen yet.

Not sure how they did the random sample or what their participation rate was. Randomly selecting who one asks to participate is important, but not as important as making sure that those who do participate are actually also random. In situations like this, sometimes participants have an agenda. I can believe that folks who lost someone would be more willing to participate and tell their stories than those who haven't lost anyone, who might not feel they've got reason to talk to the interviewers. This is "non-response bias".

It's the same phenomenon that makes people participate in follow-up surveys: Neutral folks don't bother, those with an axe to grind or those who want to sing someone's praises participate. One sees bimodal distributions, validating the "two different types of respondents" hypothesis. But neuteral folks don't participate. Non-response bias.

That said, the death estimate is too huge to be ignored. Even if their respondents were somewhat biased, it's still a staggering number and it's much bigger than the propaganda the US and Iraqi puppet governments have been spouting. The article makes a good point about pre- and post-invasion death rates from violence as evidence that things were not, in fact, this bad before we got there, repressive regime and all.

And it raises the question of nationalism as the last "acceptable" in-group out-group value distinction: We value American lives more than Iraqi lives. If we'd lost hundreds of thousands of Americans, we'd not have as much division, I think, over whether this war was a good idea or how it's going.

If W really does believe in God and heaven and all that, I can't see how he can sleep at night, with all that blood on his hands. Unless he thinks God really wanted all those people to die. That's not the kind of God I'd worship and follow.