Saturday, January 07, 2006

100,000 yet again

I can't believe it! People are still repeating the "magic number" of 100,000 from the Johns Hopkins study and still pretending that it represents killed Iraqi civilians. Gilbert Burnham, of Johns Hopkins, is even inflating his own numbers and arguing that we, the US, have continued killing at the same rate, so 300,000 must be a better estimate than President Bush's 30,000. The BBC, certainly no friend of the War in Iraq, has already weighed in with a careful estimate of 25,000 civilians, mostly killed by AIF. Take a look at these BBC summary tables before you choose any number. Note especially the pie chart showing percentages of women and children. The prevalence of men makes it hard to sell this as a civilian count.

I really have no axe to grind in this argument. I am frankly in favor of our actions in Iraq. I certainly wish we had made no mistakes, but war doesn't usually work that way. The number doesn't sway me one way or the other because I don't know how many would be a reasonable number. I suspect we are a lot more careful with Iraqi lives than Saddam was. How many did he flush away fighting Iran, invading Kuwait and putting down insurrections? I am virtually certain that none of our prisoners will be dispatched feet first in a wood chipper.

My big complaint is that the 100,000 number is being used for political rather than statistical reasons. It was a SWAG anyway. There were so many problems with that study that it should serve as the horrible example in statistics classes. The biggest practical problem is that six interviewers, probably Sunni since they weren't killed, snuck into Iraq with Burnham in the dead of night and proceeded to do their interviews, or not do them, without supervision. What kind of results can be expected from such an arrangement? The study had all the trappings of Science, but I strongly doubt, at this point, the presence of scientific scruples among the principle investigators. I have written in the past about this and about the specific statistical problems here, here andhere.

I believe that Iraq has been ruled for decades by a minority Sunni Arab mob, not dissimilar to the KKK in attitude and behavior. Such a longstanding injustice has been hard to uproot, there as it was here. Most of the Iraqi casualties today, as you can see in the BBC article, are due to those who continue to resist this change, using despicable tactics that deliberately target innocents. The US cannot be blamed for this. In fact, I believe we err on the side of excessive gentility. To put things in perspective, Abu Ghraib was inexcusable, but it was not Mai Lai. Anti-Iraqi Forces are committing a Mai Lai every month.

1/7/2006 5:45 PM

UPDATE 1/12/2006
Andrew Cockburn has some analysis worth reading on the subject. I'd have to do a lot of work to determine whether there's some merit to the statistical suggestions he's quoting. Even so, it wouldn't address the main concerns I have that 1) the authors assume a single source of change acting over interchangeable time periods, and that 2) they have undue faith in the interviewers and subjects, who have every reason to modulate their responses. I have also ranted at length in the past about numerous statistical problems, distortions of the findings, and other issues, and with the cavalier use of a memorable point estimate. Links above.

Links to this post

Links to this post:

Create a Link

0 Comments:

Post a Comment

<< Home