654,965

Thats out of a pop of 24MM. The US is over 295MM, so percentage wise, it would be like someone killing 8MM Americans.

It's even worse than that. The 650,000 figure is the number of excess deaths, so to get the real number of occupation dead, you need to add the number of people who would be killed by Saddam or the sanctions in a comparable period of time.

Before the war, Iraq had a very high death rate compared to other countries in the region. Going by the regional average, the number of excess deaths is even higher (I think it works out to something like 800,000, but don't quote me on it).

Probably exagerated, but we have no way of knowing how much. I'm very amused by the "Those number's can't be accurate, becasue the insurgents would _kill_ anyone who tried to get accurate numbers!" argument that some off the wing-nuts have been spouting. :-/

No, it's probably not exaggerated, at least not to any serious degree. Mark Chu-Carroll wrote that it tends to get overestimates, but that it tends to still be the most accurate estimate around.

I think it's interesting that a common rebuttal of that number, including Bush's "it's not credible" comment, doesn't critique the methods of the study, but instead relies on the common creationist argument from incredulity.

I don't think I need to say anything about the quality of a mind that judges credibility not on the arguments but on the conclusion.

If you want to hear from the people who actually conducted the study, and what it is like to have to visit 50 random locations in Iraq, listen to the This American Life episode where the study authors talk about how they get the actual count from family interviews.

This interview is about the study they did a few years ago that found about 100k dead 2 years ago, but the study methods are exactly the same as this current study. It was a great episode I thought.

http://www.thislife.org/pages/descriptions/05/300.html

This is beyond appalling. I knew it was bad, but this is horrendous. I don't think humans are good at even picturing that number of people, never mind to truly contemplate their deaths. I hope something can be done to stop this insanity soon, and hold accountable those responsible ...

As mentioned before, check out Mark Chu-Carroll's analysis on GoodMath/BadMath. It's a good discussion of the study itself, as in the particular way they extracted the numbers from their data.

In particular, 'cluster sampling' tends to overestimate, so the number published in the Lancet study should probably be considered an upper bound. The lower end of their confidence interval (about 450k) is probably closer to the real estimate.

Mark also comments on the critiques of the study, and why so many of them are wrong. He also discusses a bit about the other death-toll estimations, and how they intersect with this number (essentially, the methods used in studies such as the one at IraqBodyCount are known to underestimate, so at best they should be seen as a lower bound on the data).

It doesn't look like my link worked - go here:
http://scienceblogs.com/goodmath/2006/10/the_iraqi_death_tally_study.php

I'm reminded of The Americanist, who trolled this blog a few months ago defending Richard Cohen's "math is for losers" argument on the grounds that all American students needed to learn was how great the US was.

In case anyone still remembers: this is the best argument for teaching people math that I've seen.

Remember when I got ripped to shreds for saying 100,000?

Just asking.

655,000 seems like a high guess to me, though. But, heck, I'm not there; what do I know?

I've read the Lancet paper that argues for 200,000 without Falluja -- and that was years ago. I fear 655,000 is completely realistic.