Body Counts, And More Damn Lies... Errr, I Mean Statistics

First things first.
For those of you who don't know, the Lancet, a medical journal, has published a study in which they claim that their best guess as to the numbers of Iraqis killed since the beginning of the war is about 655,000.

I was gonna do a whole long post about how this is obviously bullshit.

Until it turned out that someone else already did it for me.
So, instead, I'm going to post big chunks of their article with my snide commentary just to help it along. Sound like fun? Good.
As Stephen Colbert puts it: "Strap yourselves in, America, you're about to get a truth-o-cution!"

A new study has been released by the Lancet medical journal estimating over 650,000 excess deaths in Iraq. The Iraqi mortality estimates published in the Lancet in October 2006 imply, among other things, that:

1. On average, a thousand Iraqis have been violently killed every single day in the first half of 2006, with less than a tenth of them being noticed by any public surveillance mechanisms;
2. Some 800,000 or more Iraqis suffered blast wounds and other serious conflict-related injuries in the past two years, but less than a tenth of them received any kind of hospital treatment;
3. Over 7% of the entire adult male population of Iraq has already been killed in violence, with no less than 10% in the worst affected areas covering most of central Iraq;
4. Half a million death certificates were received by families which were never officially recorded as having been issued;
5. The Coalition has killed far more Iraqis in the last year than in earlier years containing the initial massive "Shock and Awe" invasion and the major assaults on Falluja.

If these assertions are true, they further imply:

* incompetence and/or fraud on a truly massive scale by Iraqi officials in hospitals and ministries, on a local, regional and national level, perfectly coordinated from the moment the occupation began;
* bizarre and self-destructive behaviour on the part of all but a small minority of 800,000 injured, mostly non-combatant, Iraqis;
* the utter failure of local or external agencies to notice and respond to a decimation of the adult male population in key urban areas;
* an abject failure of the media, Iraqi as well as international, to observe that Coalition-caused events of the scale they reported during the three-week invasion in 2003 have been occurring every month for over a year.

In the light of such extreme and improbable implications, a rational alternative conclusion to be considered is that the authors have drawn conclusions from unrepresentative data. In addition, totals of the magnitude generated by this study are unnecessary to brand the invasion and occupation of Iraq a human and strategic tragedy.

Before we get started, this article whose content I am reposting is from Iraq Body Count, a site which has been keeping a triple-fact-checked body count of Iraqi casualties since the war began. Note that unlike the Lancet study, IBC's work comes from death certificates issued; witnessed deaths; police and media reports; and is cross-checked by no less than three different team members before its data is added to the body count.

As such, I am more inclined to trust their estimate of figures than that of the Lancet, which according to their own article, covers a range of 392, 979 – 942, 636 for excess deaths since the invasion. (An excess death being one over the statistical norm for a peacetime period in the same area.)

Let's address this in several steps, shall we? First, the margin of error. This is the big bugaboo in this study; the margin of error in the study's results is 549,657. That is, for those of you not math- or calculator-equipped, a 58.3 % margin of error.

FIFTY-EIGHT PERCENT.

The defenders of this study have gone on and on about the fact that "cluster-sampling" (the method used to come up with these numbers,) is a standard, acceptable statistical practice. This is true; however, to those staunch defenders I would note that if properly used, it usually results in a margin of error closer to FIVE percent. Look at the election day "guesses" on TV; this same method is used to determine their numbers. Note the little blurb at the bottom of the screen, saying the margin of error - 5%. 3%. 4%. Tiny percentages of error.

This says that although the methodology of the study might very well have been a standard, valid method, it was misapplied with a vengeance by the Lancet researchers.

I want to emphasize that number again. FIFTY-EIGHT PERCENT margin of error.

The much ballyhooed 655,000 number that the Lancet is using as their official "estimate" of excess deaths in Iraq is the median number.

Now, the lowest estimate number, the not - quite - 400, 000 one, is already vastly higher than any estimate thus far published, even by those speculating wildly, but I'd actually be prepared to accept it as somewhat believable, simply because the Iraqis don't like us much and don't report in when they're injured; but 650, 000? You have got to be fucking kidding me.

Let me put it in perspective.

The Lancet study delivers a guesstimate of 1,000 deaths per day in Iraq. The U. N. figures, and those of the Iraqi Ministry of Health, come to about 80 per day. I'm prepared to accept that a lot of deaths go unreported, but 920 per day? 12 times as many as ARE reported?

Bullshit.

You know what I'm gonna say.
But Wait, There's More!™

IBC points out this gem:

If 600,000 people have died violent deaths, then the 3:1 ratio implies that 1,800,000 Iraqis have by now been wounded. This would correspond to 1 in every 15 Iraqis.

And follows that up with this:

This yields a revised Lancet-based estimate of 800,000 wounded over the equivalent period for which the MoH has been collecting this information centrally. In that same two-year period the official total of wounded treated in Iraqi hospitals is recorded as 59,372.

Whether hospitals can provide a comprehensive tally of violent deaths or not, their knowledge of seriously injured should be much more complete.

Accepting the Lancet estimate would entail concluding that at least 740,000 wounded Iraqis (90% of the total) were not treated or, if treated, not recorded in any way, throughout a 2-year period beginning in mid-2004. It may be that many injured anti-occupation combatants have avoided hospitals to prevent identification or arrest, but they are hardly likely to account for more than a small fraction of this discrepancy. It would further imply that approaching 90% of Lancet's deaths are also of combatants.

Note that this implies that far, far, far more injured Iraqis have been simply walking around, wounded, rather than being treated. The Ministry of Health estimates 60,000 injured have been treated in the Iraqi hospitals.

Not 1.8 million. 60,000. That's a HUGE difference, and makes the Lancet study questionable in and of itself.

But Wait, There's More!™

The Lancet study also assumes that the Ministry of Health figures are false, anyway; they claim that they relied on death certificates for 81% of their "confirmed" deaths, but in fact the Ministry of Health has records of issuing only about a tenth of the death certificates counted by the Lancet researchers. The Lancet dismisses this as unimportant, claiming that:

"Even with the death certificate system, only about one-third of deaths were captured by the government's surveillance system in the years before the current war, according to informed sources in Iraq. At a death rate of 5/1,000/year, in a population of 24 million, the government should have reported 120,000 deaths annually. In 2002, the government documented less than 40,000 from all sources. The ministry's numbers are not likely to be more complete or accurate today."

Except they're lying outright; the Ministry of Health recorded 84, 205 deaths in 2002, excluding Kurdistan. Which means that the Ministry of Health figures are 70% of the Lancet's estimate for that year, not a third.

This means that unless the accuracy of the Ministry of Health's recordkeeping has completely fallen off - and there's no evidence that it has; they use the same methods now that they did 4 years ago - the Lancet study's figures should be reflected in a Ministry of Health casualty figure of about 460, 000.

Sadly, it's not. The Ministry of Health recorded 115, 785 deaths in 2005 - 320 per day. Note that this isn't "war deaths," but TOTAL deaths. The Lancet study, which purports to be estimating the number of WAR DEATHS, is estimating at a minimum - a MINIMUM - four times that many.

Let's take a look at what happens when you add the other 30% to the MoH's figures, shall we? According to the existing correlation between the MoH's figures and those of the Lancet's BAD-ASS SCIENCE, you'd get a figure of approximately 165, 407 total deaths in Iraq for 2005. That's a lot, but that is ALL deaths, not "war deaths."

Let's see how that adds up, shall we? If we assume that this death rate has held relatively steady - it hasn't, but let's anyway - since we invaded in 2003, we get a total since - the - invasion death toll of 496, 221. That's just over the absolute lowest figure the Lancet study estimates as the number MORE deaths there have been since the war started than there normally are.

The Lancet is estimating that there were more war deaths in Iraq in 2005 than there were actual, total deaths in Iraq in 2005. Read that again, more carefully.

THE LANCET IS ESTIMATING MORE WAR DEATHS THAN THERE WERE TOTAL DEATHS.

Does anyone still believe this study isn't completely full of shit?