Monday, March 13, 2006

The Civilian Death Count

Back in late 2004, there was a report published in the British medical journal Lancet saying that 100,000 people had died as a direct result of the Iraq War. Now the estimates are running as high as 250,000, with an average of around 150,000.

There was a lot of skepticism about the initial report, both when it came out, and today as well. No less a luminary than Michael Fumento has criticized it.

Some of the criticism, including the Fred Kaplan Slate piece I referenced, was on the basis of statistical problems with the study (although it ought to be pointed out that by leaving Fallujah out of their results, the team doing the study probably reduced the results somewhat, which would offset at least some of the factors that might exaggerrate their count). Often brought up were the disparities between this and the Iraq Body Count, which counts fatalities in Iraq by scouring news reports. (Estimates were between 14,181 and 16,312 deaths by October 29, 2004, and are between 33,489 and 37,589 as of today (March 12, 2006)).

Some people, such as Penraker have (in an article linked two paragraphs ago) insisted that the invasion actually saved lives, and that the people killed in the conflict are fewer than would have been killed if Saddam had remained in power.

The most obvious reason, in any case, why people discount the Lancet study is that there are nowhere near that many reports of deaths on record. So one wonders where all of these "phantom deaths" come from.

So let's first discuss the problems that have been mentioned with the Lancet study, and then discuss where those deaths might have come from.

A large part of the criticism was in the range of deaths presented. The range was somewhere between 8000 and 194,000 in the original study, or 98,000, plus 96,000 or minus 90,000. (why it wasn't 101,000, ± 93,000 is not spelled out; it apparently was not a perfectly symmetrical bell curve).

A lot of people dismiss the study as impenetrably broad based on this fact, but what needs to be considered is the way that statistics works. A 95% confidence interval means that the chances that the number will be within a particular range is 95%. However, the chances are not equal that it will be any point in the range. 95% confidence represents 2 standard deviations.

So what was the death toll as of the time of the Lancet study? If we assume an assymetrical bell curve with a mean of 98,000, and a standard deviation of 45,000 down and 49,000 up, there is a 68% confidence level that the result will be between 43,000 and 147,000. The chance that the deaths would be under 8,000 is only 2.5% (1 in 40) with a corresponding 2.5% chance that they were greater than 194,000.

For a discussion of standard deviations, click here.

Another thing to consider is that the study did not try to determine the gross number of deaths caused by the invasion but the net number. This means that if the death toll were 20,000, this means 20,000 more died than would have died if we had not invaded (assuming that the mortality rate in Iraq would have held constant at what it was from in the 15 months prior to the war). So if we save any lives from Saddam, the war killed those in addition to the 8,000 to 194,000 "excess deaths."

One issue that have with most pro-warriors when they are trying to talk about how many Iraqis were saved by the invasion is that they take the number of people who were estimated to have been killed during Saddam's reign, or at least the last 12 years of it, and extrapolate from that, without considering whether the rate of killing has gone down toward the end of the Saddam era. I would think that Saddam was not killing anywhere near as many people in the three years before the invasion as he did in his heyday, which would reduce the number of people he would likely have killed from March 2003 on, were he still in power. But I could be wrong. Anyone have any statistics to say one way or another?

So now to the issue of where the deaths come from. How can there be 100,000 deaths in October of 2004, or 150,000 to 250,000 now, if reports only show ~15,000 then and ~35,000 now? Obviously, there would have to be some unreported deaths. If one thinks about it, this is more than likely.

While Fred Kaplan's piece in Slate suggests that the Iraq Body Count is more reliable,

There is one group out there counting civilian casualties in a way that's tangible, specific, and very useful—a team of mainly British researchers, led by Hamit Dardagan and John Sloboda, called Iraq Body Count. They have kept a running total of civilian deaths, derived entirely from press reports. Their count is triple fact-checked; their database is itemized and fastidiously sourced; and they take great pains to separate civilian from combatant casualties (for instance, last Tuesday, the group released a report estimating that, of the 800 Iraqis killed in last April's siege of Fallujah, 572 to 616 of them were civilians, at least 308 of them women and children).

the fact of the matter is that reporting based on news reports is a surefire way to underestimate the death toll in a country as chaotic as Iraq. (Pro-War ideologues, of course, deny this and to the contrary insist that Iraq Body Count gives an exaggerated picture of the death toll. Penraker referred to it as "always suspect," although he didn't say why; presumably he believes that the reports on deaths are evil media lies or else that insurgents, particularly foreign ones, are being counted as Iraqi civilians) In fairness, Kaplan acknowledgs this:

The group also notes that these figures are probably on the low side, since some deaths must have taken place outside the media's purview.

However, he "allows" for unreported deaths by simply assigning a number to them; perhaps as many as the reported deaths (15,000 at the time). The fact is, he doesn't have any particular basis for estimating unreported deaths, or if he has one he doesn't tell it to us.

So how did these deaths occur? It appears at first to beggar belief that such a large number of deaths are occurring outside the radar. After all, it would seem that the media have a lot of access to Iraq and can find out anyhting they want, right? But that so many deaths could occur unheralded and uncounted becomes much more believable when one realizes that there are other major things that are occurring in Iraq that get very little press.

Tom Engelhardt and Michael Schwartz have commented on one such thing, something that in fact may explain the death count.

What people don't realize about Iraq is that there is an air war going on. Despite the way that talk of Bush's "new strategy," of more intensive air power and less intensive foot patrols (as which Schwartz hiself refers to it), tends to suggest that the U.S. has been sparing in its use of planes in the past, in point of fact air power has been a major part of this war since at least mid-2004. (If you'd like to see some article from August 2004 on, and you don't want to bother searching the piece linked to in the previous paragraph for links, then here are a few links from that piece to click on: here, here, here, or here). The "new strategy" essentially just tinkers with the air-to-ground ratio, it doesn't represent a qualitative change.

The clear import of the Schwartz and Engelhardt piece is that air attacks have caused large numbers of civilian casualties that have gone unreported due to logistical difficulties (i.e. when a bomb blows up a building, it can be difficult to establish who died; if the attack occurs outside of areas where reporters or even coalition infantry go, the attack might not even get reported).

So it is not unreasonable to assume tha the Lancet study is more accurate than it is being given credit for being.

That is NOT all. There is more. But I will leave that for another post.

That is all - for now.

No comments:

There was an error in this gadget