« Much More Alexsey Vayner | Main | What the EU Costs Us. »

October 11, 2006

Deaths in Iraq

The Lancet will be publishing an updated version of that report about civilian deaths in Iraq.

Here's the Wall Street Journal's take.

600,000 deaths from violence, 55% of all deaths being caused by violence.

Here's Tim Lambert.

Given my cock up last time I don't think I'll be making any further comments upon this.

October 11, 2006 in Military | Permalink


TrackBack URL for this entry:

Listed below are links to weblogs that reference Deaths in Iraq:


On the face of it, this report seems plain idiotic. The 600,000 isn't the number of dead since the invasion, it's the excess due to violence. If we assume Iraqi life expectancy in the absence of violence to be 50 years, then 2% of the population dies every year. That's 600,000 p.a. And somehow we're supposed to have overlooked a death toll that is suddenly 30% higher? You just can't hide or ignore those sorts of figures. Iraq Body Count's methodology has been attacked for using self-reporting and double counting, and even they say the number's more like 50,000. Maybe that's a sound estimate. I'd even buy the Lancet's previous inflated number of 100,000. But six hundred? Pull the other one. That's Rwanda-sized. Also, rather interesting timing on the release of this study, don't you think?

Posted by: David Gillies | Oct 11, 2006 5:16:07 PM

[And somehow we're supposed to have overlooked a death toll that is suddenly 30% higher? ]

There are quite a few people who have, in fact, noticed that things have got quite a lot worse in Iraq over the last couple of years.

Posted by: dsquared | Oct 11, 2006 6:07:07 PM

David Gillies: you are a fucking idiot.

Posted by: Charlie Whitaker | Oct 11, 2006 8:10:37 PM

Mr Whitaker: what an erudite rejoinder.

One does not have to think that all is rosy in Iraq to question the magnitude of this claim. Of course people have been dying in Iraq at a rate greater than would have been the case were the country at peace. But we are expected to believe that excess deaths are in the same region as WW2 US and UK deaths combined, and no-one noticed.

Posted by: David Gillies | Oct 11, 2006 8:43:58 PM

and no-one noticed

Like I said.

Posted by: Charlie Whitaker | Oct 11, 2006 8:49:52 PM


I believe David is making the point that everybody would have to have missed the 500 deaths per day this figure would require, with the required funerals, bodies, media reports etc etc.

As always as regards Iraq, emotion will come into play, as exhibited by Charles Whitaker above, those who want to believe these somewhat implausible figures will, those who don't, won't. Reason and logic often takes second place.

Posted by: Andrew Paterson | Oct 11, 2006 9:07:24 PM

Huh? More fuckwits?

500 deaths per day, in a country with a population of around 26 million, is one excess death, per day, per 50,000 population.

Dresden-type events are not required to produce the 600,000 reported. But these are nonetheless violent deaths, and the violence has been sustained for several years now, so it seems likely that the Iraqis have noticed that something is up even if you haven't.

Don't kid yourself that you're being reasonable or logical. You're not. You want to argue the findings: email the epidemiologists who conducted the study.

Posted by: Charlie Whitaker | Oct 11, 2006 9:43:52 PM

Are you going for the 'most belligerant blog poster award' Charles? Yeesh.

"Don't kid yourself that you're being reasonable or logical. You're not. You want to argue the findings: email the epidemiologists who conducted the study."

And here was I thinking that blogs were meant for discussion. I personally don't put much weight on extrapolated studies such as this, being an analyst I'm more interested in concrete facts and when they're not available I have a hard time ignoring my BS detector.

Posted by: Andrew Paterson | Oct 11, 2006 11:58:19 PM

The more deeply I delve into this 'study', the more statistical jiggery-pokery emerges. I believe this will be one of those things that becomes definitively debunked within scientific and technical circles but nonetheless gains an unshakeable hold on the public imagination. Extrapolating 547 deaths to 650,000 is a stretch even for 'epidemiologists' (the bunco artists of the scientific community). The Lancet gave credence to the MMR/autism link and has been at the forefront of the ludicrous Environmental Tobacco Smoke brouhaha, which gives us a datum for assessing the quality of its peer review process. Now we have a minuscule survey, part of which was undertaken in some of the most violent areas of Iraq (Baghdad, Fallujah) with people who have little incentive to tell the truth, and it yields a figure that says Iraq's excess mortality is comparable to that of Japan during WW2 (except in three years instead of eight). One person in forty in Iraq has died as a result of the war? 'One death per 50,000 per day' might sound like it is lost in the noise until you consider it is 130% more than the baseline rate of 5.37 per 1,000 p.a.. It is certainly credible that casualties in a warzone are two and a half times higher than in peacetime. Lord knows, they can be a lot worse than that. What is not credible is that is has taken three and a half years for the scale of this surge in mortality to come to light, in an era of omnipresent media coverage and a largely hostile press. The UK lost a much smaller percentage of its population during WW2. I don't think it was the spring of 1944 before the Lancet did a survey and found people noticing that they hadn't seen some of their neighbours in a while.

Where the Lancet people studied statistics is unknown (to me, at least), but if you have a sequence of numbers, all of roughly the same order of magnitude, and then one comes along that's ten times higher than the rest, you don't leap on it with an "aha!" and use it to browbeat the authors of the other findings (unless you're a low-sodium fascist working for the HSE, for example). No, you think, "hmm, outlier," and backtrack to find out where you went wrong. I know the study was supposed to be 'using the best methodology under the circumstances.' That's like saying, 'best North Korean luxury automobile'.

OK, time to stop the flogging. Poor old Dobbin's ribs are starting to show. Charlie Whitaker can call me all the names he wants, but I'm sticking to my guns: this is a bit of pseudo-statistical flummery so flawed as to be worthless.

Posted by: David Gillies | Oct 12, 2006 7:13:00 PM