More on the Lancet study of Iraqi deaths

The right wing meltdown over the Lancet study was predictable. These ideologues operate in a perpetual state of denial. Like their beloved president, they reflexively subordinate science to politcs and brook no dissent.

When the wingnuts saw a conclusion they didn't like and immediately denounced the scientists as hacks. The Gateway Pundit even put up a picture of the lead author and encouraged his readers to deluge the guy with angry email.

The wingnuts don't seem to understand that not all evidence is created equal. They cling to the Iraq Body Count estimate, even though they can't explain why the a database of media reports should carry more weight than a systematic study of the population of interest.

The Lancet study is quite simply the best evidence we have about the death toll attributable to the invasion. The Iraqi investigators conducted interviews at 1849 households that contained
12 801 individuals. The scientists went all over Iraq, except two regions that were too dangerous visit.

The takehome message of this study is that vastly more people have died in Iraq than the government or the media have told us. Sure, the study has a wide margin of error. We can be pretty confident that between 400,000 and 800,000 excess deaths have occurred since the invasion. Even the lower bound is shocking.

The alternative estimates are based on news reports or official figures. It may come as a big surprise to some people, but not every death gets reported. Last year, the United States averaged about 44 murders a day, every day. Does each of those deaths make the news? Not even close.

Unlike their colleagues in Iraq, journalists can travel freely in the US without fear for their safety. They aren't embedded with anyone or confined to their hotel rooms. They don't have to preserve operational security or face censorship. Even so, only a fraction of the murders that take place on any given day even get written up in the local paper.

The official figures are compiled by people with a strong interest in downplaying the death toll. Furthermore, different ministries don't necessarily talk to each other. Some morgues are run by the government, others by militias. Of course, not all the bodies will make it to the morgue. Even if when they do, the morgues are usually off-limits to journalists.

The Lancet study is a foundation for further inquiry. No scientific study is definitive. However, the right wing is terrified of honest debate. They can't even engage with this study because they've got sticker shock.

See also: Mark Chu-Carrol of Good Math/Bad Math reviews the Lancet study and deems the study a pretty solid upper bound for Iraqi deaths so far.

The Questionable Authority takes a hard look at the numbers. Read on and learn, wingnuts.

 

Comments

[RickD]

Mark Chu-Carrol makes this important point

When I compare this to what the US government has been trying to feed us, I find that I trust these results much more: this study is open and honest, tells us exactly how they gathered and analyzed the data, and is honest and forthcoming about its limitations and flaws. In comparison, the official US estimates are just black-box numbers - our government has refused to provide any information on how their casualty estimates were produced.

Science is based on an open examination of methodologies. What has struck me about the Burnham et al. article in Lancet is that it represents a systematic attempt to provide a firm estimate for the number of dead in Iraq. There are plenty of caveats for the number they provide, but from what I've seen thus far I haven't seen anybody put forward any other number based on any kind of study that is remotely as sophisticated.

[Njorl]

An August defense department report to congress is indirectly supporting some aspects of the Lancet study. "Measuring Security and Stability in Iraq", section 9010 reported the following observations:

-Attacks increased to 800 per week by May of this year
-90% of those killed were killed by executions, which are much more common to death squads than attacks.

If you assume 1 death from each attack, and 9 times as many from executions you get 8000 violent deaths per week.

I realize that those assumptions are not ideal. Some attacks do result in execution style deaths, which would tend to lower that 8000 number, but weighing against this, I believe most attacks result in more than one death. The Lancet article's numbers for the most recent death rates predict about 1000/day (rates escalated over the course of the occupation) give or take 50%. That would be 3500-10500 per week currently.

I suppose someone enterprising could do a survey of reports of casualties from attacks, and get deaths/attack, and percentage of execution style deaths per total.

[Ted Stein]

While you are right that the Lancet study is the most scientific and up to date estimate of civillian deaths in Iraq, it is not, ummm, based on anywhere near 13,000 interviews.

link 1
The study, published by The Lancet, was based on a survey of 1,849 households at 47 random locations in Iraq this summer

Link 2
The U.N. numbers come from records of deaths; the Hopkins numbers comes from calculations derived from a random sample of 1,849 households in 47 neighborhood clusters across Iraq.

[Josh Jasper]

Between 400,000 and ...

This sentence just sort of trails off.

[Trystero]

I have some vague memory that the wingers were telling us that Saddam was killing 10K people a year (mass graves, blah blah blah). So the 30K number from IBC etc lets them claim no net harm was done.

Huh. Wingers with conscience? Or just a desire not to be seen as a monster? Either way it explains the meltdown.

Paraph. "600K is clearly too high. It can't be that high because...it can't."

[Lindsay Beyerstein]

You're right there were nearly 13,000 people in the households surveyed, but only one interview per household:

Three misattributed clusters were excluded from the final analysis; data from 1849 households that contained 12,801 individuals

[Joerg]

It's totally clear to anybody scientifically minded that the Lancet numbers are the new benchmark for discussion. they are plainly the only numbers that (i) attempt to measure actual (as opposed to reported deaths, (ii) are transparent about how they did it.

I do have relatively minor issues with the study. (Disclaimer: I am an academic statistician, and I have the actual study at hand.) As I wrote in some other post here, the 655k headline number adds 601k violent and 54k nonviolent excess deaths - but the latter are not significantly different from zero. I think the authors would have been well advised to remove them from the tally, then their headline number would be "only" 601k. Interestingly, the lower 95% confidence bound actually goes *up* now, from 392k to 426k; this may surprise at first but comes from the fact that nonviolent excess deaths might well have been negative.

Also, http://scienceblogs.com/authority/2006/10/the_iraq_study_-_how_good_is_i.php
observed that the authors use an Iraq overall population number of 26.3M, which is on the high end of available estimates, with the UN rather giving 24.7M. To insure themselves against claims of blowing up numbers, they should arguably have used the lower one (as they also did in their earlier study). This brings the headline number to 564k. The new lower 95% CI is easily computed by an according rescaling and comes out at 405k. Still higher than the 392k from their headline result!

So I do have qualms about the 655k number, but then, we all knew that this should not be taken too seriously. And, the reasonable lower bound of 400k ("reasonable lower bound" being a fair translation of 95% CI into everyday language) is actually robust to these issues.

One other thing. The study found only 1 violent death pre-invasion and consequently, contrasts post-invasion violent deaths with a pre-invasion violent death rate of zero. That's obviously not literally true, but if pre-invasion violent death rates had European, or even U.S., levels, it would be close enough. However, does that ring true as a description of pre-invasion Iraq? I do not have alternative numbers, but I see the worry that the contrast might reflect a measurement issue (imperfect recall, imperfect dating of deaths) or even Saddam keeping a low profile pre-invasion. Does anybody know a source for Saddam "victim counts" over a slightly longer time frame?

To be sure, this is going to bring the numbers even remotely down to the IBC order of magnitude. Davies in the Guardian already nailed the real point: Even with reasonable criticisms being taken into account, the ca 300 excess deaths in 1849 households (12801 individuals) are statistically incompatible with a five-digit overall death toll.

[Joerg]

correction to last paragraph: of course I meant to write "To be sure, this is not going..."

[Lindsay Beyerstein]

Thanks for the link to the Questionable Authority post. I added a link to it in the post above.

[Anthony Damiani]

I am deeply ashamed.

I had initially supported, and indeed publicly argued for the Iraq war because I believed the deaths from sanctions should weigh heavily on our conscience, and that, regardless of the presence of WMD or the deaths that would be caused in an invasion, the death rate over a five or ten year span would be dramatically lower. I supported going to war because I believed that fewer people would die. Whether due to a fundamental miscalculation on my part or the spectacular and unimaginable incompetence of the Bush administration, that has simply not been the case. It appears that I was not merely was wrong, but disastrously so.

May God forgive me, and may God forgive America.

[Fitz]

I was perusing your links below and cant seem to find this piece by Richard Nadlerin NRO. He disputes the numbers Lancet arrives at and compares them to the numbers endorxsed by the Brooking Institute & the Center for International and Strategic Studies.

http://article.nationalreview.com/?q=YmRmY2VhMDJiMjVmMGM3YjU3NzhhNTRiNTM4N2UxZjM=

The Brookings Institute’s Saban Center updates its database weekly, based on several sources:…. Brookings published a “low” estimate of 43,300 civilian casualties of the post-war, and a “high” estimate of 62,000, cumulative through summer, 2006

“The Hopkins researchers chose their “base-line” for pre-invasion Iraq carefully: January 2002 through March 2003. They chose a period of time in which Baathist violence against the Kurds was restrained by a U.S.-imposed no-fly zone in the north; a period of time after the extermination of hundreds of thousands of Shia and Marsh Arabs in the south.

Burnham and associates carved out a brief period of enforced peace within a 25-year regime singularly dedicated to war and internal slaughter. They called it a “baseline,” and they compared that baseline against a period of war.

As a result, 300 of the 302 conflict-related deaths they report are “post-invasion.”

I was wondering what responses were adequate considering this particular pointed critique. I don’t find Nadlerin’s particular analysis to be among These ideologues operate in a perpetual state of denial. Like their beloved president, they reflexively subordinate science to politics and brook{(ings) Institute} no dissent.

[Lindsay Beyerstein]

The title of the article says a lot about the editorial standards at the NRO: "Fixed Findings: Another cooked up study from the Lancet"

Not exactly in the spirit of mutual intellectual respect. I'll go read the article.

[Njorl]

I think the criticism about the 14-month period being cherrypicked for its low death rates is not really fair. You could go back years and have about the same rates. I think by 1997 the effects of the post-gulf war rebellions and retributions and refugee crises had disappeared, and the mollifications to the oil-for-food program had started. The authors could have gone back 60 months instead of 14 and probably gotten the same number. The problem is, if you go back too far, the demographics of the country you are studying might be different.

[togolosh]

Anthony - I am in the same boat as you. Pre-war I believed that the combination of deaths due to sanctions (which are directly connected to our country's actions) and the evil of Saddam (an evil likely to get even worse when one of his sons succeeded him) made war the least bad option. It's easy to look at war as always unacceptable because it causes immediate and acute suffering. When the alternative is chronic suffering, perhaps leading to more deaths than the war itself, war is IMO the least bad choice. I didn't buy the WMD arguments since it was clear even before the war that intelligence was being cherry picked. I made the error of thinking that PNAC and fellow travellers, having been banging the drums for war for over a decade, would have given considerable thought to postwar reconstruction. Stupid me.

[Lindsay Beyerstein]

Nadler uses bigger words than some of his allies in the blogosphere, but he's trotting out the same specious arguments and baseless accusations of scientific malpractice.

His overall point is that Saddam killed a lot of people, too! If you pushed the baseline all the way back to 1980 and included the Iraqi casualties from the Iran-Iraq war, then the baseline would be much higher.

However, at the moment we chose to invade, he wasn't slaughtering people on masse. The period before the invasion was relatively peaceful. The US and its allies were doing a good job keeping Saddam in check with the no-fly zone, etc. Which makes our rush to war appear even more absurd.

The point of establishing the baseline is to say, "if the status quo had continued, how many deaths would we have expected." There's no reason to think that the situation would have deteriorated from enforced peace to all-out-civil war in three and a half years if we hadn't invaded.

[pseudonymous in nc]

“The Hopkins researchers chose their “base-line” for pre-invasion Iraq carefully: January 2002 through March 2003.

Er, no. They divided the post-invasion period into three chunks, and used a similar-sized chunk before the invasion. If Nadlerin is imputing a particular motivation to the JHU team's decision, he should have the spine to say it explicitly, instead of weaselling about it.

In fact, his argument is doubly tosh: if it was a carved-out period of relatively low deaths, should not the US have attempted to sustain containment? It's remarkable that people invoke, for instance, the slaughter of Kurds in 1991 as a casus belli as if the slaughter was ongoing and a humanitarian intervention were necessary. They should have had the spine to say that past murders were only a retributive reason for war. HRW said at the time that there was no humanitarian basis for the invasion.

Naderlin appears to wish that time could somehow be jiggled around so that the period before the invasion might magically become 1991, not 2002.

[DavidByron]

The Good Math/Bad Math site's comments are not worth linking to. His statement that 600,000 is an upper bound proves (as he admits) he hasn't read the study or even as much as a report of it because that figure is easily exceeded by the upper end of the range (which is, what? 800K+ ?)

He also says,
Population-analysis sampling based techniques like this do tend to produce larger numbers than other analyses

I don't know what he means by that but it gets a lower result than the obvious method of just assuming that the death rate for the sample is equal to the death rate for the whole country. I think the analysis for the earlier report (98K deaths) gave a far lower result than just looking at the average. The researchers also just ignored the Falloojeh death toll as an "outlier" which had the effect of halving their estimate.

In effect this report is likely already low-balled somewhat.

Why do liberals do this all the time? Do they think lying to low-ball the figures will make them more credible with conservatives? It doesn't. Ever. They just take the number you low-balled and low-ball it all over again.

Ask yourself WHY so-called liberals were using ridiculously inaccurate figures such as 50K before this report came out when better estimates were easily calculable. WHY was this a surprise to so many?

Maybe by "other analyses" he means Iraqi Body Count. If so he's wrong. IBC is not a bad estimate or a good estimate. It simply is not an estimate at all. It's an exact count. And it's not a count of the death toll but of a tiny subset of the death toll. It's absurd to compare them as competitors, yet we see that again and again. even liberals asking which is the "better" estimate.

[Lindsay Beyerstein]

I was wondering where he got the idea that 600,000 was the upper bound.

[Fitz]

That’s a fair minded critique of Nadlerin’s point concerning the baseline. I always believed that we had him adequately contained and that this was a war of choice. However, Saddam & his regime were an unpredictable foe, no one expected his invasion of Kuwait & his WMD program was larger (after the first Gulf War) than any intelligence agency imagined at the time.
It’s obviously difficult (rather impossible) to predict the future. That said his relative (enforced) passivity in the pre-invasion years should be gauged along with some acceptance of his volatility and ruthlessness toward his subjects.

[DavidByron]

Anthony,
I believed the deaths from sanctions should weigh heavily on our conscience, and that, regardless of the presence of WMD or the deaths that would be caused in an invasion, the death rate over a five or ten year span would be dramatically lower. I supported going to war because I believed that fewer people would die.

Well obviously your stance was utterly immoral in as much as you decided you had the right to decide another nations fate. War is the ultimate crime, if you felt bad about the genocidal sanctions you could certainly have opposed them without endorsing aggression.

However in fact Bush had declared in public that the sanctions regime would not end after the occupation. Therefore if you were listening you'd have known he promised to only increase the death rate. He has been true to his word.

[DavidByron]

togolosh,
It's easy to look at war as always unacceptable because it causes immediate and acute suffering. When the alternative is chronic suffering, perhaps leading to more deaths than the war itself, war is IMO the least bad choice.

The choice was not yours to make.

You listened to people you knew to be liars and ignored the people whose life you would be snuffing out. There is no excuse and it isn't a question of your stupidity. It was an immoral choice regardless of the accuracy of the data you trusted. You don't get to kill people because you happen to think it's for a good cause.

[pseudonymous in nc]

That said his relative (enforced) passivity in the pre-invasion years should be gauged along with some acceptance of his volatility and ruthlessness toward his subjects.

Of course. I recall that one argument for a March 2003 invasion, as opposed to retaining the option for when a security-based or humanitarian reason was that it would cost too much to keep those American troops sitting in Kuwait on combat ready-status, twiddling their thumbs.

But that's the problem with pre-emptive wars of choice: they don't provide defenders with easy recourse to how bad things were in the immediate time before.

[DavidByron]

Joerg, I don't think most of your quibbling over the numbers is worth replying to, although I could. It's distastefull, especially if you are American, to be seen to try and low-ball the size of your country's genocide. I don't get the impression you are trying to do that exactly. But this isn't the time for strictly accademic interest.

I will comment on this:

The study found only 1 violent death pre-invasion and consequently, contrasts post-invasion violent deaths with a pre-invasion violent death rate of zero. That's obviously not literally true

Obviously you don't know if it is true. That statement is simply a statement of your own bias. Taking the results without bias they show that Saddam was just not going around killing people as American propaganda had claimed. It was just one more lie.

[DavidByron]

Lindsay you are missing the point on the baseline death rate.

The point of establishing the baseline is to say, "if the status quo had continued, how many deaths would we have expected."

We want a death rate relative to "normal" peaceful conditions not a death rate relative to the genocidal embargo period.

Iraq went from the frying pan to the fire alright but we want to know how many people died in the fire NOT how many MORE people died in the fire than in the frying pan.

Therefore to get a true death toll you need to add to the Lancet figures the rate that the sanctions represented in excess of "normal" peaceful conditions for Iraq. About 400,000 dead, for a total of about 1 million.

To do otherwise is a moral absurdity.

Let's say Bush had droped a nuclear bomb on Iraq prior to the invasion. That would have raised the pre-invasion death rate wouldn't it? That in turn would have lowered the post-invasion death toll count. Kill enough Iraqis pre-war and you could even claim Bush saved lives by invading.

Which is absurd of course. The war wouldn't be saving lives in that hypothetical scenario. It would simply have been killing less lives than the pre-war period.

So the war has killed about 1 million. But only 650,000 more then the US would have killed with sanctions.

[Alon Levy]

If you want a non-inflated baseline, just use the regional average death rate, which is about 4/1,000 (prewar Iraq was 5.5). That would increase the number of excess deaths by 40,000 per year.

It's distastefull, especially if you are American, to be seen to try and low-ball the size of your country's genocide. I don't get the impression you are trying to do that exactly. But this isn't the time for strictly accademic interest.

It's also distasteful to try shutting people up just because their numbers don't accord with your bias. Now is the time to tell the truth; if your ideology is out of sync with them, it's time to change your ideology.