Learning to Count

President Bush’s statement acknowledging 30,000 civilian casualties in the war in Iraq has been, predictably, getting a great deal of play in the blogosphere today. I’m not a regular reader of US political blogs, so my apologies if I’m making a connection that’s been made elsewhere.

30,000 civilian casualties seems like a very low number. It’s close to the number being offered by the website Iraq Body Count, which takes reports from the press, the US military and the Red Cross, adds them up and reports the numbers. While this sounds like a reasonable way to count war dead, it’s a really, really bad method for getting a comprehensive count of people dead as a result of the US invasion of Iraq.

Basically, the IBC number is the “reported dead”. According to their methodology page, the staff of the Iraq Body Count project take reports of civilian dead in the mainstream press, check to ensure the reports appear in two media sources, and record the numbers of dead reported.

But not every death in a war zone gets reported. If a bomb hits a building and people die, but journalists don’t see it, there’s a good chance those deaths don’t get added to the totals. And these numbers don’t include civilian deaths from less direct causes – an increased number of murders in a post-Saddam Iraq, deaths from illnesses caused by disrupted food and water supplies, etc.

The way you get a comprehensive death toll from a war is to do an epidemeological survey, choosing households at random and interviewing them to see who died in a period before the war and in a period after the war, and what the causes were. Dr. Les Roberts of Johns Hopkins University used this method to count war dead in the Democratic Republic of Congo, and came up with the stunning figure of 1.7 million deaths in two years, only 200,000 from violence. This figure was widely reported and cited as evidence that the bloodiest war since World War Two had been largely ignored outside of Africa.

As it turns out, Dr. Roberts also conducted the same sort of study in Iraq, in the summer of 2004. Surveying 988 Iraqi households (7,868 people in 33 neighborhoods around the country) carefully chosen to provide a statistically significant random sample, Dr. Roberts and his team came up with the stunning news that the death rate in Iraq shot up from 5% before the war to 7.9% in the eighteen months after the war. This rise in death rate implied roughly 100,000 civilian casualties in the war in Iraq.

The respected British medical journal, The Lancet, reported the results of the study, which were picked up by a few US newspapers, but largely didn’t enter the national debate. The timing of the announcement was fairly poor – the numbers were released just before the US presidential election and it’s hard to avoid the perception that Dr. Roberts wanted his results to affect the outcome of the 2004 elections. Right-wing pundits seized on the small sample size of the survey and the long error bars (because the survey relies on statistical extrapolation, the figure of 98,000 dead is the top of a bell-curve distribution of possible accurate figures). Worried the study was partisan and biased, Rob Stein, reporting on the story for the Washington Post, called a military expert at Human Rights Watch, Marc E. Garlasco, who declared, “These numbers seem to be inflated.”

“This American Life” ran an excellent hour-long radio show on the Iraq extrapolation a few weeks back, where they interviewed Garlasco, who admitted that, when asked for comment by Stein, he hadn’t read the study and was giving a “snap reaction” to the figure of 98,000 dead. Since reading the study, Garlasco admits that Roberts methods appear to be sound.

Why does it matter whether there are 30,000 dead or 98,000 dead?
Or 196,000 dead? When Dr. Roberts released his study, estimating 98,000 deaths, the Iraq Body Count site estimated 15,000 dead – using a crude extrapolation forwards, perhaps Roberts’ study, conducted today, would show a doubling in body count.

It’s not just a matter of justice or respect – it’s also a matter of how policy gets shaped in the future. As the US looks into strategies of pulling out of Iraq, it’s possible that, as reported by Sy Hersh in the New Yorker, US troops will be replaced with an increased use of air power.

While I’d be very glad to see our troops come home, increased air power has a price. According to the Lancet study, 84% of violent deaths were due to actions of coalition forces – 95% of those deaths were the result of airstrikes, missles or helicopter gunships. In other words, when we fight the Iraq war from the air, civilians die, and we don’t hear about all those deaths. And if we don’t hear about them, we don’t make plans for troop pull-out with those future deaths in mind.

My friend David Weinberger wants to know why we don’t want to know how many died in Iraq. Part of what’s so moving about the This American Life radio piece is that Garlasco (the guy who trashed Roberts’ study before reading it) joined Human Rights Watch precisely so he could ask the Pentagon that question. His previous job? He was the guy who identified precision bombing targets for the Pentagon. In other words, he had to calculate whether a target could be bombed with a minimum of civilian casualities, which is hard to do when the military doesn’t try to count casualties after a bombing.

It’s hard to have a debate about what to do about the morass we face in Iraq. But it’s lots harder to have it if the numbers we’re working with may be low by a factor of six.

This entry was posted in Media. Bookmark the permalink.

15 Responses to Learning to Count

  1. Thanks for this, Ethan.

    To clarify: I was asking why our administration doesn’t have its own estimate of the number of Iraqi civilians killed and instead relies on “media estimates.” I have no reason to think that citizens don’t care, but I find it appalling that our leaders obviously don’t give a shit.

  2. Ethan says:

    It’s a useful clarification, David – thanks. But until I see more US citizens getting pissed that our leadership doesn’t seem to be counting civilian dead, I’ll continue to be worried that we citizens also don’t care enough about civilian dead.

  3. And I disagree with David’s logic above– I had responded such on his blog.

    Also, Slate’s war correspondent Fred Kaplan was one journalist who criticized the Lancet study— he has been very skeptical of the war and is no “right wing pundit.”

    Lastly, I’ve argued on a Radio Open Source threaD that the metaphysical aspects of the studies are quite different. The Lancet study concluded; The IBC is still being ongoing.

  4. Pingback: links for 2005-12-15 at connecting*the*dots

  5. Ethan says:

    Hi Jon – no doubt that IBC and Lancet are metaphysically different. The Lancet study was designed as a prelim study with the intent of carrying out a larger study at some point in the future (preferably when the country’s safer for researchers.) So there’s the possibility that the Lancet method can be used in the future.

    I’m not saying that everyone who criticized the Lancet study was a right wing pundit. I am saying that I don’t think it got careful enough consideration and that it looks methodologically sound to me.

  6. Jonathan says:

    There was never great media coverage of Saddam’s massacres of his people, such as gasings of tens of thousands, and strapping people to trees as they cut them alive for the public to watch.

    Or coverage of the 200,000 murdered Mongolians, 800,000 disfigured, in the past two decades, under nazi China. Or the hundreds of thousands of forced abortions, while they are awake, of Tibetans. Or 10,000 executions of political enemies.

    What makes you think the public ever cared?

    If they did, they would seek out injustice and fight it.

  7. Ethan,

    I followed your link to the IBC site but I am still uncertain about this count. Does this 30,000 figure include people who may have been ‘enemy combatants’? I know that the figure used in the Palestinian conflict are stretched beyond credibility to include terrorists and ‘collaborators’ (having been murdered by other Palestinians) as Palestinian casualties.

    I agree that civilian deaths are unacceptable where civilians are blameless (rather than active adversaries) and desire an accurate understanding.

    By your own logic if you would add imputed deaths to these numbers is it also reasonable to deduct deaths not suffered any longer as the result of dismantling Hussein’s murderous regime?

  8. Ethan says:

    Hi Stuart – the IBC figure, as I understand it, is the count of every death of Iraqis reported in two or more media outlets. I don’t know how – or if – it draws a distinction between combattants and bystanders – in a guerilla war, that’s a very difficult call to make. I agree with you – some of those deaths are going to involve combattants, even if IBC is attempting not to count them.

    The Lancet report, which is the number I think has gone under-reported, is a pure statistical extrapolation based on interviewing households on deaths that occurred. I doubt the Lancet study is able to meaningfully distinguish between combattant and non-combattant death… that would require some percentage of households to acknowledge that their lost relatives were killed while participating in the insurgency, which is a hard question to ask in a door to door survery.

    Your suggestion in the last paragraph is actually very consistent with the Lancet survey strategy. Because the survey asks people to list deaths in their households for two years prior to the invasion and for two years after, it should register deaths from Saddam’s government. In other words, the Lancet study is seeing a significant increase above that baseline.

  9. hi ethan,
    i think it is best to follow the ‘hospital-morgue-trail’ numbers.
    for example info like this:
    IRAQ: “Morgue can’t cope with numbers of bodies arriving on a daily basis, doctors say”
    http://www.irinnews.org/report.asp?ReportID=50700&SelectRegion=Middle_East&SelectCountry=IRAQ

  10. Danny Yee says:

    Anyone curious about the Lancet study should check out Tim Lambert’s blog Deltoid. He has a whole section devoted to posts debunking bad media coverage of the study.

  11. Pingback: …My heart’s in Accra » DRC - 3.8 million dead. Darfur?

  12. Pingback: …My heart’s in Accra » Is Israel a problem for the Democratic Republic of Congo?

  13. Pingback: …My heart’s in Accra » New estimates on deaths in Iraq

  14. Pingback: …My heart’s in Accra » 5.4 million dead in Congo. Believe it or not, it could have been worse.

Comments are closed.