Iraqi Mortality Studies

A new epidemiological study of Iraq by the Iraqi government and the World Health Organization sees 151,000 or so dead through violence since the Iraq War began and "also found a 60 percent increase in nonviolent deaths -- from such causes as childhood infections and kidney failure -- during the period." The news account contrasts this with the earlier statistical survey that found 655,000 "excess deaths" during the war period, of which 601,000 were from violence.

Unfortunately, I don't understand from the coverage how different the total fatality rate in these two studies is. In other words, how much of the disagreement is about what to attribute an increase in the death rate to (violence or other) and how much is about disagreement over how many people died. This section implies, however, that the disagreement is primarily over causes:

Les Roberts, an epidemiologist now at Columbia University who helped direct the Johns Hopkins survey, also praised the new one. While both found a large increase in mortality, his found that much more of it was caused by violence.

"My gut feeling is that most of the difference between the two studies is a reluctance to report to the government a death due to violence," he said. "If your son is fighting the government and died, that may not be something you'd want to admit to the government."

Further difficulties include the fact that many people have been displaced by the conflict and also I assume that some entire households have been wiped out. It remains noteworthy to me that while the US military insists that it takes measures to minimize the civilian death toll, it doesn't take any measures to quantify the civilian death toll, which makes it impossible to know what their measures accomplish. Step one in trying to increase blog traffic, for example, is to measure blog traffic.

Comments

Step one in trying to increase blog traffic, for example, is to measure blog traffic.

Gosh, Matt, maybe you should go into the consultant business. All consultants ever seem to know is that nothing exists unless it can be counted.

But this is just wrong. Step one in trying to increase blog traffic is to post pictures of nude women. Step one in minimizing the civilian death toll is to stop killing civilians.

Posted by peeposaur | January 10, 2008 9:54 AM

Of course, the fact that the Lancet study has been completely discredited does not even seem to have entered into the left wing blogosphere's consciousness. But, hey, it's not like the Reality Based Community actually cares about, well, reality.

Posted by Al | January 10, 2008 10:02 AM

Because, you know, murdering "only" 150,000 innocent people for no good reason is just fine, in Wingnut World.

Posted by Steve LaBonne | January 10, 2008 10:08 AM

Also, the two studies don't seem to be measuring the same thing at all. While the Lancet study broke down deaths by violent/non-violent, I don't believe they distinguished civilians from combatants. At which point saying that "Your study showing that 600K people died violently is wrong because my study shows that only 150K civilians died violently" seems kinda dimwitted.

Posted by LizardBreath | January 10, 2008 10:16 AM

completely discredited

Eh, not so much.

Posted by Matt Weiner | January 10, 2008 10:28 AM

Thanks, Al. Some lawyer read some right wing news sources that claim that the Lancet study has been discredited (which is exactly what we would expect them to say) and then he comes here to let us all in on the secret. For those of us who do live in the real world, we know that this issue has been hotly debated. Unfortunately, the methods of statistics are not particularly accesible to most of us - at the end of the day I depend a lot on expert sources. My sources say the study was fine. I realize you have your own sources, I just question their judgement based on their track record (and yours, subsequently).

Posted by mpowell | January 10, 2008 10:30 AM

I don't know the methodology for this most recent study, but if I recall correctly the Lancet study was criticized because it wasn't completely random; homes near major roads were more likely to be surveyed. Given that you could reasonably assume violence would be more prevalent near major thoroughfares, it would skew the mortality higher than a truly random survey would show.

Posted by omellet | January 10, 2008 10:34 AM

Yes, the alleged "discrediting" of the Lancet study is just standard wingnut BS. And LizardBreath just explained why the two studies are not necessary in conflict at all.

But morally speaking it shouldn't even matter, as I tried to point out. Unless you think we should stop prosecuting murderers who kill "only" one victim.

Posted by Steve LaBonne | January 10, 2008 10:35 AM

The IFHS and Lancet studies measured exactly the same thing. The are both household cluster sample surveys of "excess" mortality from the invasion. The Lancet found 655,000 excess deaths through June 2006, of which 601,000 were from violence. The IFHS study found 151,000 excess deaths from violence for the same period--a horrific toll, abeit one fourth the Lancet total. The IFHS sample size was bigger, and the geographic distribution and trends of violence in their study generally conform to non-Lancet estimates (although the total magnitude of deaths in the IFHS is higher). In short, the IFHS is a far superior study.

Also, neither the Lancet or the IFHS study distinguished between civilians/noncombatants and combatants. The only group to specifically count civilian deaths in Iraq Body Count, which estimated 48,000 deaths based on media reports through June 2006 (the current IBC total is around 87,000 as of Jan 2008). Based on estimates of combatant casualties (e.g., Iraqi Army during the invasion, insurgents, and Iraqi security forces since the invasion) provided by independent groups and the U.S. military, about 26,000 combatants were killed through mid-2006. That means that the IBC total through mid-2006 plus combatant estimates suggest 74,000 total deaths at that time (both civilians and combatants). If these ratios are applied to the IFHS study, it suggest that the IBC media-based estimates of civilian deaths are off by about half . . . meaning the current civilian death toll may be as high as 170,000.

Posted by CK | January 10, 2008 10:47 AM

The new NEJM study wasn't able to survey 11% of its proposed areas because it was too dangerous (and, as a corollary, I would say that dangerous areas are likely to be ones with the highest violent death rates). Instead, it relied on Iraq Body Count (IBC) data for those areas.

Yet the IBC obviously under-reports deaths (eg. if a tree falls in the forest and no one is there to hear it, does it make a sound? IBC says it didn't fall at all...), and most likely, seriously so.

Therefore, the NEJM study is substituting data that seriously under counts violent deaths in the very areas where we would expect violent deaths to be highest.

That might be just one reason why Lancet and NEJM don't jibe.

Posted by McKingford | January 10, 2008 10:50 AM

Given that you could reasonably assume violence would be more prevalent near major thoroughfares, it would skew the mortality higher than a truly random survey would show.

I could? I confess to knowing squat about this stuff, but that seems sort of questionable on its face. Wouldn't major thoroughfares be patrolled more often? And even if there is more violence along a thoroughfare, wouldn't the fact that it's more densely populated cancel that out to a certain extent? If I live on a block with 50 households and there are 10 acts of violence (assuming each is against a member of a different household) in one week, my household only has a one in five chance of being victimized. If I live on a block with 5 households and there are only 2 acts of violence in one week, my household is twice as likely to be victimized as under the previous scenario.

Again, I have no background in this stuff. But it doesn't seem like a slam dunk to me that over-sampling households on major thoroughfares will skew the death count upwards. I could be persuaded by evidence, of course.

Posted by Jake H. | January 10, 2008 10:51 AM

I've read someplace that the IFHS study couldn't survey some areas of Iraq because they were to dangerous for the personnel to enter. If true, that would suggest that IFHS study underestimates mortality due to violence, given that one of the distinguishing characteristics of "dangerous areas" is that they have relatively high levels of mortality due to violence.

But I think that a better explanation is that the Lancet study was run by bunch of pansy Islamofascist homosexuals who hate freedom.

Posted by afferent input | January 10, 2008 10:56 AM

Also:

Step one in trying to increase blog traffic, for example, is to measure blog traffic.

Or you could just post something negative about Ron Paul...

Posted by afferent input | January 10, 2008 10:59 AM

The IFHS sample size was bigger, and the geographic distribution and trends of violence in their study generally conform to non-Lancet estimates (although the total magnitude of deaths in the IFHS is higher). In short, the IFHS is a far superior study.

The IFHS (aka NEJM) study determined a pre-war death rate 1/3 that of the WHO estimate. It also shows a *steady* rate of violence from 2003-2006. So, in both respects, it does not conform to non-Lancet estimates (or, ahem, *reality*).

And, since the IFHS study was conducted by Iraqi government representatives (the Lancet study was not), it likely suffers from response bias, as respondents are less likely to want to acknowledge a violent death to a government worker.

Posted by McKingford | January 10, 2008 11:07 AM

Well, the definitional causes of "excess deaths" are always a little fuzzier than the actual numbers of such deaths.

For example, I think it's pretty solidly established that Soviet Communism killed at least about 25M civilians. But my impression is that only about 10% or so were probably killed violently, e.g. nine grams in the back of the head, the rest dying by hunger, cold, or disease.

And the really stupid/crazy economic policies of Mao's Great Leap Forward probably killed about 35M people or so, but my impression is that almost none of these---maybe 1%---died "violently."

Similarly, most people seem to say that around 3M civilians have died in Congo during recent years because of the civil war. But I doubt that more than a tiny, tiny fraction were actually "violently killed."

Posted by RKU | January 10, 2008 11:12 AM

The comparable number you get from the new survey to the Lancet 650,000 excess deaths, is 420,000 excess deaths.

Posted by Tim Lambert | January 10, 2008 11:17 AM

RKU,

See also Darfur, where, for all the hysteria about genocide, it is only a fraction of the deaths there that are the result of direct violence.

But on the issue of Iraq, both the IFHS and Lancet study confirm that *violence* is the leading cause of death for those aged 15-59, so, "excess" deaths are mostly violent ones...

Posted by McKingford | January 10, 2008 11:19 AM

Matt, that analogy doesn't hold water. In one case, you're trying to increase blog traffic, and in the other you're trying to decrease civilian deaths. Not the same at all.

Posted by Mac | January 10, 2008 11:26 AM

Yes, Matthew, it truly gives the lie to the American military establishment's stated objective of minimizing civilian casualties that they take essentially no steps to quantify them. Two-faced, lying mothers, sad to say. And I grew up in the service, so f**k all of y'all that take umbrage.

Posted by DCBob | January 10, 2008 11:28 AM

And you could count on one hand the percentage of Americans who give a rusty about the near-holocaust we have inflicted on innocent Iraqis.

The Americans evidently dropped 40,000 pounds of bombs yesterday in that new offensive. Wonder how many civilians they bagged.

Posted by bob h | January 10, 2008 11:50 AM

The IFHS used IBC data to impute figures for missing clusters, but they did not use the IBC data itself. Instead, they used IBC's estimates of the distrubtion of fatalities across Iraq's provinces to estimate the ratio of deaths and then apply these ratios to estimate deaths in areas not covered by the survey.

Hence the comment on the study in the WaPo article today: "Overall, this is a very good study," said Paul Spiegel, a medical epidemiologist at the United Nations High Commission on Refugees in Geneva. "What they have done that other studies have not is try to compensate for the inaccuracies and difficulties of these surveys, triangulating to get information from other sources."

Posted by CK | January 10, 2008 11:50 AM

It's important to understand what's going on when Al says stuff like "completely discredited". Since further academic review of the Lancet/Johns Hopkins study endorsed its methodology, it's apparent that Al is referring to the fact that in right-winger's minds saying the words "completely discredited" is all that's required. It's a mix of magic and a kind of urban folklore reports of Muslim divorce: saying the words is all that's important.

Posted by Jeffrey Davis | January 10, 2008 12:16 PM

Thinking a bit more about this "definitional" dispute raising an interesting related possibility.

It's important to realize that every objective policy analyst will admit that 20th Century Communism achieved some very, very good and positive things in many places. The biggest perceived minus is that it killed about 100M innocent civilians.

Now it seems that the argument made by Al and his friends is that this widely-accepted figure for the Communist body-count should be reduced by about 95%. Such a huge downward revision would certainly cause me to heavily reassess my own view of the net merits of Communism.

So maybe there's something else going on here. Everyone knows that the bulk of the noisiest neocons are all either former Communists or at least come from a Communistic family background. (For example, Bill Kristol's father and mother first met at a meeting of young Trotskyite Communists.)

So perhaps Al and his neocons are therefore really aiming at rehabilitate the public image of Lenin, Stalin, Trotsky, and Mao...

Posted by RKU | January 10, 2008 12:45 PM

Here's the NEJM article presenting the IFHS results. I have a fundamental comprehension problem with it.

The third paragraph of the Abstract says: the overall rate of death [in Iraq] per 1000 person-years was 5.31.

Compare that with the overall rate of death in the United States in 2004, per the National Center for Health Statistics:

Number of deaths: 2,397,615

Death rate: 816.5 deaths per 100,000 population

Life expectancy: 77.8 years

Correct me if I'm wrong, but 816.5 deaths per 100,000 population in one year = 8.165 deaths per 1000 person-years, or more than 1.5 times the Iraqi mortality rate.

Even before the war, that wouldn't have made sense. But given the war, it is completely incomprehensible.

Does anyone have an explanation for this? Because I sure don't.

Posted by low-tech cyclist | January 10, 2008 1:14 PM

Few people remember that the original Lancet study had a very large uncertainty. The authors cited a 95% confidence interval between about 400,000 and 940,000 excess deaths. The overly precise number reported by the press (654,965) was simply the center point of that range.

Posted by Jinchi | January 10, 2008 1:16 PM

Does anyone have an explanation for this? Because I sure don't.

The Lancet study authors responded to this point directly, citing U.N. numbers:

[M]ost Middle Eastern nations really do have lower death rates than most European countries, and in fact have lower death rates than 5.5. Jordan's death rate is 4.2, Iran's 5.3, and Syria's 3.5. The reason for the lower rate is simple: Most Middle Eastern nations have much younger populations compared to most Western nations. Obviously, the elderly die at a greater rate than young people and account for a disproportionate number of deaths.

http://www.slate.com/id/2154203

Posted by Jinchi | January 10, 2008 1:29 PM

low-tech cyclist, there's no great mystery there. The proportion of old people in the US population is greater than it is in Iraq's.

Posted by Kevin Donoghue | January 10, 2008 1:30 PM

CK,

I don't think there is any dispute that IBC under-reports violent deaths (the only issue is by how much). Having said that, I don't see how using IBC data, in either raw or imputational form, will not cause an under representation.

Nevertheless, I don't want to go overboard in criticizing the new study. All we have is the abstract - the study itself won't be published until the end of the month. Unlike the neocons, I am not wedded to conclusions, and am open to the possibility that this study is, in fact, better. But I am not yet prepared to dump on the Lancet study, since I see two major problems with the IFHS study that will need to be more fully hashed out: a) the use of IBC data, and b) response bias arising from the use of Iraqi government surveyors.

As for the Speigel quote, I have serious issues with his objectivity, given that he also said (about the new survey): "this does seem more believable to me". Believable? WTF? It's one thing to say a study is better because it is more representational or complete or uses better criteria. But to laud it for a conclusion that most closely jibes with one's subjective preconceptions is utterly unscientific and, well, unobjective.

Posted by McKingford | January 10, 2008 1:37 PM

Jinchi and Kevin - that's interesting, and their cites of the known mortality rates of nearby countries that aren't at war are convincing.

But I'm still curious: wouldn't most countries with a much younger population than ours have such a distribution, at least in part, because people die at a much younger age than in the U.S.? There's no question that a much younger population, together with the same life expectancy, would result in a much lower death rate. It's just that I wouldn't expect to find the two together.

I'm not still arguing the prewar death rate. I'm just wondering why it's true, not questioning its truth.

Posted by low-tech cyclist | January 10, 2008 1:40 PM

Few people remember that the original Lancet study had a very large uncertainty. The authors cited a 95% confidence interval between about 400,000 and 940,000 excess deaths. The overly precise number reported by the press (654,965) was simply the center point of that range.

This is a red herring, because the Lancet methodology is consistent with sampling generally.

You know when you read opinion polls, for instance, that say Obama is at 37%? The small print you don't read is that there is a 95% confidence level of 37%, but that the possible intervals is 16% to 58% (but only a 1 in 20 chance).

The large discrepancy is *not* the product of (or the same as) a "very large uncertainty". Quite the opposite: there is a very large possible variable, but a high confidence (ie. 95%) that 37% is the true number (+/- 3%). Nobody ever says "Obama could actually be at 58% - the polls say it's possible!", even if this is technically true.

Posted by McKingford | January 10, 2008 2:01 PM

Sorry to follow up...

Nobody ever says "Obama could actually be at 58% - the polls say it's possible!", even if this is technically true.

Nor does anyone discount the likelihood that Obama *is* at 37% because of the large possible (but unlikely) interval.

Posted by McKingford | January 10, 2008 2:09 PM

wouldn't most countries with a much younger population than ours have such a distribution

Not necessarily. That depends on a combination of birth rates and death rates (not to mention migration in and out of the country). If couples in the U.S. started having an average of 4 or 5 children each our population would become much younger on average.

What you really need to answer your original question is a comparison of the death rates for specific age groups. If you randomly selected 1000 Americans and 1000 Iraqis you'd find that more of the Americans had died one year later. But if you randomly selected 1000 Americans and 1000 Iraqis (aged 18-25) you'd find that more of the Iraqis had died a year later.

Iraqis don't make it to old age as often as we do and that, ironically, keeps their overall death rate lower. (Well it did before the war anyway).

Posted by Jinchi | January 10, 2008 2:14 PM

Of course, the fact that Al is completely discredited can be easily measured by reading every one of his comments.

Deltoid has good discussion, as ever. And those who aren't complete fucking hacks will look at the methodology, rather than making bullshit claims of fraud from day one.

But what all the surveys show, in basest terms, is that the choice of words to describe Iraq from a humanitarian standpoint is somewhere between 'disaster' and 'catastrophe'. Unless you're a total fucking hack like Al.

Posted by pseudonymous in nc | January 10, 2008 2:17 PM

McKingford, I don't know where you get your polling numbers from, but the Lancet study itself:

For the single most important category— the total number of deaths by violence during the war—the confidence interval ranges from 426,369 to 793,663. That means that we are 95% certain that the correct number is between those two, and 601,027, is the statistically most probable number. The likelihood that another number is the correct number decreases very rapidly as one moves up or down from the figure of 601,027.

I realize that doesn't mean that 426,369 is as likely as 601,027 - but it's still a pretty huge range (and the likelihood of it being exactly either of those numbers is approximately zero). It's a statistical measure that also assumes no fundamental flaws in the methodology itself. If you prefer error bars drawn at the 65% or even 50% confidence level we're still talking about a significant uncertainty.

Posted by Jinchi | January 10, 2008 2:34 PM

Tim Lambert,

The comparable number you get from the new survey to the Lancet 650,000 excess deaths, is 420,000 excess deaths.

What is the basis for this claim? The authors of the new study state:

The best estimate on the number of deaths from March 2003 through June 2006 based on the IFHS data analysis and comparisons with other sources is ... only one fourth of that reported by Burnham et al. [in the Lancet study]

Posted by Mixner | January 10, 2008 2:45 PM

Jinchi,

The Lancet study authors responded to this point directly, citing U.N. numbers

And Fred Kaplan responded to their response, citing the U.N. estimate of pre-war Iraqi mortality of 10, almost double the 5.5 value the Lancet authors use.

Posted by Mixner | January 10, 2008 3:47 PM

Keep in mind that this 151,000 violent civilian deaths figure is just for the first 5/8ths of the war (so far), so the total for all 4.8 years and counting would be approaching a quarter of a million.

Posted by Steve Sailer | January 10, 2008 3:47 PM

I estimate, from testimony before Congress, that the U.S. military has been firing about 275,000 bullets in combat per _day_.

http://isteve.blogspot.com/2007/08/how-many-bullets-are-we-firing-in-anger.html

Posted by Steve Sailer | January 10, 2008 3:52 PM

I blogged at length in October 2006 about the controversial Lancet study that claimed 600,000 deaths by violence in Iraq. I eventually realized it was probably exaggerated because, paradoxically enough, Iraq was too violent for the study to have been carried out the way the authors claim:

"Maybe what happened is that the interviewers didn't actually go much door-to-door at random, but instead arrived in a neighborhood, put the word out, and then either had people who wanted to talk to them come see them or were invited to the homes of people who wanted to see them. That might account for the very high % of people with death certificates available.

"Or it could be that the interviewers got in contact ahead of time with neighborhood leaders to see if their presence would be welcome to reduce their chances of being killed. (That's not good random surveying hygiene, but are you going to blame them?) Then, in a neighborhood where the local big shot wanted their presence, he might have passed the word around to aggrieved families to get ready to tell their stories to the interviewers when they showed up. This could cause a bias upward in the number of deaths reported.

"The more I think about the mechanics of carrying out the survey on the street without getting killed, the more I suspect that the Iraqi interviewers didn't actually implement the purely random survey design that the American professors from MIT and Johns Hopkins dreamed up for them. It would be nuts to to let luck determine which streets you'd choose, as the report claims they did. You'd want to only go where you knew you'd be safe. Then you'd tell the Americans you did exactly what they told you to do."

I have not subjected the plausibility of the new study to extensive analysis.

Posted by Steve Sailer | January 10, 2008 3:54 PM

Nature's online site has a report discussing the latest study: http://www.nature.com/news/2008/080109/full/news.2008.426.html

Key graphs:

Jon Pederson of Fafo, an Oslo, Norway-based organization that gathers information about living and working conditions, helped run a similar survey in Iraq in 2004. This United Nations-led project looked at 22,000 households and estimated that, up to then, between 18,000 and 29,000 had died.

Pederson says the WHO survey was methodologically robust, but its results are a little on the low side of what he would expect. Household reporting always underestimates mortality compared to a straightforward body count, which is very difficult to do in times of conflict, he notes.

and:

Guha-Sapir has recently published a working paper analysing many estimates — including the Iraq Body Count, Johns Hopkins and Fafo surveys — and has come up with a death toll of around 139,000. From that basis, she thinks the WHO’s total is about right.

But, she adds, she thinks the WHO's sample is too small to come up with a definitive number, especially in Iraq where many people have been displaced, and violent deaths tend to be clustered in small areas. “The sample size is not really good enough to represent extremely variable conditions,” she says.

Posted by Jinchi | January 10, 2008 4:25 PM

The 150,000 figure is for violent deaths only. Total excess mortality for the period (which includes both violent & non-violent deaths), according to this study, is about 400,000.

Posted by Peter H | January 10, 2008 4:41 PM

McKingford said: As for the Speigel quote, I have serious issues with his objectivity, given that he also said (about the new survey): "this does seem more believable to me". Believable? WTF?

This is one thing I used to stress, back in my math-teacher days: to not just follow the calculations blindly, but have some reality check for them at the end. For instance, if you're integrating a function f(x) from a to b, check your result against (b-a)*(max of f on [a,b]), because you should know that if it's bigger than that, it's not believable.

So you should take a look at your numbers, see the implications of them, see if that's believable.

A case in point:

Jinchi said: The Lancet study authors responded to this point directly, citing U.N. numbers

Mixner responded: And Fred Kaplan responded to their response, citing the U.N. estimate of pre-war Iraqi mortality of 10, almost double the 5.5 value the Lancet authors use.

If the background mortality rate was 10/1000 rather than 5.5/1000, it would make a larger number of excess deaths more plausible, rather than less so, from a straight commonsense point of view.

Let's say the prewar population of Iraq was 27 million. If the prewar mortality rate was 5.5, the mortality rate needed to increase by 132% on account of the war to account for 655,000 excess deaths. But if the prewar rate was 10, then it needed to increase by 'only' 73% to account for that number.

I can't tell you whether either one's inherently believable, but an increase of 73% or more is inherently more believable than one of 132% or more.

Posted by low-tech cyclist | January 10, 2008 5:02 PM

If the background mortality rate was 10/1000 rather than 5.5/1000, it would make a larger number of excess deaths more plausible, rather than less so, from a straight commonsense point of view.

Huh? The higher the pre-war mortality rate, the lower the "excess" number of deaths for a given post-war mortality rate. As Kaplan points out, using the Lancet study's post-war mortality rate of 13.3, if the pre-war mortality rate were 10 (the U.N. estimate) rather than 5.5 (the Lancet authors' estimate), that lowers the "excess" mortality rate from 7.8 to 3.3.

Jon Pederson of Fafo, an Oslo, Norway-based organization that gathers information about living and working conditions, helped run a similar survey in Iraq in 2004.

That would be the same Jon Pederson who worked for the United Nations Development Corporation, and who Kaplan cites to suggest that even the U.N estimate of the pre-war mortality rate may be too low. Again, the higher the pre-war mortality rate, the lower the "excess" mortalities for a given post-war mortality rate.

Posted by Mixner | January 10, 2008 5:28 PM

Re: I don't think there is any dispute that IBC under-reports violent deaths

Seems to me the opposite would be true: violent deaths, even in chaotic conditions, are going to stand out more than deaths from "natural" causes. I also wonder at this insistence that most deaths have to be due to violence when it's always been true, even in modern times, that most civilian war deaths are the result of the disruption of war, leading to starvation and disease. Iraq was already in rough shape in that respect in 2003 due to the embargo and Saddam's policies. How can anyone doubt that the aftermath tippped an awful lot of people over the edge of survival and this is where the true death toll is coming from? And it's an easy one to miss since these are "natural" causes after all. People generally take a lot more notice of a murder than they do of someone slipping away from, say, typhoid or pneumonia.

Posted by JonF | January 10, 2008 6:11 PM

Estimated mortality due to violence of YR 2003 Iraqi household members exceeds 3%. That from a UN-sponsored recent survey of refugees living in Syria. (See table on last page.)

http://www.mcclatchydc.com/iraq/story/23159.html

Unfortunately, the sample was small, and details of the sampling method are absent; but the extremely high mortality is consistent with what might be expected among impoverished refugees.
I know of no other estimates from this segment of the Iraqi population. It would be wonderful for a Bayesian to construct some plausible synthetic estimates from the several sources. But, it is wondrous enough that the current estimate of excess mortality is such a relief to those at pains to amateurishly discredit the 2006 estimates that they are silent on how the new estimate hilights the the incompleteness of IBC numbers, which are 50% lower.

http://media.mcclatchydc.com/smedia/2007/12/14/16/IPSOS-II-Survey-Dec07.source.prod_affiliate.91.pdf

Posted by DemostiX | January 10, 2008 6:19 PM

The problem I have with this study is the involvement of the Iraqi Ministry of Health which has consistently underreported the level of violence in the country month to month for years.

They are a severely biased and incompetent source of figures.

So unless this study was sufficiently moderated by WHO personnel, I find it highly unlikely to be accurate.

Still, I will wait for the Lancet people to release some sort of comprehensive analysis of the study's methodology and results before I dismiss it.

The problem is that any lower figures, while nice for the Iraqis, will be seized on and compared to the mythical Saddam kill rate as somehow "justifying" the war by the right wing nuts.

Posted by Richard Steven Hack | January 10, 2008 9:58 PM

As for the pre-war mortality rate, keep in mind that we're talking about a rate based on the effects of sanctions allegedly responsible for killing half a million or more Iraqis.

So to suggest that the pre-war rate is somehow "normal" is stupid in the extreme, unless that pre-war rate is ALSO pre-sanctions.

Posted by Richard Steven Hack | January 10, 2008 10:14 PM

A corollary occurs to me: If the pre-war or even pre-sanctions rate was so low that it disrupts the notion that the war was bad for the Iraqis, then what happens to the justification that Saddam was such a bad guy because he was killing all the Iraqis before the way?

If his killing all the Iraqis didn't show in the pre-war mortality rate, then how does the post-invasion mortality rate figures justify the invasion?

Or are the right wing nuts going to argue that the reason the pre-war mortality was so high was BECAUSE of Saddam killing all the Iraqis? I still don't see how that justifies the war that RAISED the pre-war mortality to current levels.

Slight problem here for the right wing nuts, but I'm sure they'll cope with some bullshit.

Posted by Richard Steven Hack | January 10, 2008 10:18 PM

What I find funny are the people who complain about the involvement of the Iraqi ministry in this study, but who don't care at all that a Saddam-era Baath official was part of the Lancet group (indeed, the Baathist was the sole person responsible for actually conducting the data gathering).

Of course, the left wing was always just fine with the Baathists. It is people actually democratically elected that the left have problems with.

Posted by Al | January 10, 2008 10:34 PM

References, Al, references.

Provide them or blow away.

Posted by Richard Steven Hack | January 11, 2008 12:08 AM

What I find funny are the people who complain about the involvement of the Iraqi ministry in this study, but who don't care at all that a Saddam-era Baath official was part of the Lancet group (indeed, the Baathist was the sole person responsible for actually conducting the data gathering).

This is what Les Roberts, one of the co-authors of the Lancet, says in a letter to the Wall Street Journal:

Your suggestion that our Iraqi colleague Riyadh Lafta was suspect because he recorded child mortality during his career is particularly ironic. He was one of few professors in the country that never joined the Baath Party.

Posted by Peter H | January 11, 2008 1:06 AM

Yup, nitwit Al believed another WSJ editorial rather than check the facts.

Typical right wing nut behavior.

And we're supposed to treat anything they say as somehow credible.

Posted by Richard Steven Hack | January 11, 2008 4:33 AM

Step one in trying to increase blog traffic, for example, is to measure blog traffic.

See, that's where you're wrong. Step one in trying to increase blog traffic is to talk more about Ron Paul.

Posted by brooksfoe | January 11, 2008 11:21 AM