Asymmetrical Information

"Germany, with a prewar population of just about 80 million, suffered 1.8 million civilian deaths during six years of invasion", I'd like to point out that the Germans were doing most of the invading - the Sudentland,Czechoslovokia,Poland (shared with Russia),Demark, Norway, France, Belgium, Netherlands, Greece, etc. + Italy into North Africa. The bomber war really didn't get going until mid-1943.

And the death rate among the Russian population, civilian and military, was many times greater than any other belligerent.

"Germany, with a prewar population of just about 80 million, suffered 1.8 million civilian deaths during six years of invasion"
....
"How likely is it that Iraq has lost a higher percentage of its civilian population in three years--especially given the vast advances in medical care,?"

I do not know. But here is one possibility: Germany had a much higher percentage of its pop. in the miliarty. So those deaths would not be counted as civilians. And many would have occured in battles outside Germany. Plus many civilians were Jews and Roma killed internally via the holocaust. Also not counted. In otherwords, not as many civilans left to kill. In Iraq, there is more or less no state-supported military nor is there a holocaust. So all deaths can be resported as "civilian". Also, many deaths might not be iraqi, but mujahadeen from other countires.

So the figues can be seen as reasonable.

But either way, its still a quagmire.

Germany lost, IIRC between 4-8 million troops in WWII. That still leaves somewhere between 72 and 75 million civilians to kill.

Another peculiar thing - many parts of Iraq are (relatively) peaceful, for instance the Kurdish zone and the Shiite areas around Basra.

It'd be a hell of a lot more convincing if he stated which parts of the study's methodology he disagrees with.

The numbers may be too high but does it make any difference? We already know there are tens and probably hundreds of thousands of excess deaths. If the real number is 300,000 instead of 650,000 do your conclusions about the war change?

Plus I'd love to read a critique that is based on something esle than "it sounds strange to me".

Some idle speculations...

Another possibility (this is just idle speculation) is that the other estimates for German WWII noncivilian deaths or American Civil War deaths were not analogous to the Lancet paper. The Lancet paper is I think is fundamentally a delta-- all deaths minus projected deaths, whereas I would guess that the other estimates may have been attempts at a direct count of casualties directly attributable to war. As such, there may well be "marginal" deaths that would be attributed to the war in the Lancet study, but wouldn't be in the other studies.

Here is one hypothetical: what would such a study show for *US* civilian casualties? At first blush, it would seem to be zero, but such a study might detect a non-zero death rate, which could possibly be real (either something direct, such as an increased suicide rate due to war stressors, or a cascade effect such as diminished health expectations due to military family loss of income or stresss effects on the general civilian population.

Even if such a study showed a non-zero death rate, would it really mean anything? I think the answer to that would depend from person to person. Heck, if it showed a negative death rate, I wouldn't be surprised either.

Dean--suicides and other stressor deaths go down during wartime (absent serious material deprivation). . . people stop focusing on themselves, and social solidarity makes people healthier.

Dean, AFAIK the numbers were derived by comparing pre- and post- war populations. No one was keeping tallies on how many German civilians they killed.

Are there any numbers for total deaths in Iraq during that period? Are 1% of the deaths in excess or 50% or 90%?

About my probably bogus explanations for a hypothetical US civilian casualty rate (how ethereal can I get?)--

I don't really doubt that you are almost certainly correct, that such an analysis would show a negative civilian casualty rate. What was more interesting to me is: how do we interpret this number? If it were negative, does this mean war has a positive side benefit that we should care about? If it were positive, should that get added as a cost to the war? Should we really associate a statistical trend to the war itself?

The reason why I think that the above is a reasonable intellectual exercise is that I think it shows the difficulty of trying to figure out what the Lancet study means. Personally, I suspect that the science behind it is a best faith attempt, and that the number, by which I mean this delta, is probably as good a number as you can get using this approach. More importantly, what does this number mean? I think that it probably doesn't measure civilian war deaths in the way we think about it, but rather something else. Here is (yet another annoyingly hypothetical) an example-- let's say that pre-war, the death rate due to petty crime consequences in Iraq was very low due to draconian law policing, but that in the post-war period, due to a less authoritarian approach to crime prevention, that death rate is much higher. Is that a war casualty, or is that a more general societal effect? I know that the current situation makes talking about that in a detached way really hard, but I think it would be reasonable to disagree about the meaning of that.

As a postscript, I suspect that the Lancet's study is not comparable to the sources for the numbers that you mention if only because epidemiology methodologies have changed and I suspect improved in the last hundred years.

Jane-- read your second update, and that I do find persuasive: it is quite possible that the error bar is far larger than Lancet is admiting too. I still think that even if the methodology were rock solid, we would still have a hard time figuring out what it meant.

Jane Galt, The Lancet study does not claim the excess deaths are all civilian, in fact they are largely men of military age. So why are you comparing to civilian deaths in Germany?

Another possible problem with "excess deaths" is that you shouldn't trust the pre-war Iraqi death rates. As it stands, the reported death rate was dramatically below the US, UK, Germany, France, Brazil, and Mexico. (8.1,10.3,10.3,9.0,9.3,10.1 per 1000) {I used the easiest to find rates over the past four years for each country, which isn't strictly sound but considering the lack of huge disasters in any of the reference countries I suspect isn't the cause of an error}. The reported Iraq number appears to be around 6.4.

When I was looking at the numbers originally I suspected it might be a function of age, which is why I added two relatively younger countries--Mexico and Brazil.

So under the sanctions regime (which was allegedly killing thousands) Saddam had apparently a drastically better death rate than the US, UK, Germany, France, Brazil or Mexico. That is drastically as in 20% better than the US and 35% better than the UK and Germany. (Am I doing that calculation right? Normally I'm math-confident but that seems ridiculous.)

We'll know how many excess deaths there were when Iraq calms down, the refugees return, and they get a decent census; not before.

Was there a good prewar census with which to compare it? It seems as if it would have been difficult to conduct a reliable census given the Baathists' desire to overstate the Sunni population, the lack of access to Kurdistan, the desire on many people's parts to keep their heads down and not catch the regime's notice, etc.

Apparently we're killing more people in Iraq than are dying there:

http://ace.mu.nu/archives/200298.php

Defeat Zombie Terrorism.

Jane,

I don't know Medpundit but the previous Lancet study was by no means discredited. I know a lot of right-wingers said a lot of things but Daniel Davies at Crooked Timber showed most of them simply did not understand the statistics behind the study. I'm thinking of Shannon Love from Chigao Boyz but there were several others as well.

Want to update my death rate comparison. The numbers used by the Lancet study are even lower than the numbers reported by Saddam.

According to page one of the report:

A sample size
of 12 000 was calculated to be adequate to identify a
doubling of an estimated pre-invasion crude mortality
rate of 5·0 per 1000 people per year with 95% confidence
and a power of 80%, and was chosen to balance the need
for robust data with the level of risk acceptable to field
teams.

Their reports 'found' a death rate of 5 per 1000. Therefore they find a pre-invasion Iraqi death rate of 1/2 of the German, French, Mexican and Brazilian rates and 2/3 of the US death rate.

Germany, with a prewar population of just about 80 million, suffered 1.8 million civilian deaths during six years of invasion ...

If you are drawing this from the Wikipedia figure, that's not what the sources say:

Civilian war dead include 360-370,000 killed by Allied Strategic bombing within the 1937 German boundaries. About 600,000 Victims of Nazi persecution from 1939 to 1945 including 300,000 political prisoners. Rűdiger Overmans estimates 1,100,000 German civilian dead in eastern Europe as the result of the 1945 military campaign and the expulsion of Germans after World War II. Postwar deaths of about 500,000 due to famine are not included in these losses. The genocide of Roma people was 15,000 persons. Jewish Holocaust victims totaled 160,000.

In short, this tally includes part of the Nazi genocide and selected types of collateral impact on civilians, not total non-military deaths.

Indeed, if you think about it, it would be unprecedented if a country of 80 million people at peace suffered 1.8 million civilian deaths over six years. Assume a 0.8% death rate (Sebastian Holsclaw's post gives the current US death rate as 8.1 per thousand). Then you get 0.8% x 6 = 4.8% x 80 million = about 3.8 million dead, at least twice that number.

Turning to Sebastian's problem: how is it that the Lancet death rate in pre-war Iraq seems too low? That's an interesting question but I think seeing the details of his calculation of the 6.4 number might help me on that.

Medpundit's last paragraph really does make the best point - "as good as we can do" simply ain't the same as "reliably accurate".

In any other context, I'd expect mostly-to-almost-entirely-unverified self-reported claims to be roundly rejected as even close to accurate or plausibly scientific data.

(Full disclosure and gigantic parenthetical: I supported the Iraq war. I still do, in fact, nasty as the aftermath has been, because I consider the alternative as far worse for both the world (in general and most especially Iraq's neighborhood) and Iraq.

The terrible human toll there doesn't sting my conscience much - though it engages my sympathy and desire to do whatever we can to alleviate it... as I think is being done - precisely because I compare it with the alternatives over time, and because I find it difficult to castigate or blame myself for supporting an action whose primary negative consequence is the vile and cowardly attacks of others on innocent people.

If US troops, as a matter of policy and/or under orders were blowing up or otherwise murdering random innocent Iraqis (for fun or intimidation or whatever purpose), I'd rail against the war and the State for issuing such policy or orders. But that isn't the case, by any evidence I've seen, and those that do so on their own get arrested, investigated, and seem to end up in prison fairly reliably.

(The issue of "torture" is not this issue, since what rough interrogation is authorized by policy and order is explicitly to be aimed at those seriously thought to be guilty, rather than to intimidate the innocent or the whole polity by a show of undiscriminating terror. That's another discussion, as this parenthetical is already long enough to be an entire blog post.)

The "excess deaths" caused by FREs/Al Quaeda/Iranian-backed elements, well... I can't say as I hold them as being my responsibility, even in part.

The deliberate targeting of civilians by evil men, designed to pressure decent people into giving in, or terrify the population into submission or to embrace Any Offerd Order As Better Than Chaos, are not something the decent should feel guilty for, precisely because of the actions' nature and intentions.

If this appears hard-hearted, so be it - being otherwise seems to me to be more deadly to the innocent and decent people of the world, and I'm concerned with [likely] outcomes and incentives more than I am with emotional impact.

Feelings (especially feelings, alone) aren't a great guide to policy, I think.)

In the first survey, the one that estimated roughly 100,000 excess deaths
in the first twenty-one or so months of the war, a very small group of
pollesters asked a relatively small group of households to list the number
of people that had died in their households in the preceeding three years.
Something like a normalized forty or so more people died after the war than
before. If the same was assumed to be true of the rest of the population
then we'd have 100,000 people dead that wouldn't have been if it weren't for
the war.

It seems like a straightforward calculation, but there are some interesting
assumptions behind it. For one there is the whole question of what 'household'
means. The people making the study had a narrow definition: the people living
in a dwelling plus those that used to live in that dwelling but had died. Did
the people answering the pollesters have the same definition? Suppose you
have a son that died. But your son hadn't been living in the same building
for some time. When asked if someone in your household had died, are you
really going to leave out your son?

The authors of the Lancelot survey assume you would leave out your son.
I don't believe that this is what people would really do. So in fact the
survey wasn't a 'household' survey; it was somewhere between a household and
extended family survey. Since extended families are much larger than households
the practical impact would be to significantly lower the number of excess deaths
when extrapolated to the population as a whole. All this is obvious or
would be obvious if one was actually doing such a survey. An ethical
researcher would have attempted to estimate the magnitude of such an effect
and included a second estimate for the number of deaths making the assumption
that extended families were being surveyed instead of households.

Another interesting assumption is the belief that everyone will answer honestly.
Obviously some people are and were deeply committed to one side or the
other. If a person is willing to aid and abet those striving to kill
americans can there really be any doubt they would lie if there seemed a benefit to
lying? Given the small number of actual reported deaths it would have only
taken a handful of people answering disingenuously to completely throw the survey.

The previous Lancet study WAS discredited. The 655,000 civilian death estimate is also suspect.

OK, not "suspect," but total crap. The Lancet has an ideological agenda these days.

One thing is true: civilian death tolls are very high, and civilians are bearing the brunt of the war deaths by a huge margin, since they are the prime target of insurgents, and are likely to be in the way during coalition counter-attacks. The fact that certain parts of Iraq are producing ghastly civilian tolls three years after the invasion does not speak well for American and Iraqi strategy. Draw your own conclusions.

Let's see if I can summarize. One can argue (and apparently more than a handful are arguing) that the study is flawed, attributable to left-wing observer bias, to left-wing agendas in that segment of the medical community which carried out the study, to left-wing electoral motives, and to general left-wing nefariousness.

Great. I'll accept that this study is bunk, produced solely for political gain and partisan ax grinding by the left.

Thereafter, please point me in the direction of the conservative study (since US conservatives are apparently the only group that operates honestly and with integrity) that passes all the first-level smell tests and has been carried out rigorously to produce robust results.

I promise I will believe that number. And I will be hornswoggled if it changes my opinion one iota that a) the primary rationale for the war was mistaken, b) my support for starting the war was therefore misguided, c) our war effort, led by the Bush administration, has grossly mismanaged, and d) whatever number of excess deaths we agree on - 250,000, 100,000, 10,000 or whatever - is far too many for the wretched situation into which this misadventure has devolved.

And, yes, it feels cheap even to wax sarcastic about 'excess deaths' like they're some kind of poker chips.

Cheers,

Alkali, you're making my point for me . . . knock more people out of the German sample, and the comparison only looks worse.

James: fine, compare it to the American civil war. How come they're missing more people than armies that sent brigades in massed formation into the teeth of modern artillery?

R. Adrian,

No it was not discredited at all. It was, as Tim Lambert points out "flypaper for innumerates".

Alkali, you're making my point for me . . . knock more people out of the German sample, and the comparison only looks worse.

I don't think you get it. Total German civilian mortality was much larger than the civilian mortality due to the handful of causes identified by your source. To make an apples to apples comparison, you need the first number.

Rofe: I'd guess that the best estimates of violent deaths come from the death certificates. The Lancet Study says they got certificates for 80-90% of deaths. Assuming that that represents the number of deaths that are certified, and taking hte actual death certificate figures from the NEw York Times, that gives us an upper bound of about 250 violent deaths a day, which is less than 1/3 of what the Lancet study claims; and a lower bound of about 115 violent deaths a day. That's a lot, but it's also a lot more believable.

I'm not interested in mortality at the hands of the German governemtn; I'm interested in mortality from war violence/famine. A totalitarian government can, if it wants, kill a lot more of its own citizens than any war machine; it's a lot harder for citizens to run and hide from their own government. The point is: is this a reasonable figure for violent death of civilians in war? The answer seems to be "no".

Plus I'd love to read a critique that is based on something esle than "it sounds strange to me".

In this case, I would accept that as a valid statistical method. When our grantees send us reports that say obviously ridiculous things, we send them back rather than publishing them. It is usually something simple but very important, like the citations of the pre-invasion death rate.

Concerning the previous Lancet report, am I mis-remembering that as giving a death count of 100,000, with a margin of error of ~95,000? I always find reports with p=0.49 (or otherwise borderline statistical significance) to be suspect, since it is easy to tweak a few assumptions to get from there from p=0.51.

How quietly do you need to mumble the disclaimer when you say, "We found 100,000! give or take 100,000"? Sure, it could be 200,000, but...

Further to Sebastian Holsclaw's question, the Lancet paper reports:

Our estimate of the pre-invasion crude or all-cause mortality rate is in close agreement with other sources. [footnote 18,19]

Turning to those footnotes:

[18] CIA 2003 Factbook entry for Iraq. http://permanent.access.gpo.gov/lps35389/2003/iz.html (accessed Oct 2, 2006).

[19] US Agency for International Health and US Census Bureau. Global population profile: 2002. Washington, DC: US Census Bureau, 2004.

Looking at the CIA 2003 Factbook entry, updated 8/1/2003:

Death rate: 5.84 deaths/1,000 population (2003 est.)

The US death rate from the same source:

Death rate: 8.44 deaths/1,000 population (2003 est.)

Why the difference? Possibly Iraq has a lot more kids than the US. Also from the same source:

[Iraq:]

Median age: total: 19 years
Population growth rate: 2.78% (2003 est.)

[US:]

Median age: total: 35.8 years
Population growth rate: 0.92% (2003 est.)

This statement bothers me:

"The average report is of about 30 civilian casualties a day--a horrifying number
that should sting the consciences of those who advocated war."

Who should we be blaming for this? Those that advocated the war or those that
are doing the killing (or aiding and abetting the killing)?

There's something absurd about blaming those who are attempting to defeat those
doing the killing for the acts of the killers.

Yes, there's an argument that if the United States is incapable of crushing
the monsters that stalk Iraq today that it shouldn't have stirred them up
by trying to destroy them. There's a pragmatic argument that if one knows this
is going to be the result ahead of time, then yes one should just stay away.

But this isn't the real world. No one has such foreknowledge. In fact
even worse having such an attitude biases the future towards an even worse
state than it is today.

Jane,

Very good. The back of my envelope says for three years the total is 125,000. To what end, please?

Yes, I am pleased that Saddam Hussein is out of power and that his murderous sons are dead. But for the life of me I don't see any kind of endgame that's worth the price.

Cheers,

Jane:

I am making exactly that point.

Your suggestion is that the Lancet reported civilian death rate must be too high because it exceeds the implicit death rate you derive from 1.8 million German civilian deaths over 6 years of WWII in a country of 80mm people.

Doing the math, that implicit German death rate is something like 1.8mm/6 years = .3 million/year / 80 mm people = 4 deaths per thousand per year. am saying that your reported German death rate is too low because it only collects a particular number of civilian deaths, not all civilian deaths from all causes like the Iraq number you are using as a basis of comparison. The right German number simply has to be higher than that.

Yes, Alkali, but there is no government in Iraq trying to exterminate the population. THere is only bog-standard civil war. That's why the lower figure is relevant, and the higher one is not.

Looking at the CIA 2003 Factbook entry, updated 8/1/2003:

Death rate: 5.84 deaths/1,000 population (2003 est.)

Looking at the CIA 2006 Factbook entry:

5.37 deaths/1,000 population (2006 est.)

So why would the authors of the Lancet paper rely on the 2003 CIA Factbook to boster their case aas to the 2003 death rate without admitting that the 2006 Factbook completely undermines their theory?

If we rely on the CIA factbook, there have been NEGATIVE "excess deaths" in Iraq.

"Why the difference? Possibly Iraq has a lot more kids than the US."

I already thought of that. That is why I included Mexico and Brazil.

Rofe -- no, because that's the current rate. The earlier rate, it is generally agreed, was lower. So back of the envelope, the figure is probably below 100K.

400K died in the civil war--was it worth it? It depends on whether you think slavery is really, really bad, or not worth 400K lives. I go with the first. Obviously, that depends on the outcome--good state or good partition=worth it; horrifying and endless civil war=not worth it. But it's too soon to tell on that.

Rofe, "This is a bad study" is not the same as saying "All liberals are liars". It's not even the same as saying "We were right to go to war in Iraq." The point Jane was trying to make with the opening quotation was that the numbers should stand or fall on their own, without reference to which political side they appear to support.

BTW - the WHO's statistics give a pre-war death rate for Iraq of about 9 per 1,000, which is significantly higher than the 5.0 or 5.5 in the Lancet study. link

Jane, a small point: Do you think that the majority of casualties in the Civil War came from massed formations charging into rifles and artillery? (a la Gettysburg?)

I get the feeling from what little I've read about the US Civil War that that aspect is overemphasized in movies and whatnot. I also get the feeling that the generals got pretty wise to that idea by 1863.

I dunno -- perhaps others would like to comment?

Actually, the majority of deaths were from disease and hunger. But the military operations were certainly at least as productive of mass death as the current terror attacks.

From yesterday's WP:

And neither does Michael E. O'Hanlon of the Brookings Institution, which also tracks Iraqi deaths.

"I do not believe the new numbers. I think they're way off," he said.

Other research methods on the ground, like body counts, forensic analysis and taking eyewitness reports, have produced numbers only about one-tenth as high, he said. "I have a hard time seeing how all the direct evidence could be that far off ... therefore I think the survey data is probably what's wrong."

However, several biostatisticians and survey experts were supportive of the work.

"Given the conditions (in Iraq), it's actually quite a remarkable effort," said Steve Heeringa, director of the statistical design group at the Institute for Social Research at the University of Michigan.

"I can't imagine them doing much more in a much more rigorous fashion."

He said the study made "minor departures" from the standards generally used in national surveys for choosing what households to interview. Whether those departures, brought on by wartime conditions in Iraq, introduced a bias in the results is impossible to measure from the data alone, he said.

Frank Harrell Jr., chair of the biostatistics department at Vanderbilt University, called the study design solid and said it included "rigorous, well-justified analysis of the data."

And Richard Brennan, head of health programs at the New York-Based International Rescue Committee, said the study's survey approach was typical.

"This is the most practical and appropriate methodology for sampling that we have in humanitarian conflict zones," said Brennan, whose group has conducted similar projects in Kosovo, Uganda and Congo.

"While the results of this survey may startle people, it's hard to argue with the methodology at this point."

Donald Berry, chairman of the statistics department at the University of Texas' M.D. Anderson Cancer Center in Houston, said he believes the study was done "in a reasonable way." But he said the range of uncertainty given for the estimates was much too narrow, because of potential statistical biases in the survey.

While it's impossible to calculate a better range that accounts for that, he said, it wouldn't be surprising if the low end dropped about four-fold to 100,000 deaths. A wider range of uncertainty would make the 655,000 figure less meaningful, he said.

How come they're missing more people than armies that sent brigades in massed formation into the teeth of modern artillery?

Jane, I think the factor you are missing is that a significant number of deaths are inter-faction genocide.

Armies are small, relative to population--even with near-universal service, the Confederacy fielded only 25% of the population, and only a small portion of that Army was involved in high-casualty battles. Sherman and Sheridan were criminals and terrorists, but their goal was terror, not wiping out civilian populations.

If your goal is to kill all the "others," you don't get something like the War Between the States--you get something like Rwanda, where 10% of the population died in a couple months.

GT, all the other statisticians are saying is, essentially "We don't know any way of getting better numbers." That isn't at all the same thing as saying hte numbers are good. Sometimes the correct answer is not "here's the best figure I can come up with"; it's "I don't know".

Jane,

Yes, but they are also saying that there are established and tested procedures for cases like this and this study follows such procedures. If you impugn this study you are saying all such other studies in Sudan and elsewhere are equally invalid.

Why don't you just multiply the posted Iraqi death rate (5.84/1000) by the population (24M, iirc) and what do you get ?

128,800 deaths per year

TOTAL

times 3 is 386,640.

They should shoot the peer reviewers.

Thanks to the other posters citing the varying death rates in Iraq pre-invasion. This shows the real difficulty in calculating the increase in deaths due to the war, which the authors of this study should have done. A more honest approach would have been to calculate the difference based on each of the estimates. That they didn't do this exposes them to claims of bias.

Horrors no! You don't mean to suggest that we don't have an accurate count of deaths in Sudan! Say you don't mean it!

Of course we dont know how many people are being killed in Sudan. Any "scientific" study of hte matter is garbage. What we do know, from eyewitness accounts, is that there are gross and widespread human rights abuses. That's all we need to know. Putting numbers on it is false precision.

Mrs. L,

You are correct. But my comment wasn't entirely directed at Jane's original post. There are indeed folks who make just the kinds of points you highlighted. If I'm not dyslexic, I read some of them upthread.

Jane,

I think we may be talking past one another. Yes, I believe the trade off in our Civil War was worth it. But I would have to say that a good state / good partition in Iraq, though desireable in and of itself, might not be worth the price the Iraqis are paying IF the war could have been managed better and casualties - whatever the 'true' number now - could have therefore been kept lower.

Think of it this way - assuming that the outcome of WWI was a 'good', does that result make pointless suicidal mass assaults worth it?

Cheers,

Not sure claiming it's all garbage is much of a methodological critique.

I'm wondering how many were being killed specifically by the power structure in Iraq before the war... 10's of thousands of Kurds alone were killed, about a million displaced in 1987-88, and that's only from one very small minority (albeit a disproportionate one for this type of question.) Why are we not looking at that type of action as the baseline? And this was BEFORE they retaliated against the Kurds for their support of the US in Gulf War 1.

Becuase "about 655,000 Iraqis have died above the number that would be expected in a non-conflict situation" isn't a useful baseline anyways: if you have an insane meglomaniac in charge of the government, it's not exactly a normal "non-conflict situation."

I'm not saying that, even including all of these critiques, this balances out the deaths we have now. If, however, you are weighing the lesser of two evils, than you have to admit that both weren't any good. The fact that our mishandled invasion is killing civillians doesn't mean the war was in itself a bad decision (even though it was) - but our mishandling of it was.

Why doesn't someone talk about the excess deaths that occurred versus a projected baseline of deaths if we had done the war with sufficient forces, and had managed it based on sound principles from the beginning? Hint, it's not becuase it's such a difficult figure to come up with, even though it is. It's because it helps neither side of the arguement as much, even though it's probably the most useful figure for _fairly_ critiquing Bush - something noone seemse interested in doing.

Jane Galt said in comments above:

"... The point is: is this a reasonable figure for violent death of civilians in war? ..."

Why do you keep claiming this is a figure for civilians? It isn't.

GT - Yes, there are established procedures for this type of study. However, following methodology does not always produce credible results. Sometimes following established protocol only puts the gloss of science on BS. When a study such as this one seems to defy reason, it's proper to be skeptical. That's particularly true in this case. Before 9/11, similar studies -- all no doubt scrupulously following established forms -- claimed that the sanctions on Iraq were killing hundreds of thousands of Iraqi children. Before the invasion of Afghanistan, experts, based on their scientific models, predicted more than a million deaths from the shortages of food, energy, and medical care the war would cause. None of these claims proved true. Turns out Iraq was giving researchers false data in an effort to turn public opinion against the sanctions. The shortages predicted for Afghanistan were nowhere near as severe as the scientists had assumed for their modeling. Given this track record of accuracy, maybe we should start reviewing the methodology. It's not producing reliable results.

Also keep in mind that with Germany, you're comparing the proportions of the pre-war population that died, whereas with the U.S. Civil War you're comparing absolute numbers. The 600,000 figure would mean the proportion of the Iraqi civilian population that died is several times lower than the proportion of the Soviet population (nearly 7%).

To Sebastian Holsclaw: The Qatari mortality rate is 4.61, the UAE is 4.26, and Kuwait's is 2.42 (per 1000 population). That doesn't make a pre-war Iraqi death rate of 5.0 seem so unreasonable.

How many of the "critics" of the study have actually read it? And, if you haven't even read it, do you realize that you are being willfully stupid to comment on something you haven't read? I am just curious how low public dialogue has sunk.

David Walser,

Okay, so you have two examples in which "the scientists" were wrong. Based on that, you want to start reviewing "the methodology."

What part of that methodology do you have problems with? Statistics as a whole? Cluster sampling? If you have problems with them, then by all means let's talk about these problems.

Furthermore, how are these examples comparable with what we're talking about here? In the case of Iraqi children, researchers were being provided with fraudulent data--do you think that was happening here? As for Afghanistan, making projections is very different from assessing what's currently happening.

To Sebastian Holsclaw: The Qatari mortality rate is 4.61, the UAE is 4.26, and Kuwait's is 2.42 (per 1000 population). That doesn't make a pre-war Iraqi death rate of 5.0 seem so unreasonable.

According to what source? The CIA Factbook? Is it reliable? If so, then why does it say that the POST-war Iraqi deathrate is 5.37?

Note also that among the 10 countries (and pseudo-countries) with the lowest death rates, according to the CIA Factbook, are the West Bank and the Gaza Strip. Do you believe that the West Bank and the Gaza Strip have among the 10 lowest death rates in the world? I don't.

the last thing I want to do is get into a calculus of death. however, for anyone who thinks that the number is "too high to be credible":

a sample of deaths in world war I(combined civilian and military, rounded off)-

France: 1.4 million

Italy: 650,000

Romania: 611,000

Russia: 3.7 million

Serbia: 1.1 million

UK: 733,000

United States: 126,000

Austria-Hungary: 1.5 million

Bulgaria: 350,000

Germany: 2.5 million

Ottoman Empire: 2.5 million

What were they saying again about the death toll being too high?

Charles Giacometti says if you haven't read it you can't critic it.

Fair enough. As long as the inverse is also true. If you haven't read it you can't cite or promote it.

Ultimate, whatever the number, is the answer is "so what?". Not so what as in who cares, but, so what's the conclusion? Would you suggest it's not worth it if 600k died? Would you still suggest it's not worth it if 100k died? How about 20k? 10,000 people?

Will you play moral calculous and assume that if 100,000 people died as a result of the invasion and 100,001 people would have died if we didn't invade, therefore invading against the UN and world's opinion was worth it? My hunch is those that gleefully cite this statistic to show what a bad guy Bush and his ilk are would never countenance invading Iraq for virtually any reason other than if Saddam was directly involved in 9/11.

And even then, had Saddam been involved in 9/11, would it then be "worth it" to inadvertently kill 600,000 had nothing to do with 9/11 just because Saddam might have been involved?

No, it wouldn't be "worth it". The answer is few things are ever "worth" the taking of human life. But unfortunately it must happen from time to time and that's the paradox that all the peace loving hippies just can't wrap their brains around.

Sean,

At the risk of earning Charles Giacometti's ire, I'm going to admit I've not read the report. However, I don't think I need read a report to have valid reasons for questioning the report's conclusion. Just this week, I reviewed a business valuation report prepared for one of my clients. A preliminary scan revealed I was going reject it. (The appraiser used 32.9% as the expected annual return for an investment in the entity. The entity's assets consist of nothing more than a diversified portfolio of US equities.) Close examination revealed that the appraiser had meticulously applied standard practices in producing the report. In this case, the problem was not the methodology, but the quality of the assumptions used.

My focus on the report's methodology in the Lancet report was in response to GT's line of posts that also focused on the methodology. He seemed to be saying that it was wrong to question the result if the study was conducted using accepted methodology. My point was that good methodology does not always yield reliable results. It's something I preach to my staff all the time: We can't accept "x" as the answer just because the computer says the answer is "x". If "x" defies reason, experience, and intuition, we've got to do more work. In this case, the Lancet numbers defy historical precedents and logic. Until someone can cogently explain why the death rate in Iraq should be higher than in prior conflicts -- when everyone agrees the physical damage caused by the war is substantially reduced from that experienced in prior conflicts, we've got to question the report's conclusion. I don't care how many decimal places the authors' used for their calculations, the answer just defies all reason and needs to be rejected.

A better comparison of the rate of deaths per total population is WWII Japan versus Iraq. Using the Lancet's number of 655,000, Iraq has an overall rate of 2.6% excess dead per total population. Japan, through all of WWII, had a rate of 3.1%. This figure is based on both military and civilian Japanese killed thus avoiding the confusion regarding how many of the dead were civilians. While the Japanese rate is somewhat higher, I feel the similarity is beyond belief. My calculations and sources can be found here.

Kieran Healy seems to be saying just because the number appears improbably high, that doesn't mean the number is wrong. So what? It sure doesn't help the Lancet's claim.

However, when one adds to the questionable validity of the number, the release of a similarly dubious Lancet study immediately before the 2004 election, motive must be considered.

I speak only for myself when I say the Lancet and this study stink to high heaven.

Moronic brownshirt fucks - every. Fucking. One. Of. You.

"I agree that there are a lot of deaths. But I don't think that there are as many as the Lancet study makes out. Not because I am comparing them to one-time events like Nagasaki, which I agree is silly. But even comparing them to other long term wartime figures makes it look to big."

Rwandan civil war, 1 month of killing in a country of 8 million produced about 500,000 deaths.

Iraqi civil war, 2-3 years of fighting producing 650,000 deaths in a country of 22 million is not astonishing at all.

Saying the numbers look too big to be true has no merit whatsoever as an argument. There may be valid arguments to make, but that one is utterly valueless.

A contrarian thought:

If the number is even close to being accurate, wouldn't that be a point in favor of the neo-cons and staying there? Is there anyone that believes that if we leave, they will stop killing themselves?

If the sample in the Lancet study is properly representative and the statistical methods are reliable, then well over 100,000 residents of Baghdad have been issued death certificates for death by violence between October 2004 and July 2006.

Accordingly, it should be trivially easy to verify the Lancet numbers; there should be a huge paper trail available in the Baghdad area to show a weekly average of more than 1000 death certificates issued for deaths by violence.

The Lancet paper *in no way* is claiming it measures civilian deaths. It is measuring all deaths, including the Iraqi army which got pounded like a kid's first toy drum during the initial invasion.

The methodology was sound, and the death rate pre-war was measured, and agreed with multiple other sources, and thus, was likely to be correct.

Similarly, the study confirmed the expectations of the 2004 study (using a different population and slightly different technique), and thus we have more assurance that both are correct.

The study is strong evidence, and deserves a lot more than "I don't like it", or "it appeared in The Lancet" as a challenge. To try to counter it with such talk is to hide from the evidence for fear it might say something unpleasant. One can ethically be suspicious, and determined to dig for the truth... but to turn away, and insist it must be bad without finding serious problems with it is an act of moral cowardice.

Oh, and for those who missed it, and think the peer reviewers should be shot: the deaths reported in the headlines were the *extra deaths*, the ones that we *beyond* the number you'd expect if the death rate stayed constant.

As for what this says about the moral basis for the war, I suppose it depends on how much you value human life, and how much of a threat Iraq posed to us before we invaded.

Since the threat was barely existent, if you think killing 650,000 people is acceptable, if it wards off a barely existent threat, I suppose you're on solid moral ground, for some very bothersome ideas of morality.

Looks like a simple case of Bayesian updating. We all start this little foray into confirmation with a set of beliefs about the number of Iraqis who have died since the invasion. We get these beliefs, at least in part, from knowledge of past events like those being mentioned here – WWII, the American Civil War, Vietnam, Rwanda. We compare those to the numbers produced by the study and we adjust our prior estimates of Iraqi war dead. The amount by which we adjust our priors is subject to a number of factors, prominently including the quantity and perceived quality of the data that gave rise to our priors, our beliefs regarding the validity of the methodology used by the research team from the Lancet article and, not least, various types of bias.

Those who opposed the invasion (or who have decided to oppose it in retrospect after some of the outcomes are more or less known) tend to nearly completely discount all of the priors and heavily weight the Lancet study in their estimates. This gives rise to silly statements such as “you have to accept the estimate unless you can find a flaw in the methodology”. This is logically equivalent to saying that you have to completely reject the Lancet estimate unless you can find flaws in all the methodologies of your priors. Both statements cannot be right. But they can both be wrong.

This is all fine, as far as it goes. The problem is that humans are not very good Bayesian updaters. They tend to place way too much emphasis on the most recent number they have heard, even when it is completely unrelated to the issue in question. So the 655,000 Iraqi casualty number will likely gain great currency (think about the number of homeless during the Reagan years, the number of deaths caused by Love Canal or Agent Orange) even though it is far removed from all other estimates, because it appeals to those who have a particular motivational bias and so will repeat it. The repetition means that it will take much less effort to recall the Lancet number than those of their priors. And since humans tend to assess probabilities based on cognitive effort, they’ll go for the 655,000 number, and that will drive their decisions.

And that’s why media bias matters, no?

Regarding your claim:

"Bayji, a major oil centre, has 60,000 inhabitants. As anyone from a town that size can tell you, it wouldn't go unnoticed if it was losing 1,600 people a year to murder."

The Lancet study claims that the war has led to the demise of 2.5% of Iraq's population over 3 years. Agreed? I hope so, you wrote it in your post.

2.5 % of 60,000 (do it on a calculator: 0.025 x 60,000) = 1500 deaths.

Over three years.

Now get out the calculator again. 1500 deaths over 3 years (1500 / 3) = 500 deaths per year.

I think Ayn Rand would be very disappointed by your lack of competence in basic mathematics.

It is fairly well acknowledged that epidemiologists are false mathematicians, on par with marxist economists. Far too many of them really don't understand how to create a study, nor why a conservative approach to "statistical proof" is necessary. CSPI et al love epidemiological studies and use them heavily in pursuit of their missions, which is a contrary indicator of their accuracy.

Do we know what the right number is? No, and we won't for a long time. What do we know? Iraq is not a nice place, especially Baghdad, Ramadi, Falluja, etc. Significant portions of the rest of the country are not as bad. Many people have died, some for the better (Uday, Qusay, ex-baathists...) and far too many for the worse (regular folks trying to get on with their lives).

What will we always disagree about? Were we right? What should we do now? Does terrorism exist? If it does exist, how best to deal with the problem?

I know my answers. I know the answers of most of the people here. They're entirely predictable and tiresome. But as a quick hint, one of the biggest crimes of WWII was that the US stopped when it met the Russians in Germany. Several nukes dropped on Moscow and multi pronged assault (Germany, Japan, China, India, Persia, Turkey) would have solved MANY problems. Further back, the abandonment of russia to the communists post WWI was another crime. But then I actually believe in freedom, unlike the leftists and liberals here who still long for the great dream and believe in the romance of the revolution. It's fairly easy to guess what my preferred solution to our current problems is.

Here's something about the scale of (violent) death in Iraq, and the tracking/concealment of it.

Please note that it doesn't deal with non-violent death rates, which the Lancet studies covered.

http://www.macleans.ca/topstories/world/article.jsp?content=20061016_134735_134735

See D. M. Pettilli's comment (it's down in the comments) at
http://medpundit.blogspot.com/2006/10/lancet-strikes-again-i-admit-this.html
(hat tip InstaPundit and Stephen St. Onge)

I was thinking the same thing but he said it first. I'll
say it again in my own words. Briefly recapitulating, the
method of the study was to interview randomly selected
households and ask if anyone in their household had died (and
when) over a significant period of time including over a year
before the american invasion.

Reported deaths from before the invasion were then
assumed to be the baseline normal rate and any number
greater than that was then blamed on the invasion.

The authors then attempted to verify most of these claims
of death by checking official records. They were able
to verify eighty percent of what their interviewees
claimed. In fact their verification success rate was even higher
than that, perhaps 90%, because for some claims there was no
attempt to check official records.

Here's the incongruity. Official iraqi records do not
show 700,000 additional people killed, or anything remotely
like that. For this recent Lancet study to be true, then
Iraqi government records have to be radically and dramatically
wrong.

And yet the study claims to have selected a large number of
people at random and was able to verify between 80 and 90
percent of their claims with government data.

There doesn't seem to be any way that both of these claims
can be true at the same time.

I'm trying to imagine how the study could possibly have come
up with this result.

So far I have two hypotheses.

Either (a) the Iraqi government death statistics are
basically correct and the study was in error because the
authors assumed they were sampling households when in
fact they were sampling extended families (see my comment above)
and further the average extended family was about ten times larger
than the household size leading to a ten-fold error in the death
rate estimate, or (b) the study is fraudulent and someone just
got caught faking it.

The Lancet paper . . . is measuring all deaths, including the Iraqi army which got pounded like a kid's first toy drum during the initial invasion.

Please read the Lancet study carefully. It clearly says 112,000 of the 655,000 excess deaths date to prior to October 2004. This means over 540,000 of the deaths cannot be remotely attributed to the pounding the Iraqi army took.

Alternatively, the study attributed 31% of excess deaths by violence to Coalition action, which means, of course, 69% of Iraqis killed in Iraq -- 400,000 of them -- were killed by Iraqis or Al Qaeda in Mesopotamia. And accordingly, not by the pounding the Iraqi Army took at the hands of the Coalition.

The Lancet paper is accordingly very clear the vast majority of its claimed excess deaths are not Coalition-caused Iraqi Army casualties.

The 650,000 number was extrapolated from 547 deaths in their survey and 92% of the 547 deaths in the survey had death certificates to back them up.

Meaning the Iraqi death records do a pretty accurate job of representing the total number of people who have died in Iraq in the last 3 years. Except the Iraqi death records doesn't back up the 650,000 excess death total.

There's the catch 22. If you want to claim that the number of deaths aren't reflected in the Iraqi government records but at the same time your survey shows that the records do indeed back up those records.

The Lancet study on the 540,000 Iraqi deaths that occurred because of the sanctions, when the fine folks at Lancet goal was to get the sanctions lifted against Iraq, are no longer useful to them so they no longer mention that and what the continued (also, phony) death rate caused by those santions if we hadn't gone to war and the sanctions were still in place.

doesn't=don't

Knob Jockey!

"Here's the incongruity. Official iraqi records do not
show 700,000 additional people killed, or anything remotely
like that. For this recent Lancet study to be true, then
Iraqi government records have to be radically and dramatically
wrong."

Are you sure about that?

Please indicate how you know that official Iraqi records do not show these 700,000 (350,000-950,000 really) excess deaths. Where have you seen an official complilation of Iraqi death records? It has been my understanding that no such task has been done. No one has accumulated the information because no one with authority and means to do so has an incentive to do so. It may even be true that no one has the ability to do it in the present conditions. They can't keep the power or water on all the time. It's not as if every death certificate gets logged into a national death-watch database. Nobody is notifying the Social Security Commission to stop sending the dead guy benefits.

"The Lancet study on the 540,000 Iraqi deaths that occurred because of the sanctions, when the fine folks at Lancet goal was to get the sanctions lifted against Iraq, are no longer useful to them so they no longer mention that ..."

First, it was not the Lancet, it was the author of a paper published in The Lancet. Second, the author of that study, in a letter published by The Lancet, withdrew his conclusions in 1996.

Who is Chris Murtaugh?

I don't know how vital documents are handled in Iraq. Based on the statement that the families had death certificates for 80% of the deaths it seems to me that doctors or some similar professionals do issue death certificates to next of kin. Just because a doctor gives the family a death certificate doesn't mean that this certificate will also be recorded by some governmental body. The US government has collected vital statistics since 1909, but that doesn't mean that certificates of death or birth weren't used in the US before 1909, just that then weren't tabulated.

It's quite likely that with the collapse of the government in 2003 nobody is tabulating vital statistics.

The fact that the Iraqi government issues death counts based on morgue intakes and hospital deaths strongly suggests death certificates are not tabulated--else why not simply report them? Obviously people who are killed immediately from attacks won't go to the hospital at all. Immediately identifed bodies probably won't go to the morgue, they will be likley be picked up by undertakers or familiy members. So these deaths are not going to be measured by morgue+hospital counts.

The Lancet paper . . . is measuring all deaths, including the Iraqi army

It isn't. By definition, someone who is in the army is not part of a household, so would not be covered by the survey. It also omits people in prison, which could bias the estimate either way AFAICT.

By definition, someone who is in the army is not part of a household...

Aside from the question of whether an Iraqi army really exists, this makes no sense whatsoever.

What difference does it make? 600,000, 100,000. America has NO RIGHT to roam this planet and kill innocent people that never threatened America.
This is called murder in the fisrt degree!!

Warmongering Lunatic:

Please read the Lancet study carefully. It clearly says 112,000 of the 655,000 excess deaths date to prior to October 2004. This means over 540,000 of the deaths cannot be remotely attributed to the pounding the Iraqi army took.

It's considered good manners not to suggest someone read something "carefully" when what they have said is completely consistent with the something they are being asked to read.

For example, the new study suggests that the 2004 study, showing 98,000 excess deaths was a bit low; the new study suggests 112,000 excess deaths.

Those deaths include the invasion, and likely include large numbers of Iraqi soldiers. Hence, as I said:
The Lancet paper *in no way* is claiming it measures civilian deaths. It is measuring all deaths....

When I want to point out that the study isn't restricted to *civilian* deaths, I point out that it expressly includes *military* deaths.

You are correct; the study suggests that another 500+ thousand deaths have occurred since then, and that a majority are not due to direct action by coalition forces.

But if you cause an avalanche, it does little good to say that you didn't expect the specific results of that avalanche. You either do what you can to patch things up, and resolve not to start more avalanches, or you ignore the results because you don't like them.