pageok
pageok
pageok
The Iraq Casualty Numbers:

Daniel Davies is impressed with the Lancet study estimating 655,000 "excess" Iraqi casualties since the U.S.-led invasion. Based on his experience conducting surveys in Iraq, Steve E. Moore is not.

Of Moore's criticisms, the most significant seem to be 1) the relatively small number of cluster points used for the survey (which, he claims, is more important than the number of actual interviews in the sample); and 2) the failure to obtain demographic data on those interviewed so as to verify the representativeness of the survey sample.

the key to the validity of cluster sampling is to use enough cluster points. In their 2006 report, "Mortality after the 2003 invasion of Iraq: a cross-sectional sample survey," the Johns Hopkins team says it used 47 cluster points for their sample of 1,849 interviews. This is astonishing: I wouldn't survey a junior high school, no less an entire country, using only 47 cluster points.

Neither would anyone else. For its 2004 survey of Iraq, the United Nations Development Program (UNDP) used 2,200 cluster points of 10 interviews each for a total sample of 21,688. True, interviews are expensive and not everyone has the U.N.'s bank account. However, even for a similarly sized sample, that is an extraordinarily small number of cluster points. A 2005 survey conducted by ABC News, Time magazine, the BBC, NHK and Der Spiegel used 135 cluster points with a sample size of 1,711--almost three times that of the Johns Hopkins team for 93% of the sample size. . . .

With so few cluster points, it is highly unlikely the Johns Hopkins survey is representative of the population in Iraq. However, there is a definitive method of establishing if it is. Recording the gender, age, education and other demographic characteristics of the respondents allows a researcher to compare his survey results to a known demographic instrument, such as a census. . . .

while the gender and the age of the deceased were recorded in the 2006 Johns Hopkins study, nobody, according to Dr. Roberts, recorded demographic information for the living survey respondents. . . .

Without demographic information to assure a representative sample, there is no way anyone can prove--or disprove--that the Johns Hopkins estimate of Iraqi civilian deaths is accurate.

These are also criticisms that I have not seen responses to elsewhere (e.g. here). No charges of fraud here, just poor methodology.

I do not mean to diminish the tragedy of civilian Iraqi deaths. I agree with Moore that "there have been far too many deaths in Iraq by anyone's measure," and I certainly believe that the Bush Administration's poor policy decisions and execution bear a significant portion of the blame. Nonetheless, Moore seems to mount a serious challenge to the validity of the Lancet estimates.

UPDATE: Many commentators note that the realtively small number of clusters is accounted for in the study by providing a confidence interval — this is true, and it accounts for the huge range of estimates presented (plus-or-minus 40-some percent of the point estimate). So, while the study is evidence that mortality rates in Iraq have gone up post-invasion — a point I did not think was in dispute — it does not provide a particularly reliable estimate of excess deaths. Of course, as Frank Cross notes here, Iraq is not exactly the easiest place to perform this sort of study these days.

Tim Lambert offers additional responses to Moore's critique here.

AntonK (mail):
As well as the methodological problems with this study there is, of course, that nasty old "common sense" test, which this study fails miserably:


1) The research team's method involved interviewing 1,800+ Iraqi families and asking them how many family members died in the year before the US-led invasion and how many in the three years since. To do this kind of work properly, you have to interview a statistically representative sample. This the research team almost certainly failed to do.

Michael O'Hanlan of Brookings, editor of the Iraq Index and a sharp critic of the Bush administration's war effort in Iraq, has on this count dismissed the Johns Hopkins teams work as worthless.

2) The 600,000 number flunks the common sense test. As Tim Blair points out in his excellent blog , 600,000 is more dead than all the Germans killed by allied bombing during World War II. It is ten times the number killed at Hiroshima. As he could have also pointed out, it is more casualties than Iraq suffered in eight years of war with Iran.
3) Even supposing the number had any validity at all, what would it prove? Only that the Iraqi insurgents were an even more murderous, evil collection of throat-slitters, torturers, and car-bombers than previously understood - for after all, almost all these killings are their work. It would not however prove (as the authors of the study do not hesitate to claim) that the US and UK and Australia and their coalition partners should withdraw from Iraq and turn the country over to the killers. Rather the reverse I should think: It raises the question- Can you imagine what these murderers would do if the coalition were not there to defend against them?


From:

Oh, and where are all the bodies?

See also this devastating refutation by Iraq Body Count (no fan of the war): http://www.iraqbodycount.org/press/pr14.php
10.18.2006 9:38am
DavidBernstein (mail):
Well, I pointed out in a recent post that Davies felt free to make up stuff without any evidence about me, so I don't see any reason to credit anything he says.
10.18.2006 9:41am
Al Maviva (mail) (www):
It's not that implausible. You'd merely have to kill, on the average, a thousand people a day, every day, for two years. Nothing to it.
10.18.2006 9:56am
Eric Anondson (mail):
It's not that implausible. You'd merely have to kill, on the average, a thousand people a day, every day, for two years. Nothing to it.

... without the media noticing the dramatic depoulation happening in front of their eyes.
10.18.2006 10:02am
alkali (mail) (www):
Cluster sampling means that instead of picking people randomly out of the population, you randomly choose "clusters" of people and you interview either all members of the cluster or some number of persons chosen randomly from within the cluster.

Is there some statistical reason why 47 clusters can't be sufficient?

By way of comparison, it is my understanding that public health researchers (like the World Health Organization) use a method called "30 by 7" -- 30 clusters with 7 people in each cluster -- to perform estimates of (for example) the vaccination level of a population.

To use Moore's example, If you were going to study a junior high school using cluster sampling, I imagine you would use something like "tables in the cafeteria" or "homerooms" as clusters. It strikes me as unlikely that you would survey more than 47 clusters (rather than just sampling the entire population or surveying all of them).

Moore writes elsewhere in the article:

The UNDP's survey, in April and May 2004, estimated between 18,000 and 29,000 Iraqi civilian deaths due to the war. This survey was conducted four months prior to another, earlier study by the Johns Hopkins team, which used 33 cluster points and estimated between 69,000 and 155,000 civilian deaths--four to five times as high as the UNDP survey, which used 66 times the cluster points.

This is a misleading comparison. UNDP was counting war-related deaths; the "Lancet" study was counting all deaths. If a child dies of typhoid, UNDP didn't count it. If someone is robbed and murdered, UNDP didn't count it. That is not a criticism of the UNDP study or the Lancet study, but it shows why it is not useful to compare the two. (It also shows why Tim Blair's comparisons are not informative.)
10.18.2006 10:32am
debunking troll:
Another reason to be suspicious. They counted 600,000 excess deaths. They claim that 92% of the time, they were shown a death certificate.

This suggests a couple of possibilities:
a) Their sample is representative, implying that one can find 540,000 excess death certificates.
b) Their sample is not representative.

I suppose that there is also the possibility that their sample is representative of deaths but not death certificates, but that seems rather unlikely. As is the possibility that 540,000 death certificates were issued but not recorded.
10.18.2006 10:42am
Michael Poole:
alkali- With a smaller number of cluster points, you need to take greater measures to be sure they are representative of the entire population. Moore cites Appendix A to the JHU/Lancet study as listing a number of other studies that use several times as many cluster points to count deaths (on in the Congo used 750 cluster points). The only other study Moore mentioned that used <=50 cluster points was one for Kosovo, measuring a population: 1.6 million.

As quoted above, Moore is very skeptical that the Lancet study can show that its samples are representative. The researchers did not even collect demographic information that would let them adjust for selection error.
10.18.2006 10:47am
Phineas Page (mail):
Actually, they claim that they were shown a death certificate 80% of the time. They asked for a death certificate in 87% of cases, and were presented with one in 92% of those cases.
10.18.2006 10:56am
alkali (mail) (www):
1) Is there a reason to suspect that the Lancet's selection of samples are not representative?

2) How low does the number of samples have to be to make it necessary to start checking for representativeness?

(By way of example, if I was trying to measure household income in New York and picked ten blocks as "clusters" in which I would try to conduct interviews, I assume that I would want to do something on that score. I am not sure that I would need to do that if I picked 30 blocks, particularly if I looked at the data after the fact to eliminate outliers, which I believe is what the Lancet did -- in 2004, they eliminated Fallujah as an outlier on the high side.)
10.18.2006 11:05am
billb:
It's been reported elsewhere that death certificates are highly likely to be possessed by the family but not recorded in a central office due to bureaucratic failures (see the thread at Good Math, Bad Math for a link to a Crooked Timber thread on the subject). Of course, these reports are unsubstantiated so apply grains of salt as necessary. It is, however, at least food for thought.
10.18.2006 11:05am
frankcross (mail):
The inadequate number of cluster points argument is just flat wrong. As a matter of the theory of statistical sampling and considerable practice. The fact he makes this argument suggests either a bias or that he doesn't know what he's talking about.
10.18.2006 11:14am
MartinM:
Number of cluster points is accounted for in calculating standard deviation. That the low end of the CI is still bloody huge shows that the number of cluster points cannot account for the results, no?
10.18.2006 11:17am
Mitchell Freedman (mail) (www):
I am struck by the fact that those who would most likely rip into Chomsky for his article wtih Ed Herman analyzing initial reports of Khmer Rouge related deaths in Cambodia in 1977 are falling all over themselves trying to discredit the Lancet report into oblivion.

The truth is that the Lancet report may be reasonably challenged, but the report is not ridiculous. Wouldn't "common sense" tell us there's at least 300,000 "excess deaths" in Iraq since Bush's people unhurled the banner "Mission Accomplished"? That's quite a stunning figure itself in this context.
10.18.2006 11:17am
Anderson (mail) (www):
Adler: check your title's spelling.

I'm statistically illiterate, but I expect Crooked Timber will have something up on the cluster-points issue in a day or two.
10.18.2006 11:18am
WHOI Jacket:
Mitchell,

So, US Forces are basically engaging in Khmer Rouge style genocide as opposed to, say, not actively trying to kill more Iraqis than Iran did in the 80's?
10.18.2006 11:33am
Richard Aubrey (mail):
There are two issues the study mentions but which get, IMO, inadequate attention.

One is that the bulk of the dead are men of military age. Considering that the usual complaints about American fighting--but nobody else's--is that most of the dead are women and children, this is unusual, at least.

Excess dead is not restricted to violent deaths, but deaths from any cause, and the change, if any, is attributed to the war. Hence, reduced access to medical care would affect people needing medical care. The young, the old, the ill. For this process to disproportionately turn up the non-vulnerable, men of military age, is curious. Reduced supply of heating fuel means trouble for the vulnerable in the winter. Not men of military age (MOMA). Poor nutrition, the same. MOMA should be able to get by.
From which we conclude that the study, if true, finds the allies have killed a lot of fighting men, which is a feature, not a bug.

The other problem is that the difference issue requires two points, before and after. Even if the "after" point is accurate, an inaccurate before point would make the results meaningless. The study uses a before death rate of 5.5/k/a. The world average is just under twice that. Corrected by comparing with nations whose populations are quite young, like Iraq's, does not help. They, too, have much higher death rates. Ours is over 8/k/a. The before rate needs to be investigated and possibly corrected.

But, if true, it means the sanctions were a great help to public health, not a murder of millions.
10.18.2006 11:37am
byomtov (mail):
Isn't it the case that the size of the sample - the number of clusters - is reflected in the confidence interval?

In other words, it is unreasonable to criticize the point estimate in isolation based on sample size, without considering whether the CI has been properly calculated. If it has been, where is the complaint?
10.18.2006 11:38am
Clayton E. Cramer (mail) (www):
As I pointed out here, subtracting the excess death counts from their 2004 paper would indicate 6,028 excess deaths weekly from 9/1/2004 to 5/20/2006--or 861 deaths per day. Now, it is at least possible that this happened--but where's the media coverage? As I pointed out here, when there has been media coverage of the increase in violence, 100 deaths a day is identified as a dramatic increase. To get an average of 861 deaths per day would imply that there were hundreds of days with more than 1000 deaths per day.

The other problem with cluster sampling is that if this sampling in just a few neighborhoods included places where there had been a fierce firefight or bombing, there might indeed have been a lot of deaths in that location, but be very unrepresentative of the nation as a whole. Imagine if you dropped survey teams into New Orleans, and asked people to tell you about family members who had become homeless recently, or who had died by drowning. Now scale the results up to the nation as a whole.

The claim is that 2.5% of the population of Iraq has died as a result of this war. As I pointed out here, for comparison, during World War II, the Japanese, over a period of six years, lost 2.7% of their population, both civilian and military--in a war in which we fire bombed Japanese cities--killing 100,000 in a single raid--and killed 192,000 with atomic bombs. And you want to believe that somehow, in a war that doesn't involve carpet bombing Iraqi cities, where soldiers have been punished for inappropriate or criminal uses of deadly force, that we have killed almost a equivalent fraction of the population in three years?
10.18.2006 11:41am
byomtov (mail):
Clayton,

1. Again, criticisms bsed on sample size need to discuss the confidence interval.

2. It is generally known that the media very seldom venture outside of the protected areas of Baghdad. This is no doubt quite sensible on their part, but it does mean that events in other parts of the country are not well reported on. So comparing counts based on media reports with the survey is not very helpful.
10.18.2006 11:47am
alkali (mail) (www):
With respect to Clayton's points:

1) The study takes into account all deaths, including deaths from disease, accident, and violence of the non-battlefield sort which would not typically be the subject of media coverage. There are about 20 deaths every day in Montana; very few of them make it onto the evening news in any given month.

2) That is a possiblity but there is no reason to think that a disproportionate number of the 47 clusters -- which were selected from all over Iraq -- were in such places. In the 2004 study, the count from Fallujah was excluded because it was an "outlier" -- i.e., it presented an abnormally high death count.

3) Your comparison to Japan is apples-to-oranges: the number you cite for Japan takes into account only civilian deaths directly resulting from warfare, not all deaths.
10.18.2006 11:49am
Nick Good - South Africa (mail):
This Lancet Study posits a death rate in Iraq (population 27 million) higher than that sustained by either Britain (pop 47 million) and the US (pop 132 million) in WW2. Indeed in terms of death rate per month it's comparable to the combined total for both the UK and the US in WW2!

US fatalities in WW2 - 418,000. UK fatalities in WW2 450,000 including 68,000 killed in the saturation bombing of the Blitz.

WW2 was a high intensity war fought over land, at sea, and in the air by massive all arms formations against Germany, Japan and their allies in Europe, North Africa and Asia. All sides using their full industrial capacity to wage war.

Im being asked to believe that a few poorly co-ordinated Jihadis with RPGs, small arms and using IEDs are precipitating comparable carnage.

I'm not a stats expert, but I'm no fool and not completely without a handle on maths. This Lancet study rather fails the smell test.
10.18.2006 11:53am
alkali (mail) (www):
With respect to Richard Aubrey's points:

1) There is a civil war going on in Iraq and no effective civil authority in much of Iraq. Accordingly, significant numbers of men of military age may be dying from (i) fighting with non-US forces and (ii) general criminality.

2) There was an oil-for-food program before the war. Clearly, Saddam skimmed off a chunk of the oil money, but it did appear that enough food was being purchased to avoid famine, etc., during the pre-war period considered by the Lancet study.
10.18.2006 11:54am
MartinM:
The use of 'smell tests' to discard the results of a study with which one can find no methodological issue itself fails the 'smell test.'
10.18.2006 11:56am
alkali (mail) (www):
Nick Good: see my comment #3 on Clayton's points. The Lancet study takes into all deaths, not just deaths directly resulting from warfare. Your comparisons are therefore not well taken.
10.18.2006 11:57am
frankcross (mail):
The well known Iraqi blogger Zeyad lights into critics of the study as implausible, saying that they are ignorant of the situation in Iraq, where bodies continually appear in dumpsters.

But then he says he too thinks the number is too high and that the true additional number of deaths is probably 300,000-400,000.

There seems to be nothing really wrong with the methodology or implausible about the study (to those on the ground), but it is extremely difficult to conduct a good study under these circumstances. The number shouldn't be worshipped as accurate, but it's pretty clear that a huge number of deaths have occurred, somewhere in the hundreds of thousands
10.18.2006 12:10pm
Steve:
I haven't seen this criticism mentioned anywhere, but I wonder if there's something of a self-selection problem. Supposedly the interviewers went door-to-door in the selected neighborhood, explained their purpose, and if the occupants of a given house didn't want to participate they simply went to the next house. It's at least plausible that Iraqis who had lost a family member would be more likely to want someone to know about it. I wonder if there's anything to this point.
10.18.2006 12:19pm
Commenterlein (mail):
As several other commenters have already pointed out, the number of clusters is reflected in the fairly large confidence intervalls around the point estimate of 655,000.

The fact that Mr. Moore, who is described as follows

"... a political consultant with Gorton Moore International, trained Iraqi researchers for the International Republican Institute from 2003 to 2004 and conducted survey research for the Coalition Forces from 2005 to 2006."

doesn't know this means he is clueless and should not be writing about empirical research. The fact that Mr. Adler approvingly links to Mr. Moores hackery does not reflect well on Mr. Adler.
10.18.2006 12:20pm
alkali (mail) (www):
Steve writes:

I wonder if there's something of a self-selection problem. Supposedly the interviewers went door-to-door in the selected neighborhood, explained their purpose, and if the occupants of a given house didn't want to participate they simply went to the next house. It's at least plausible that Iraqis who had lost a family member would be more likely to want someone to know about it.

This is a good question, although it seems it could cut the other way as well (i.e., maybe people who have lost a family member would be less willing to talk to a survey team).
10.18.2006 12:22pm
Steve:
On a separate note, it seems to me that Moore's self-professed shock that demographic data was not collected from respondents seems exceedingly misplaced. Is there any reason to think that whether the respondent is a man or a woman makes it more or less likely that they have suffered a death in the family? When you ask about household deaths, does it make any difference whether Dad or Grandma is the one answering the questions? I don't see how demographic data helps confirm a representative sample at all.
10.18.2006 12:24pm
MartinM:
Supposedly the interviewers went door-to-door in the selected neighborhood, explained their purpose, and if the occupants of a given house didn't want to participate they simply went to the next house


But the response rate was very high - over 95%, as I recall. Not enough room for significant bias.
10.18.2006 12:25pm
Richard Aubrey (mail):
alkali.

Well, I don't suppose it matters much who kills the buttheads (MOMA) who are killing Iraqi civilians in sectarian warfare and general criminality. It remains a feature, not a bug. The point was that as a matter of excess deaths, it doesn't fit the implied narrative of poor and helpless civilians being killed, and, indeed, requires one to think against the narrative.

The problem with the second point is that the deaths of half a million kids, at least, was an article of faith among the libs opposed to sanctions. At least until the choice became sanctions or war, in which case the half a million dead kids got resurrected, or ignored. I forget which. If the Lancet is correct in its before death rate, the libs were lying like effing rugs. Surprise.
Other sources than the CIA factbook, the Lancet's source, put the before rate at close to 9/k/a. If true, this would substantially reduce the excess deaths, even if the after rate were held the same, accurate or not.
10.18.2006 12:34pm
WHOI Jacket:
Here's another real question?

It seems that the US is in a Lose/Lose situation here regardless. Say that troops don't go down the streets and destroy all who threaten the peace, then in a week from now, the pissants blow up a truck in the market place taking 2 dozen women and childern. Those deaths get counted into this total.

On the other side, say that troops engage in heavy fire with an insurgent group and wipe all 100 of them out in an air strike. Then, a week later when a survey group comes by and asks the sister of J'ohn Al-Smith, the leader of the group wiped out last week, if she has any dead, she'll say yes and that will go into the total.
10.18.2006 12:34pm
A.S.:
Alkali writes: UNDP was counting war-related deaths; the "Lancet" study was counting all deaths. If a child dies of typhoid, UNDP didn't count it. If someone is robbed and murdered, UNDP didn't count it.

Alkali's objection is misplaced.

The Lancet study didn't count "all" deaths. It counted EXCESS deaths since before the war. With the obvious implication that the excess deaths result from the war. I mean, is alkali trying to fool us that the alleged 655,000 didn't come as a result of the war??? Please.

Later, alkali says that the UNDP only counted deaths that are "directly" related to the war. What evidence has alkali for this proposition? None. Because it is wrong. The fact is that alkali has no idea what the UNDP "counted" as "war-related" deaths.

Moreover, according to Lancet II, there were NO excess deaths from non-violent causes in 2003, 2004 and 2005 (and hardly any in 2006). That means, for the period studied by the UN, the number of excess deaths from typhoid (alkali's example) counted by the Lancet is ZERO.

Let's face it, beyond any reasonable doubt, the UN study showed that Lancet overstated the number of excess deaths by a factor of about 4. Divide the excess deaths shown in the Lncet study by 4 - we get, what, about 170,000 - and we're getting to about the right number.
10.18.2006 12:34pm
Jiffy:
Steve:

The response rate of those in the clusters was apparently extremely high. In fact, the Iraq Body Count cites the response rate, which they contend to be implausible, as one reason to doubt the study results. But that high rate does at least suggest that within clusters there shouldn't be a "self-selection" bias.

Overall, I think the blog discussion of the study has been very interesting. I haven't seen any methodological criticisms that appear to hold up (most of the ones raised above have been well answered in my lay person's view). On the other hand, some of the "smell test" arguments, if you want to call them that, especially those raised by the Iraq Body Count do give significant pause about putting too much faith in the Lancet results.
10.18.2006 12:40pm
MartinM:
Other sources than the CIA factbook, the Lancet's source, put the before rate at close to 9/k/a


The CIA factbook was not the source of the pre-war death rate reported in the Lancet study. That rate was calculated using the same methodology as the current death rate.
10.18.2006 12:45pm
A.S.:
The CIA factbook was not the source of the pre-war death rate reported in the Lancet study. That rate was calculated using the same methodology as the current death rate.

No, but the CIA Factbook IS used by the authors of the Lancet study as evidence that the number they come up with is legitimate.

As pointed out, the other sources show that the pre-war rate used by the authors is likely NOT legitimate.
10.18.2006 12:51pm
Steve:
Among the arguments against the Lancet study, one of my favorites is the argument that the methodology simultaneously UNDERstated the pre-war death rate and OVER-stated the post-invasion death rate. That's a remarkably facile claim, which many have made without the slightest pause.
10.18.2006 12:59pm
Tim Lambert (mail) (www):
Moore does not understand basic statistics. It really is embarassingly bad.
10.18.2006 1:23pm
Nick Good - South Africa (mail):
There's also the timing of this report as well as the last one - both just before the US election. This and the fact that the Editor of the Lancet one Dr Richard Horton, is a well known 'Stopper' - here seen at a recent Stopper Rally sharing a stage with George Galloway.

It does not really add to the credibility of the thing.
10.18.2006 1:28pm
alkali (mail) (www):
A.S.: The UNDP 2004 study did not count death from disease, accidents, infant mortality, and general criminality as war-related. See the UNDP questionnaire at page 48 (question HM05).

Increases in deaths from those causes since the pre-war period are arguably "related" to the war -- this is the import of the Lancet study, I suppose -- but that's not the approach the UNDP took.

The Lancet study did find an increase in non-violent death since the war began, but the result was not statistically significant (which is not the same as finding to a statistical certainty that there was no increase).
10.18.2006 1:46pm
Greedy Clerk (mail):
David Bernstein: "Well, I pointed out in a recent post that Davies felt free to make up stuff without any evidence about me, so I don't see any reason to credit anything he says."

Well, if David says so then there is no reason to credit anything he says.
10.18.2006 2:02pm
MnZ (mail):
Lancet? Didn't they also publish a study that attributed the deaths of 500,000 Iraqi children to the UN sanctions in the late 1990s?
10.18.2006 2:07pm
J. F. Thomas (mail):
The claim is that 2.5% of the population of Iraq has died as a result of this war. As I pointed out here, for comparison, during World War II, the Japanese, over a period of six years, lost 2.7% of their population, both civilian and military--in a war in which we fire bombed Japanese cities--killing 100,000 in a single raid--and killed 192,000 with atomic bombs.

Of course Japan never suffered an invasion of its homeland either. A more valid comparison would be to look at the countries where the ground war was fought, especially in eastern Europe. Poland lost somewhere between 15 and 20% of its population. The USSR around 15%. Germany lost somewhere between 2 and 4 million civilians (during the last months of the war the line between civilian and soldier became very blurred) out of the total 7 million they lost. That is fully ten percent of their pre-war population. The vast majority of the civilians were killed in the last six months of the war after the allies crossed the German border.

So 2.5% of the population in what has descended into a sectarian civil war is not outrageously off the mark.
10.18.2006 2:07pm
Random3 (mail):
600,000 is obvious BS, you really don't need to argue methodological flaws to come to that conclusion. You can multiply numbers together all day long and really scare yourself about all kinds of stuff that isn't going to happen. I mean where are the bodies? Where are the tens and hundreds of entire towns and villages that have been destroyed? Geez. And why would so many of the supposed dead be young men? What is up with that? If they are going to extrapolate based on no friggin real data whatsoever, why not at least say 1 million deaths? Or even do a Dr. Evil and go for 1 BILLION? Just because the population of the entire country is only 27 million is no reason to hesitate. You guys would find a way to argue yourselves around to believing it.
10.18.2006 2:29pm
Richard Aubrey (mail):
J. F. Thomas.

In your description, the supposed losses are not wildly or obviously off the mark by any large amount. In other words,they are plausible.

The problem is that in all of those cases you cite, we had bodies by the truckload and trainload.

The Holocaust imported victims and killed them before the areas of the camps were overrun. That's probably fourteen million altogether, killed in what arguably was not war, and enough to decrease the population of the victims' origins by a fair number. So to use the results you have as a result solely of being overrun by war is not as accurate as it might be.
10.18.2006 2:33pm
Duffy Pratt (mail):
I'm with the "where are the bodies?" crew. Who is doing the burying? Where are the gravesites? Where are the funerals? If the bodies are not being buried, where is the rampant disease you might expect?
10.18.2006 2:55pm
Anderson (mail) (www):
It seems to me that the claims that Iraqis have bowel movements are grossly exaggerated. I mean, where is the fecal matter? In a nation of millions, wouldn't there be *tons* of it? But I don't see that on TV!
10.18.2006 3:05pm
Steve:
What's amazing is the number of people who seem to think Iraq is just crawling with international reporters on every street corner, ready to relay details of every body that turns up just like our local small-town police blotter does. From thousands of miles away, people are like "Where are the bodies? I don't see any bodies." It boggles the mind.
10.18.2006 3:11pm
Commenterlein (mail):
Let's say that in peace time about 1% of the population dies every year of natural causes and because of accidents. That's 270,000 people per year in a country of 27m. The pre-war number in Iraq was probably lower, but the 1% number is, as far as I recall, pretty accurate for many industrialized countries.

An additional 650,000 deaths in three years threfore requires less than a doubling of the natural death rate - absolutely no reason to expect villages or cities to be destroyed, or to expect holocaust-style death camps, or piles of bodies in the streets. Just twice as many funerals per year as you would observe in Belgium, France or Japan.
10.18.2006 3:13pm
JRL:

What's amazing is the number of people who seem to think Iraq is just crawling with international reporters on every street corner, ready to relay details of every body that turns up just like our local small-town police blotter does. From thousands of miles away, people are like "Where are the bodies? I don't see any bodies." It boggles the mind.



Rather, "[p]eople are like" if the media goes nuts for 20 bodies, then they'd cream their jeans to show us 1000.
10.18.2006 3:27pm
Duffy Pratt (mail):
Commenterlein, that may be one way of looking at it, and it might be right if there were just an overall, randomly distributed increase in the death rate. But, in Iraq you are talking about concentrated, violent death for the most part.

In the Battle of Normandy, about 43,000 allied troops died, and about 15,000 French civilians were killed, over a two month period (including D-Day, which is one of the bloodiest battles in history). That amounts to somewhere around 950 people killed each day, for two months.

The report estimates almost 900 Iragis dead every day for a three year period. That's the equivilent of the Battle of Normandy going on for 18 times as long.

With that kind of slaughter, I think its fair to ask where the coverage is, where the bodies are, etc... I suppose its possible that this is happenning. But if it is, it boggles the imagination. Maybe the U.S. could cover it up, but you would think that Al Jazeera might pick up on something like this.
10.18.2006 3:32pm
J. F. Thomas (mail):
In the Battle of Normandy, about 43,000 allied troops died, and about 15,000 French civilians were killed, over a two month period (including D-Day, which is one of the bloodiest battles in history). That amounts to somewhere around 950 people killed each day, for two months.

Normandy was not one of the bloodiest battles in history. Heck, more people were probably were killed in the battle for Berlin (the Russians alone might have lost 300,000 in the final push) than the Western Allies did in the entire European campaign. The Western Allies didn't fight a single campaign in World War II that even came close to the slaughter on the Eastern Front. 500,000 civilians in the siege of Leningrad. Probably a million Russians and 200,000 Germans at Stalingrad.

And for real senseless slaughter, look at World War I. The British had 60,000 casualties (including 20,000 dead) on the first day of the Somme in 1916. Tack on another 10,000 or so dead for the Germans and the French and it might have been the single bloodiest day in the history of warfare. And while that was going on the battle of Verdun was in its sixth month. The entire purpose of that battle was to "bleed France dry", and it very nearly succeeded. The battles of 1916 cost around 2.5 million lives. World War I averaged out to about a 6000 lives a day for the duration of the war.
10.18.2006 3:53pm
Colin (mail):
Commenterlein, that may be one way of looking at it, and it might be right if there were just an overall, randomly distributed increase in the death rate. But, in Iraq you are talking about concentrated, violent death for the most part.

Isn't that just what this study is about? An overall, distributed (but nonrandom) increase in the death rate? My impression, just from skimming the comments, was that it purports to show an increase in the death rate overall, not concentrated and violent death. Violence could even be a minority cause of death; it's not hard to see how a state of de facto civil war would skew the actuarial tables.
10.18.2006 4:17pm
Duffy Pratt (mail):
Of course you are right about the Eastern Front in WW2, and about most of WWI, but it still makes the top 25 on the 20th century hit parade. See http://users.erols.com/mwhite28/battles.htm

Ever visit the cemeteries in Normandy? The report says that 18 times that number of people have been killed in Iraq. I still think it's worth asking where the bodies are.
10.18.2006 4:31pm
Duffy Pratt (mail):
Colin:

Read the report. Deaths from disease increase six fold. Infant mortality rises some. So some of the deaths come from lack of medical care. Almost all the rest of the increase is attributed, according to the reports figures, to violent deaths and accidents. That accounts for the vast majority of the excess deaths.
10.18.2006 4:36pm
TC (mail):
I am really confused. For years before the war, I thought that the mortality rate in Iraq was skyrocketing because of the sanctions on Iraq (thus, we had to enact the Oil for Palaces Program) and the crippling damage done to the Iraqi infrastructure during the 1991 Gulf War. Now I've discovered that everything in Iraq was hunky-dory until the US invasion.
10.18.2006 4:39pm
TC (mail):
http://www.fff.org/freedom/fd0401c.asp

CBS correspondent Lesley Stahl relied on this estimate in 1996 when she asked U.S. Ambassador to the United Nations Madeleine Albright,

We have heard that a half million children have died. I mean, that is more children than died in Hiroshima. And, you know, is the price worth it?
Albright answered,

I think this is a very hard choice, but the price, we think the price is worth it.
10.18.2006 4:45pm
Colin (mail):
Almost all the rest of the increase is attributed, according to the reports figures, to violent deaths and accidents.

Shame on me for not reading before I wrote. But I don't see how, even if the increase was predominantly violent, that invalidates Commenterlein's point. You wouldn't need acres of mass graves or extirpated villages to reach these numbers.
10.18.2006 4:45pm
TC (mail):
http://www.peace-action.org/camp/justice/iraqfs.pdf
10.18.2006 4:46pm
Richard Aubrey (mail):
TC. Don't be confused. The statistics are consistently in accord with the pressing issue of the day. Find that out and you'll be right with the numbers.
Next day, new issue, new numbers. But at least you won't be confused or surprised.
10.18.2006 5:05pm
SR (mail):
TC-

The Oil for Food program began in late 1995. The pre-war mortality rate for the Lancet study was calculated based on the mortality rate in Iraq after nearly five years of oil for food revenues. That there may have been a higher mortality rate in 1991, 1992, 1993, 1994, and 1995 as implicated by Leslie Stahl is irrelevant to the mortality rate in 2000 -02.

It appears that many of the criticisms of the Lancet study on this thread are based on a decision that the numbers must be wrong, and then a search for anything that resembles an argument for why we should not credit the study. But, as just about everyone who knows about this stuff says, the methodology was sound. So, unless the data collection was flawed, there is a %95 chance that the invasion of Iraq has resulted in 400,000 to 900,000 extra Iraqi deaths.

Maybe if you don't see the bodies, its because you are not looking close enough.
10.18.2006 5:08pm
Duffy Pratt (mail):
Colin:

I would agree with you if there was any reason to think that the violence were somehow random, and not localized. But we have good reason to think that most of the violence now is occuring in Bagdad, where reported deaths from violent deaths are somewhere around 80-90/day. The report implies that there are many, many other violent deaths occuring -- at least 5 times the number that we know of in Bagdad. So, that leaves open the question, where are they? where are the bodies? etc...

I'm not saying the report is necessarily wrong. But there's good reason to be sceptical.
10.18.2006 5:15pm
Richard Aubrey (mail):
SR. I believe the mortality rate for those other years is relevant.
One issue is the remarkably low mortality rate the study cites for before the war (5.5perthousandperyear). That is about half the world average. Hungary, for example, has a rate of about 13/k/a. The reason is, it is said, that Iraq has a huge number of children, who, God willing have a long life ahead of them and aren't dying in anything like the numbers of Hungary's large elderly demographic.
For the argument that the kids lowered the pre-war death rate, you need live kids. But, as others have pointed out, the sanctions killed half a million of them. Supposedly. So do we have enough kids left over to drop the pre-war death rate to half of the world's death rate? Reasonable question. If yes, then the slaughter of the innocents is most likely a lie. If it isn't a lie, we have a problem with the pre-war death rate.
Question is, which lie do the liars prefer? Can't have both.

Besides, another source, I think it was the World Fact Book (I like literal titles) said the prewar death rate in Iraq was really 8.9. That means the excess deaths could be made up mostly of the MOMA--men of military age--who will probably not be missed by the folks trying to build a civil and peaceful Iraq.

It is certainly true that wars make things tough for everybody. It might be interesting to break down death rates in the US during WW II. Considering the huge draw of medical personnel into the military, was the reduction in help available responsible for increased civilian deaths? Hard to see how it could be otherwise, but I've not seen any work on the subject.
I imagine there were excess deaths in Iraq. The point is, how many, and would anybody be interested in working on it if the results would be published exactly between US elections. Probably not.
10.18.2006 5:22pm
SR (mail):
Do we know that most of the violence now is occuring in Baghdad, or is that just where most people are paying attention to the violence (due to the fact that most of the foreign press is there)?

According to the Lancet study, there has been a marked increase in violence just about everywhere. Further this increase in violence throughout the non-Kurdish areas in Iraq is confirmed by other sources as well.

I'm not saying the report is necessarily correct. But there's good reason to be confident.
10.18.2006 5:25pm
frankcross (mail):
Well, I think the preexisting death rate is badly analyzed. Those other sources, ranging from 5 to around 8 per K are obviously of uncertain reliability.
The Lancet study itself calculated a rate of around 5/K, but that was just a point estimate that had to be bracketed by confidence intervals. There are reasons to think it might be a little low, so don't worship the 600,000. But if the preexisting death rate were 8/K, probably as high as plausible, that would mean that Lancet calculations yielded excess deaths of over 400,000.
10.18.2006 5:36pm
M. Gross (mail):
One wonders why anyone even bothers with they newest study after the author's laughably bad previous study. In their last outing, they found an unadjusted crude death rate of 12 per 10,000 for the invasion period, compared to a 7 per 10,000 rate for the same period in their second study. (Numbers rounded so I don't have to go look them up again.)

How, pray tell, did they reconcile the differences? They simply dropped clusters in the original study until the number approximated what they wanted to see.

Why, after demonstrating their methodology was hopelessly flawed several years ago, did they attempt at second study using exactly the same method?
10.18.2006 5:46pm
Mitchell Freedman (mail) (www):
WHOI Jacket:

I am NOT saying the US is the Khmer Rouge. Far from that--as humanly possible.

What I am saying is that there are plenty of Sunni and Shi'ite extremists, and even Al Queda supporters in Iraq, who are actively killing civilians as a result of a poorly planned and executed war by the US. The Lancet report is not saying the US killed 650,000 Iraqis. The report is saying that the violence following the defeat of Saddam has resulted in those numbers of excess deaths.

I am also saying that those who would completely denigrate the Lancet report are sounding like they wish to deny the significant suffering of Iraqis. That's where the analogy to Chomsky's skepticism of the initial refugee reports from Cambodia came in.
10.18.2006 5:56pm
marghlar:
That means the excess deaths could be made up mostly of the MOMA--men of military age--who will probably not be missed by the folks trying to build a civil and peaceful Iraq.

That's right. Because there is absolutely no value to the life of any Iraqi male between the ages of 18 and 45.
10.18.2006 6:08pm
Clayton E. Cramer (mail) (www):

I am also saying that those who would completely denigrate the Lancet report are sounding like they wish to deny the significant suffering of Iraqis. That's where the analogy to Chomsky's skepticism of the initial refugee reports from Cambodia came in.
I'm not denying the significant suffering of Iraqis. There are monsters running around Iraq torturing people to death with power tools, and who revel in taping themselves beheading conscious people. But the crowd that crows about the Lancet paper's quite incredible numbers are in favor of immediate withdrawal of all Coalition forces--leaving the monsters as the likely winners.

By the way, doesn't anyone find it just a bit implausible that the pre-war Iraqi death rate could be so low, when Hussein murdered hundreds of thousands of Iraqi civilians with poison gas, by torture, by forcing people to drink gasoline and then shooting them? And we are to believe that the situation today is far worse?
10.18.2006 6:10pm
te:
I am not a statistical epidemologist. Is anyone else here one?

I have seen several comments about this not passing the "smell test" and some people saying that they thought the sample was inadequate. But isn't this like all the tort reform bozos who just prattle on after a verdict that they don't like for no real reason other than they don't like the result?

I have no idea whether the 600K estimate is sound. It seems high to me. But I would be interested in reading a solid criticism of it - rather than the smell test nonsens.
10.18.2006 7:00pm
frankcross (mail):
I think M. Gross's post is just flat out false.

The 12.3 figure in the first study was through September 04. The 7.5 figure in the second study was through April 04. When you compare the same time period in the two studies, the results are similar.

Clayton, Hussein did murder many Iraqis but not in the time period immediately before the invasion, where the death rate figures are from. No one contests this.

If you want to be an ideologue, fine but do it honestly.
10.18.2006 7:44pm
Steve:
By the way, doesn't anyone find it just a bit implausible that the pre-war Iraqi death rate could be so low, when Hussein murdered hundreds of thousands of Iraqi civilians with poison gas, by torture, by forcing people to drink gasoline and then shooting them?

Although Saddam undoubtedly perpetrated horrible acts, the vast majority of the deaths that are ascribed to him were attributable solely to the Iran-Iraq war, which was long since over by 2003. While Saddam was surely torturing and killing dissidents during the postwar period, no one contends that the line stretched around the block. In a nation of 24 million, you have to make a lot of people drink gasoline to create a blip on the death rate.
10.18.2006 8:14pm
byomtov (mail):
Jonathan,

In view of your update, where you recognize that his "not enough clusters" claim is foolish, do you still feel that,

"Moore seems to mount a serious challenge to the validity of the Lancet estimates."

If not, why do you ascribe such great importance to Moore's other argument about the lackof demographic data?
10.18.2006 8:16pm
htom (mail):
I had two quarters of statistics more than a quarter of a century ago and have rarely used any of it, so I am in no way any kind of statistician.

What struck me about the survey method was the great care to do things "randomly", until we came to the second household, which was to be adjoining the first. The third to be adjoining the second, ... the fortieth adjoining the thirty-ninth, and it seems an entire cluster could be counted in an apartment building. This strikes me as being not the best possible method, and indeed, a rather bad method. Two hundred clusters of ten households would seem to be a better scheme.

It would be interesting to see a comparison of results from the first household at each cluster, the first three visited at each cluster, and the first ten. There may have been some effect caused by the long chain of visits.
10.18.2006 10:04pm
Richard Aubrey (mail):
Marghlar.

Try not to be so silly.

The dead MOMA were thugs, killing each other when they couldn't kill civilians or Americans.

They would have been a major problem for normal people trying to build a stable society.

It is not true they have no value. They have a negative value.
10.18.2006 10:13pm
Elliot123 (mail):
If I were an Iraqi man between the ages of 18 and 40, it might be very convenient for my family to have a death certificate hanging around when US troops or the militia for the other side knocked down the door looking for me.

I also wonder who was issuing all those death certificates? Did Saddam's whole government bureaucracy break down except for the coroner's office?

Exactly how does one get a death certificate in Iraq?
10.18.2006 11:12pm
vic:
i hope that the commenters accusing the study authors and publishers of liberal bias are aware that the john hopkins school of public health faculty have a healthy disdain of liberals. Not because they are conservative but because the school is so completely dominated by Trotskyites.
10.19.2006 12:35am
Brian Garst (www):
Whether their bias is liberal or not is beside the point, as they are most certainly anti-war. But even that isn't the issue. The fact of the matter is their study has been debunked not on their motives, but on its merits, or lack thereof in this case.
10.19.2006 2:38am
Commenterlein (mail):
The fact of the matter appears to be that anyone with training in statistics and empirical methods thinks that the study looks solid, and all wingers claim that the study has been "debunked".

What's telling is that there seems to be no overlap between the two groups.
10.19.2006 2:53am
Richard Aubrey (mail):
Commenterlein:

There's a third group. When I had my unfortunate experience with stat--in the Psychology Department--I learned how to make the procedures sing, dance and recite.
I also learned GIGO.
Some of us think there is a good deal of G in the study. The uses to which the study are put are, of course, not the fault of the study. BS.

One of the authors has been a raving anti-war type. If this were a revelation casting doubt on, say, the validity of Potassium-Argon dating, and we were to discover the author of that study was a featured speaker at Creationist rallies, I expect there would be considerably less deference given to the "results".

To repeat something whose importance has been brushed aside: There is strong concern about the putative death rate for before the war. Methodology and stat are useless if the quantities used are false.

I would be surprised if there were no excess deaths. I would be surprised if you could find anybody who said there were and could have been none. Remember, the 655,000 number is one of the possibilities. It could be lower, it could be higher. To insist the facts point to it being lower is not a matter of debunking the study.

To do a couple of numbers: If the "before" rate were not 5.5/k/a, but 9.5/k/a, and the "after" death rate were constant, whatever it was, the reduction in the diffrence would be in excess of 100,000.

If the highly trained Iraqi volunteers included some who knew the most important terrain in this war is the six inches between the ears of the American voter, and their results got punched up inadvertently (riiiiight), and considering the impact a couple of ringers could have--there being so few interviews overall--another 100,000 might be completely bogus.

Ringers, I hear you say. Heaven forfend. Just as an example, Time magazine recently ran an obit on one of their correspondents who was moonlighting as a spy for Hanoi. Don't say it can't be done. The ease of countering that is too easy.
10.19.2006 8:43am
Colin (mail):
The dead MOMA were thugs, killing each other when they couldn't kill civilians or Americans.

Like most sweeping assertions, that one doesn't pass the same "smell test" you're trying to apply to this study. The insurgency has victims as well as perpetrators; not all of the men of military age who have died were "thugs." Nor is it reasonable to say that they "won't be missed;" a missing generation will have serious demographic effects down the line.

To do a couple of numbers: If the "before" rate were not 5.5/k/a, but 9.5/k/a, and the "after" death rate were constant, whatever it was, the reduction in the diffrence would be in excess of 100,000.

And if that were solid data, the conversation would be over. Do you have any data? From here, it looks like you're contesting the initial values because the study's results clash with your political preconceptions.
10.19.2006 11:42am
Richard Aubrey (mail):
Colin. Wrong about my motivation.
My concern about the "before" rate stems from the fact that it is impossibly low, compared to the rest of the world. Whether that fits with my political views or not is a separate issue. I suppose you're fine with the study's before-war rate because it tends to support your political preconceptions.

The second problem with the study's before war rate is that it makes us wonder about the validity of the howls about how the sanctions killed so many. How could that be true and the before rate still be so low? Somebody's putting somebody on.

Yeah, I'm sure the study is lamenting all the Iraqi cops and soldiers killed fighting the insurgents--not. They're heroes but, despite their terrible casualties, the numbers don't make much of an impact on the 655k.

If political preconceptions invalidate an assertion, the authors of the study shouldn't have wasted the paper, considering THEIR political preconceptions. I think they're getting a huge break in that people are actually looking at the study instead of dismissing it outright because of the clear preconceptions of the authors. Which, if your principle were principled, you would have done immediately.
10.19.2006 11:54am
Colin (mail):
My concern about the "before" rate stems from the fact that it is impossibly low, compared to the rest of the world. Whether that fits with my political views or not is a separate issue. I suppose you're fine with the study's before-war rate because it tends to support your political preconceptions.

I don't have any data that suggests that it is wrong. Neither, it seems, do you, other than comparing the pre-war date to the global average. You are apparently starting from the position that the study must be wrong, and hunting for a reason why.

If political preconceptions invalidate an assertion, the authors of the study shouldn't have wasted the paper, considering THEIR political preconceptions.

One of the many differences between their assertions and yours is that they have actual data to support their statements. They also have a methodology that, from the statistical analyses offered in this thread, hasn't been convincingly refuted. You have a stong partisan outrage, but not much else. Can you see why I would credit one assertion over the other?
10.19.2006 12:04pm
Justin (mail):
Darkies aren't human, read all about it!!!!

Or that's basically the arguments made by Aubrey and Cramer, who have the unbelievable gall to call themselves Christians and act holier than thou on these boards.
10.19.2006 12:05pm
Richard Aubrey (mail):
Colin. You may not have any data. It exists, and it's been cited. But I won't bother repeating it, since you are hunting for reasons to believe the unlikely.

The data I have to contradict the before rate is that it is impossibly low considering the rest of the world. That is a fact which should be addressed. It requires, because it is an outlier, and because of the impact it has on the results, some kind of additional scrutiny.

Secondarily, if that rate is true, then the story of the murderous sanctions is called into question. My question is, Colin, which story is more important to you? You can't have both. Pick one. Let us know.
10.19.2006 12:19pm
byomtov (mail):
Richard Aubrey,

My concern about the "before" rate stems from the fact that it is impossibly low, compared to the rest of the world. Whether that fits with my political views or not is a separate issue. I suppose you're fine with the study's before-war rate because it tends to support your political preconceptions.

The CIA gave an estimated 2003 death rate of 5.84. Who exactly is ignoring data because of preconceptions?

The data I have to contradict the before rate is that it is impossibly low considering the rest of the world. That is a fact which should be addressed. It requires, because it is an outlier, and because of the impact it has on the results, some kind of additional scrutiny.

This is simpy false. The CIA factbook gives recent death rates for some other countries in the area as:

Syria 4.81
Iran 5.55
Libya 3.48
Israel 6.18
Saudi Arabia 2.58!!

Further, these numbers have been endlessly pointed out on various web sites.
Given their ready availability, I think you are the last person who should be accusing others of arguing in bad faith.
10.19.2006 12:32pm
Richard Aubrey (mail):
Other sources--one of which I mentioned--had the Iraqi before rate at 8.9. That means, taking the Iraqi population at 27 million, and the rest of the data as purported, the number is down by about 100,000.

The other countries in the area you cited weren't having sanctions killing, among other things, half a million kids. The death rate is particularly sensitive to the number of kids. In less developed countries, to the number of kids past age five, the survivors, because they aren't dying. If you have a big kid demographic, you will have a low death rate compared to some other makeup.
So what happened to the sanctions story? I guess you won't be peddling that one, will you?
10.19.2006 12:49pm
byomtov (mail):
Other sources--one of which I mentioned--had the Iraqi before rate at 8.9.

What sources? The World Fact Book that you mentioned is the CIA publication from which I drew my figures.

So what happened to the sanctions story? I guess you won't be peddling that one, will you?

I have never cited any figure related to sanctions. Got anything else to spew?
10.19.2006 1:02pm
Richard Aubrey (mail):
by.

What happened to Colin? I thought I was talking to Colin.

The fact book is CIA. I was wrong about that. I saw the other number in a different thread.

Whether you cited anything regarding the sanctions is beside the point.
Let's say it's true that 500,000 kids died, and say it was over ten years. That's 50,000 a year. Do the math. It's the equivalent of a change in the death rate of 1.85/k/a, taking the Iraqi population as 27,000. That's one reason the before rate is suspect. That would bring the non-sanctions death rate down to 3.65/k/a. That, when the sanctions were destroying, among other things, the Iraqi medical system.
Secondly, that number of kids removed from the population would not be here after the war. Since kids depress the death rate, their absence would increase it. If sanctions caused the slaughter of the innocents as advertised, that would be one reason the after rate is high. And that would be without any of those kids dying "after", since they, supposedly, died "before". Their absence reduces the offset against the deaths of the elderly, which has a higher rate.

So. If sanctions killed as advertised, we have one reason to doubt the study's results. If we do not doubt the study, the sanctions did not kill as advertised.

I don't expect there are no excess deaths. I do expect the numbers given are wrong, and for some of the reasons I've given.
10.19.2006 1:15pm
Colin (mail):
I'm here, but busy. I share byomtov's skepticism about your figures and disregard for your hand-wringing over whether the sanctions raised the death rate. Neither of us have raised it here. You've backed yourself into an argument that sounds, to me, something like, "Well, if THIS liberal lie isn't a lie, then this OTHER liberal lie must be!" Pardon me if I'm not impressed by your reasoned objectivity,

Even as to the sanctions question, your argument doesn't seem to have much relation to fact. Using entirely invented numbers, you calculated as follows:

That would bring the non-sanctions death rate down to 3.65/k/a. That, when the sanctions were destroying, among other things, the Iraqi medical system.

In other words, the "non-sanctions death rate" would have been X, when it was affected by sanctions. I question your methodology.

I appreciate that you don't like the results of the study. But your complaints are sounding more and more like pure partisan hackery.
10.19.2006 1:32pm
Richard Aubrey (mail):
Colin. The sanctions issue, if there was one, and the current debate are connected, since they each affect, or report, on demographics.
And, yes, if one is true the other can't be. Unless you can tell us how both could be true.
Yeah, if sanctions killed, then the non-sanctions death rate would have been 3.65, or something else in that neighborhood. Is 3.65 realistic in a country whose medical services are falling apart? That last seems to be true, based on the experiences of troops and reporters in the first weeks of the war.
So, I have a problem with the starting point. I have no problem with the end point, other than to say that if journalists are having trouble getting the real skinny, since they don't know who their stringers are working for, the survey might have had the same trouble. I see no reason not to address that. Remember, the Iraqis stood down an entire police brigade due to terrorist infiltration. What prevented the study from having one or two activists slide in?
I don't say they did. I do say that, given the situation, the issue ought to have been addressed.

What, by the way, is partisan about seeing lower or higher numbers? Strike that. We know that there can't be enough dead civilians for a certain group in this country. But what is partisan about seeing lower numbers? After all, if object sought through war is worth killing people, it would be pretty silly to meter the numbers and say, oops, we hit our planned excess death target. Time to go home. If you're going to do that, don't even start.

But, yeah, wrt one of your points. One lie refutes the other. Tough, that.
10.19.2006 3:18pm
Colin (mail):
Aubrey, you haven't connected the sanctions issue with the current study. Your pre-war death rate assumptions, much less the off-the-cuff calculations you're trying to do with them, are blue-sky guesses. Your assessment of the numbers is based on nothing more than your gut feelings that some inputs can't be accurate because they produce politically unacceptable outputs. Even for a blog comment, that's shaky analysis.

What, by the way, is partisan about seeing lower or higher numbers? Strike that. We know that there can't be enough dead civilians for a certain group in this country.

And that right there is when I realized you aren't even trying to be serious or objective.
10.19.2006 3:26pm
Richard Aubrey (mail):
Actually, Colin, the two are connected, and clearly, for those whose preconceptions won't be damaged by the connection.
If they died in the sanctions, they weren't available to die after the war started. Seems a reasonable view.
The guesses are based on the widely-touted claim of half a million kids dead. I used a ten-year period to accumulate that number. I could have used a different one and had a different result. But I didn't make up the half a million. Somebody else did. The point is, if the sanctions did kill, the pre-war non-sanctions death rate was practically invisible. Whatever the result, the prewar non-sanctions death rate would have been less than the 5.5, and thus even less believable without being specifically addressed.
Because the dead I'm talking about here were or are kids, they had, or would have had, a particularly strong impact on the post-war rates, so their presence or absence is particularly important.
If sanctions killed, the entire study has to be viewed as needing correction. It seems reasonable to say the reverse, which is that if the study doesn't need to be revisited on this account, then sanctions didn't kill.

Do you have a way of insuring us that, although journalists and the Iraqi policy are infiltrated, the Iraqi interviewers were not? As I say, I have no reason to doubt the post-war rates, except that they may have been rigged by one or more interviewers. That should be addressed.

"What, by the way, is partisan about seeing lower or higher numbers? Strike that. We know that there can't be enough dead civilians for a certain group in this country."

And that right there is when I realized you aren't even trying to be serious or objective.


Buddy, I have never been more serious. I've watched the left since the late Sixties. In 1987, I was in El Salvador with a bunch of the usual suspects. We learned, from unimpeachable authority--which is to say nobody doubted it--that the death squad killings were down by about 98%. The disappointment among my colleagues was obvious.
One writer whose work we used, Sarah Miles, wrote of Low Intensity Conflict, that, in Central America, it worked. And it reduced casualties. Which was a problem. Those of us, she wrote, who are used to raising the consciousness of the American people by pointing to mass casualties will have to use different techniques (paraphrase). I've been on mailing lists of people who thought I was on their side, Colin. They're not as cautious as they should be.
Yup. There can't be enough dead civilians for the left, if the US can be blamed. If not, the left doesn't give a rodent's patootie.

As regards this study, people didn't say, hey, look at this.

They said, with ill-concealed glee, "LOOK AT THIS HOW BAD IS THE UNITED STATES MURDERERS" either actually or by connotation.

Never been more serious.
10.19.2006 4:52pm
Colin (mail):
Oh. I didn't realize you have been "watching the left since the late Sixties," and have "been on mailing lists of people who thought [you were] on their side." Please forgive my rash insinuation that you were indulging in partisan ranting.

"the two are connected, and clearly, for those whose preconceptions won't be damaged by the connection. . . . Seems a reasonable view. . . . . The guesses are based . . . on the widely-touted claim . . . I used a ten-year period . . . I could have used a different one and had a different result. . . . I didn't make up the half a million. Somebody else did. . . . It seems reasonable to say . . ."

In other words, made up numbers, blue-sky guesses, and back-of-the-envelope calculations designed to refute a study that you find politically incorrect. No citations, hard data, or serious analysis. At first you were directly challenging the reported prewar death rate, only to have your numbers solidly refuted. Now you're guessing about the effect of sanctions to arrive at a sans-sanctions death rate, then guessing about the demographic effects on the statistical analysis in the study. All, apparently, to counter the Evil Liberal Death-Loving Conspiracy What You've Been Watchin' Real Close Now For Forty Years.

Nor am I impressed by your hand-wringing over the possibility of infiltrators manipulating the data. You've started from the position that the study must be wrong, and are now searching for reasons to give for that conclusion. It should work the other way around. You're saying that the prewar death rate was wrong. If not that, then infiltrators are twisting the data. If not that, then the sanctions reports were wrong. If not that, then somehow, for some reason, no matter what, whatever you've identified as the "the left" must be wrong.

There can't be enough dead civilians for the left, if the US can be blamed. If not, the left doesn't give a rodent's patootie. As regards this study, people didn't say, hey, look at this. They said, with ill-concealed glee, "LOOK AT THIS HOW BAD IS THE UNITED STATES MURDERERS" either actually or by connotation. Never been more serious.

You've dragged this thread to a nadir. I think the serious analysis above has shown that whatever criticisms may be leveled against this study, Moore's is badly flawed. Past that, all we have is your hysteria, and that's not holding my attention. I'll leave you to rant in peace.
10.19.2006 5:37pm
Richard Aubrey (mail):
Cherry picking might seem useful, Colin, but it isn't.

I didn't start from the position that the study must be wrong. I looked at it and asked what questions remain. One is the pre-war rate. (Ended my senior moment. World Health Organization--Iraqi death rate 2001 8.9.) Using that takes about 125,000 excess deaths off the results, whatever they are.
The other question is what, if any, precautions were taken to avoid ringers in the survey. If I were doing the study, I'd certainly have a procedure and detail it so people could judge whether they thought it was adequate.

It is not that either the pre-war rate was wrong OR the post-war rate was faked. It's possible both were. Or both were valid. Neither concern has been addressed except by spittle.

The sanctions being right or wrong aren't conditional on one result from the study or another.

The sanctions reports were either right or wrong, but if they were right, the study has a problem. I, personally, don't believe in all the lurid tales of how the sanctions killed huge numbers of Iraqis.

The problem is with the left trying to insist both on the murderous sanctions and the higher end from the Lancet study. That's a problem.

I don't say the left MUST be wrong. Sometimes they're right. But that's not the way to bet.

Before calculators, people often used the back of an envelope. I'm not the one who made up the half a million dead kids. I simply chose to calculate they died over a ten year period. You can pick some other period--longer than ten years might not fit the historical record--and redo the math. The result will be different, but the implication is the same.

Hysteria? Reconcile the CIA--the lefty's favorite alternative to the Britannica--and WHO. Prove there were no ringers. Easy enough questions.

If you don't like the made-up number of half a million Iraqi kids dead of the sanctions, then I suppose you can show us contemporary docs where you chided those who made them up for fibbing? If not, then the number is out there and you didn't mind it. So why are you complaining now?
10.19.2006 7:48pm
Lev:
Perhaps this is too late for this thread, but I have a question.

It seems to me as a logical matter, that a more representative sample of a population will be obtained by more clusters of fewer people than by fewer clusters of more people.

How do the statistical analyses account for differences in numbers of clusters/number of interviews? Say 50 cluster points for a sample 2000 interviews vs. 250 cluster points for a sample of 2000 interviews.
10.20.2006 1:44am
Richard Aubrey (mail):
Lev. It's pretty simple: The interviewers were shitting bricks every moment of their work. Understandbly. The fewer clusters, even if that meant more people per cluster, the less exposure to death in one or more gruesome ways.

The study took what it could get.

Your concerns can also be addressed to the journos who don't leave the green zone and depend on Iraqi stringers who, at best, know what kind of reporting gets them paid.

In 1970, a class at Ft. Bragg happened to be watching a captured film of Viet Cong activity. Too bad for one guy in the movie. He had improved his station in life by getting a gig teaching Vietnamese at....Ft. Bragg. He was not tenured.

Never discount the relative facility of infiltrating a large organization made up of bighearted Americans who think all you guys look alike. Never discount the other side's view of the utility of such infiltration.
10.20.2006 9:13am
Sceptic:
htom 10.18.2006
You have put your finger on one of the study's major weaknesses.

The other is the study's questionable baseline.

Here is some informal material questioning the baseline:
http://channel4newswatch.blogspot.com/
Note
WHO ~ 9/1000
UNICEF ~ 13/1000

N.B. CIA source supporting 5/1000 pre-invasion did NOT support Hopkins' claimed massive increase post-invasion.
10.20.2006 12:21pm
Richard Aubrey (mail):
Sceptic. The Unicef figure would reduce the pre-post difference by, conservatively, about 202,000, compared to the CIA's 5.5 figure.

I guess the folks pick the one they like, that being the one giving them the biggest difference.

Hmm.
10.20.2006 2:39pm
Sceptic:
10.21.2006 5:48am