Death rates and death certificates

Death certificate inconsistency objection in good faith, that is. Preview is my friend, but I treat him so badly.

Awhile back I read about a case in India where a man had been declared dead by his inlaws, and had to spend several years in court proving that he wasn’t.

The dead Iraqis are unlikely to get lawyers to get included in the stats. Perhaps we should set up a fund for them to do that.

Wasn’t the UN study based on Iraqi health ministry numbers? Jane seems to be suggesting that the UN did an exhaustive survey of hospitals in Iraq, but I don’t remember a survey with that methodology (not that I’ve been watching it as carefully as I probably should… quite plausible I missed it). Even this would obviously give an undercount, but I don’t know how well the error could be estimated.

grossly overcounting deaths

Come again?

So wouldn’t it be better technique to collect the data from the hospitals directly?

Sebastian-

Regarding crude death rates, the World Bank posts some numbers at its web site. Iraq´s is put at 9 for 1990; 10 for 1995; n.a. for 2000; and 9 (italicised, presumably to indicate uncertainty) for 2004.

This compares to an average for the Middle East and North Africa of 8 (1990) and 6 (2004).

But if we look at Iraq’s immediate neighbours, we get (for 2004):

Turkey: 6
Syria: 3
Jordan: 4
Saudi Arabia: 4
Kuwait: 2
Iran: 5
It would appear that the Lancet estimate of 5.5 for 2002 is quite different from the World Bank’s estimates — but at the same time it is much closer to the general level of the region in which Iraq is located.

(Btw, I checked Brazil and Mexico as well, and this source differs from the numbers given by poster sebastian holsclaw — they’re much lower, at 7 and 5 than what he reports (around 10 for both, as should be expected for countries with younger populations)).

“I think that people may be misunderstanding the issues here. Iraq is in a state of complete anarchy. It is not a place where you can just rock up to the hospitals, smile at the receptionist and pick up a neat manila folder full of the week’s death certificates. Even if anyone had the resources and armed guard necessary to carry out this little fantasy, they would use them for something else.”

Ok, this is where I have a problem. It is too chaotic for people to go to hospitals (which have a discrete small number) but they can do a household survey? That is the kind of thing that makes me skeptical.

stostosto, I had already made a correction on the Mexico numbers. (on the other thread) :)

“because you have no idea how many people use a given hospital.”

Is that really an unsolveable problem? Heck by focusing on large cities you don’t have to ‘scale it up’. You can get a hard number for discrete areas. The number might not be as flashy as 600,000, but you could say things like: in hot spots like Baghdad and Fallujah X hundred thousand people were killed. Since we all accept that the killing are clumpy, why not focus on getting really accurate numbers out of the clumps instead of crappy numbers that you can ‘scale’ up.

Daniel,

Then the person who issued the death certificate is meant to send a copy to the central government records office where they collate them, tabulate them and collect the overall mortality statistics. This bit of the death infrastructure is not still working in Iraq.

What sources do you have for this? I’ve no doubt it is true to some extent, but to what extent?

I’m also far from convinced about what you presuppose is the inherent mortal danger attached to asking hospitals to produce the copies and records that presumably they have kept, if not submitted to central government.

I’m not a right-wing nutter and I’ve been careful to say little or nothing about statistical models – I’m simply not qualified. But I don’t think it’s going out on a limb to suggest that the Ministry of Interior, for example, when producing their mortality figures, asked hospitals and morgues for their input. In fact, we know they did. So unless the argument is that hospitals are issuing certificates without retaining a copy and/or recording the issue, or deliberately concealing this data, the disparity in the latest Hopkins figures and those produced elsewhere remains unexplained to any satisfactory degree.

As per Hopkins, at least 80% of deaths are certified (92% of 87%). So 80% of deaths are certified, but as of 2 days ago, nobody knew anything about 9/10ths of them.

You don’t have to be an inveterate decent to be unconvinced.

“There could be more than one way to study this, therefore this way must be invalid and probably fraudulent.”

No, I’m saying that a study which implies a civilian death rate higher than Germany during WWII deserves a skeptical eye since we aren’t, you know, firebombing cities.

I realize that I’m being annoying, but anyway.

What we seem to have established today is one or all of three things.

1.) Some people have a pure, non-political interest in the Iraq war primarily as an object upon which to practice statistical method.

2.) Some people think that while 600,000 extra dead might be regarded as awful, if that number were lower by a factor of five or so, that would be non-awful. And it’s awful if things are awful.

3.) Some people, for whatever reason, want us to talk about statistical methods rather than the Iraq War.

I suspect that one place where falsification would take place would be at point where someone might think “If I tell the truth I might be killed.” We already know that people have deliberately falsified death statistics or simply stopped reporting them.

BTW, it’s at home, but I have a book on Vietnam which mentioned how something similar happened there. The official figures for the number of cases of some disease (I think it was bubonic plague) was actually lower than the number of cases in the files of one particular hospital. I should check the details later.

I’m a numbers fetishist myself, but John Emerson does have a good point. Iraq is a total disaster and we don’t need exact death tolls to know this.

But I’ve got my own good point which I intend to beat into the ground–any rightwinger who criticizes the Lancet numbers and doesn’t call for an independent investigation fully supported (but not run in any way) by our government is someone who is more interested in suppressing bad news than in finding the truth. The US has just been accused of causing more deaths than have occurred in Darfur–apparently it’s enough just to point out why this may not be true.

The Lancet authors, to their great credit, have called for such a study twice. Too bad their critics could not care less.

It is too chaotic for people to go to hospitals

Perhaps relevant:

Iraqi Hospitals Are War’s New ‘Killing Fields’

In growing numbers, sick and wounded Sunnis have been abducted from public hospitals operated by Iraq’s Shiite-run Health Ministry and later killed, according to patients, families of victims, doctors and government officials.

As a result, more and more Iraqis are avoiding hospitals, making it even harder to preserve life in a city where death is seemingly everywhere. Gunshot victims are now being treated by nurses in makeshift emergency rooms set up in homes.

Spartikus gives us more evidence that Iraqi statistics are no good, and that things might be perfectly all right there!

Brownie:

As mentioned in another thread, it’s documented that the Iraqi health ministry figures for civilian deaths in Anbar province in all of July, 2006 was exactly zero. Nil. Nada. Zip. The UN assumed this was because any figures were impossible to retrieve under the current conditions there, and footnoted the numbers they cited in their report accordingly.

I think we can assume the total deaths in that province, at least, was somewhat more than that. The hospital/morgue count is an undercount: can we concede that much?

I do not see the point of pretending to compare mushy numbers from Iraq with mushy numbers from Darfur and mushy numbers for the Congo. This is pushing Bush’s fuzzy math to the limit/

While we’re at it, can we also all stop referring to the 600K violent deaths or 650K excess deaths in the Lancet as “civilian deaths?” The Lancet paper does not make either claim.

Sebastian, you have consistently been misquoting numbers on e.g. German death rates in WWII.

I followed this debate on the earlier study too, and so far the only reasonable criticisms of the study raised have been: A) recall bias might be greater for pre-war deaths, and B) deaths and casualties include insurgents, not just civilians. Everything else has just been a version of “I refuse to believe this, because I am committed to my own preconceptions over empirical evidence”.

Daniel writes:

When someone dies, you get a death certificate from the hospital, morgue or coroner, in your hand. This bit of the death infrastructure is still working in Iraq.

Could you provide a citation for this claim? Something on-line would be great. But if this was something told to you by, say, a reporter just returned from Iraq, that would be good to know.

The reason I ask is that this seems somewhat inconsistent with accounts that I have read in places like Fiasco. The claim is that in many areas of Iraq (e.g., Falluja, Ramadi) there is/was no local government to speak of for long periods of time. There was no coroner or morgue to hand out death certificates.

I just want to understand the facts better.

Also, can you clarify why everyone keeps death certificates? I have read that the certificates entitle one to various benefits. Is that true? If so, is there any reason to suspect that some certificates are issued fraudulently?

Again, I highly doubt that these issues affect the overall tenor of the results, but details are good to know.

Start with posts 30 and 31, Brownie. Then read 34 and 36.

No, Brownie, you are clearly not objective.

But you also miss the point. There are lots of numbers floating around about the casualty rates (amongst other things) in Iraq. That doesn’t mean that all the numbers are equivalent and can be meaningfully compared.

First off, you have laughable numbers like the `30,000′ that the president is still repeating. This is a political number, though, and shouldn’t be taken at all seriously (why? because it doesn’t bear up under the slightest scrutiny).

Now we also have numbers like the IBC’s ~50,000. This is not even claimed an estimate of the total civilian deaths, let alone an estimate of the total excess deaths. What it is is a pretty solid number of verified civilian deaths. By nature this is an undercount, and their methodology doesn’t have anything to say meaningfully about about the amount of undercount.

Finally, you have numbers like the lancet study or the ~250/day number you quote.

We can forget the political numbers, of course. Of the others, you have to be careful first off to make sure they are counting the same things. Many people in the last couple of days have said stupid things about the comparison between IBC numbers and the lancet study.

Given that you have two numbers actually talking about the same thing, you can attempt a comparison. Here is where the thing about objectivity comes in. I’m not sure you actually understand this, so I’m going to spell it out. The objective thing to do when presented with two significantly different estimates is not to say `wow, how could the new estimate be so much larger, it can’t be right’. That’s simply nonsense. The thing to do is say `hmmm, one or both of these estimates is wrong. Lets look at the methods used and see if we can’t understand why’.

Yelling loudly that we can’t possibly have had ~750 d/day in Iraq and not seen it in the news isn’t evidence. That’s a bald faced assertion. Coming from the average north american keyboard commando who can just about find Iraq on a map in a pinch it also looks pretty stupid.

The fact of the matter is we don’t know. Even the people who *ought* to know best what’s going on over there have a pretty sketchy idea. So all we can do is look at the methodology of these estimates and try and see which is a better approach. This is why nobody can expect precise numbers. The best we can hope for is accurate numbers — and the nature of doing measurements in Iraq today means the intervals are going to be large. Period.

As it is right now, the overwhelming majority of the objections I’ve seen to the Lancet numbers boil down to wishful thinking. There are a few interesting questions about methodology, but nothing devestating. Other questions which rightly belong to a larger scale discussion of the methodology, not this study (regardless of the validity of the methodological comments).

That you don’t want the numbers to be this high constitutes no serious objection.

Can you offer a serious objection to the methodology? Can you offer any serious argument that another estimate has significantly better methodology? Or are you going to be like thousands of other jokers offering `i can’t believe..’, ‘clearly…’, ‘obviously the morgues in Baghdad’, ‘surely the media…’ and other mindless speculation about the day to day reality in country you can barely imagine, and a lot of effort has been spent to keep you fairly ignorant about?

so?

I have not read the study, so I can’t claim whether it is valid or not. See, that is how reality works. You actually read something before you decide if it is wrong or right.

Yet I’m willing to bet “Jane Galt” and most of the wingnut “critics” didn’t even read the study. I’m willing to bet that the few wingnuts who did read it simply don’t understand most of (or any) of the statistical concepts. Yet this doesn’t stop these people from declaring the results “false.” Since when is such willful stupidity acceptable?

Over on Jane Galt’s blog, one of her commenters suggests the peer reviewers should be shot. I bet that person didn’t read the paper either. Just a hunch.

As against the usual rarified discourse there?

Soubzriquet, now I remember why I haven’t been there for a year.

Hmmm. Numbers seem suspiciously similar to the last Lancet pre-election volley. Which was debunked by the UN by a factor of 4.

One suggestion – (1) Compare the number of people killed in a single car bomb to the number maimed or injured to a lesser degree. (In typical combat it’s about ten to one.) (2) Do a random sample of people in the areas that are hardest hit, and find out what percentage have minor injuries and what percentage have none. (3) Compare the numbers. If the numbers show about a fifth (or less) of the injuries that would be projected by the number of alleged deaths, then look for the biases that are increasing your supposed death rate.

If you can’t show that your numbers are reasonable in, say, Baghdad or Basra, then they couldn’t possibly be reasonable in Iraq as a whole.

re comment 11:

steve,

Daniel is right and Juan Cole is wrong.

How do I know this? Juan is explaining something that doesn’t need explaining, and his explanation directly contradicts the study he is defending. 90+% of those asked could produce death certificates, so there is no need to explain why they didn’t have death certificates (which is what Juan goes on to do). Daniel instead explains why those death certificates don’t reliably make it into the central list. No one has offered any evidence to counter Daniel’s claim that the official system of data collection is broken (and multiple people have offered support for his claim – 0 deaths in Anbar province?).

That bugged me about Juan’s post when I first read it. If I didn’t hate blogger’s commenting system I would have pointed this out there.

Re #29
One thing I have been wondering about in trying to picture how the survey was conducted, is the number of households surveyed per day.

Hm. I wonder, what’s the norm in this sort of work?

Do you have a link to this information? I have the methodological appendicies, but it doesn’t seem to have this material.

I’ve been in market and social research for 25 years and I find the figures quoted above difficult to believe for door-to-door fieldwork, particularly given the desperate conditions reported in the report’s appendix.

Yelling loudly that we can’t possibly have had ~750 d/day in Iraq and not seen it in the news isn’t evidence. That’s a bald faced assertion. Coming from the average north american keyboard commando who can just about find Iraq on a map in a pinch it also looks pretty stupid.

soubzriquet, these exchanges only work if each party deals with the things the other has actually said. You appear to have me confused with someone else.

FWIW, there are plenty of discussions about the methodology taking place all over the internet, involving people far more qualified to comment than I am. I could cut and paste their commentary and pass it off as my own if that would make you feel better? Also FWIW, one of the main criticisms I’m hearing is not that the study definitely *is* flawed, rather that its potential to be flawed (systemic sample bias, for example) is not sufficiently acknowledged by the authors. Yes, caveats are given, but in the main, the authors convey an air of surety that the nature of the methodology in the circumstances the study was undertaken should disallow: a little too much grandstanding and not enough humility.

Additionally, if you can go to YouTube and watch a video of Horton venting his spleen at an anti-war rally, suspicions are going to be raised, justifiably or not. Horton has every right to do this and more, but if you are seriously looking to avoid the controversy that surrounds your work, it’s probably not a good idea. Grist to the mill, and all that.

I believe there is at least *something* to be said for observable facts and I think you are a little too quick to discount some of the reasonable concerns being voiced e.g. death certificates. We’re told that it’s too dangerous to go to hospital to count death certificates, but it’s not too dangerous that 80% of Iraqis are going to hospitals to get deaths certified. If Hopkins is right, 658 deaths are being certified in Iraq each day, yet previous estimates of the dead don’t get anywhere near that figure. Yes, I’ve read the explanations on this thread as to why that may be the case, but I’m still not convinced. Whether that’s because I’m not prepared to be convinced or because the explanations themselves – for which no facts are being presented so far as I can tell – are less than convincing, is something I’m sure we can argue about.

Your position appears to be that unless you try to unpick the Hopkins methodology and/or embark on your own study, you have little or no right to comment. This is palpably absurd. Unqualified, statistically illiterate dissenters criticizing sampling techniques is one thing, raising questions that the Hopkins figures either prompt or do nothing to address, is quite another.

Dan, can you clear up one point for me, perhaps? The link to the study I’ve been using is:

http://i.a.cnn.net/cnn/2006/images/10/11/human.cost.of.war.pdf

There is only one mention of certificates:

“We are certain that households did not report deaths which did not occur, as 92% of households had death certificates for deaths they reported.”

Firstly, my understanding is that the certificate question was asked only 87% of the time and 92% of the respondents claimed to have death certificates. What is not clear is whether the certificates were physically presented. Or if it is clear, I’ve completely missed that reference. Do you know?

That’s not the only reference in that link.

There’s:

At the conclusion of the interview in a household where a death was reported, the interviewers were to ask for a copy of the death certificate. In 92% of instances when this was asked, a death certificate was present.

There’s the proper report here as well – somehow that link gets around the Lancet’s registration

http://www.thelancet.com/webfiles/images/journals/lancet/s0140673606694919.pdf

The death toll seems not all that implausible considering the enormous number of bullets the US military has fired in the last few years, perhaps a billion per year more than the annual rate in 2000.

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

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

The overall point, however, is that nobody else appears to be doing this kind of study because it is so hideously dangerous, which ought to tell us something.

More analysis is necessary, but, after a few hours of kicking the tires, these numbers don’t strike me as obviously implausible. I wouldn’t put tremendous confidence in them either, though, due to the savage conditions under which this heroic effort was carried out.

You’d want to only go where you knew you’d be safe. Then you’d tell the Americans you did exactly what they told you to do.

The thing is, if this is true, it would, counter-intuitively, almost certainly lead to an overcount in deaths.

The surveyors were doctors from Baghdad, who had published papers under the previous regime, and so by the standards of modern day Iraq probably count as Baa’thist. By that I am not saying anything about their politics, or accusing them of bias, just pointing out that people like them have been explicitly targetted for ‘revenge’ killings, so they are not disinterested neutrals.

Consequently, they would be least safe in Kurdistan or the south, the areas other sources indicate to be the safest now, and most violent pre-war.

This is backed up by the fact the survey reports no violent deaths in 2002, when press reports show a civil war was ongoing in the north, and that the northernmost and southermost provinces in the country were explicitly not sampled.

The real relevance of this is not to score points as to who is right or who is wrong, but to assess the current situation in Iraq, and what is to be done going forward.

In particular, if the grim picture painted by the survey is right, then the ugly but only alternative would be to keep troops in the country for the indefinite future to keep a lid on the violence until it burns out.

In contrast, if other reports are more accurate, then a withdrawl is possible, according to this kind of plan.

Please don’t let annoyance with some statistically illiterate Bushoid be too much of the determining factor on what you think should actually be done between now and the next US election.

Because after the election will be too late.

There is no way to portray the deaths of half a million people in a politically neutral way.

I think there is. Moreoever, if you are the editor of a medical journal that you know will shortly be publishing a study like this, appearing at anti-war rallies in the months before venting your spleen at Bush and Blair counts as unecessary politicization.

Again, Horton’s totally within his rights to do and say whatever he wants, but I don’t buy the notion that without the political baggage, the prospects of a more considered response to the study findings would be improved. It strikes me this ought to be entirely uncontroversial.

bi,

You do realise you sound like a three-year-old, don’t you?

Arguments that it ‘doesn’t really matter’ because one opinion poll had a badly worded question that seemed to imply that a small majority of Iraqis thought it was ‘worth it’.

1 – I certainly didn’t claim the study “doesn’t matter” because of the response to the poll questions and I don’t recall having read such a comment from any other contributor. That’s just a flat-out misrepresentation.

2 – It’s more than one opinion poll that shows majority Iraqi support for the war. Try every opinion poll where that question or similar has been asked.

3 – “Badly worded” is entirely subjective. “Not worded the way I would have preferred” is not the same as “badly worded”.

4 – “Seemed to imply” is unnecessarily vague. The consistently positive response to the “worth it” question “clearly demonstrates” rather than “seems to imply”.

5 – “Small majority” is understating things when the smallest majority I have seen in one of these polls is a 61/39 split. “Comfortable majority” is nearer the mark.

That’s 4-and-a-half errors in one sentence. Is this a record?

brownie, don’t worry, you still hold the record for errors. This is due more to volume than to skill, but that’s your strong point. As to ‘what’s next’, I had pointed to posts which rebutted your assertions.

Brendan, I’d add one more point to the deniers:

Having washed the memory of the Rwadan genocide:
“…was the massacre of an estimated 800,000 to 1,071,000 ethnic Tutsis and moderate Hutus in Rwanda, mostly carried out by two extremist Hutu militia groups, the Interahamwe and the Impuzamugambi, during a period of 100 days from April 6th through mid-July 1994.” (wikipedia,
http://en.wikipedia.org/wiki/Rwandan_genocide)

That 80,000 or more people per day, with machetes and (presumably) some small arms.

Duh, 8,000 people per day. Still adds up to far more people than in Iraq, in a far shorter time.

It was: ‘do you think it was a good thing to get rid of Saddam Hussein?’

This is a simple flat lie, a straightforward and unambiguous contradiction of information directly available on this thread.

You know that, everyone reading this and paying attention knows that. Why do you feel your position is so weak you can’t defend it in any other way?

Well it may well be a lie but it’s hardly a simple lie. The actual question asked was: ‘Thinking about any hardships you might have suffered since the US- Britain invasion, do you personally think that ousting Saddam Hussein was worth it or not?’

As I argued in the previous post, I argued that this is really a question about whether the Iraqis were pleased that Saddam Hussein was deposed or not. And the answer is of course ‘yes’ as any sane person could have predicted.

This is really rather different from Brownie’s interpretation that a majority of Iraqi’s ‘supported the war’.

[As I pointed out in the other post, this is like asking Hungarians or Czechs in 1949, “Thinking about the hardships you have suffered since the liberation of your country by the Russians, do you personally think that forcing the Germans out of your country was worth it or not?” Of course the asnwer would be ‘yes it was worth it’ and of course this would spectacularly miss the point].

In any case, this has, yet again, nothing to do with the Lancet study.

Daniel writes:

you need the death certificate to get the body buried and to claim on life insurance, get bank accounts closed, establish probate, all the grisly business. (I suspect that overworked Iraqi undertakers might be stretching a point these days, but life assurers, never).

Again, do you have a citation for these claims? Something online would be great but if this is just info told to you by, say, a reporter recently returned from Iraq, that would be fine.

Although I am not an expert in Iraqi financial markets, I was not under the impression that the life insurance market in, say, Falluja, was thriving. How many Iraqis have life insurance, for example?

Again, my point is that the more that we understand about the actual mechanics of death certificates, the better we can evaluate the Lancet study. Daniel is nice to explain things to us, but not so nice to not make clear his sources.

“The surveyors were doctors from Baghdad, who had published papers under the previous regime, and so by the standards of modern day Iraq probably count as Baa’thist. By that I am not saying anything about their politics, or accusing them of bias, just pointing out that people like them have been explicitly targetted for ‘revenge’ killings, so they are not disinterested neutrals.”

“I think that the attempt to smear the Iraqi doctors as Sunnis and Ba’athists who obviously made the numbers up is just disgusting.”

“I’m seeing rather a low degree of correlation betwen that and what I actually said. Care to reread it?”

Soru, what point are you making? If the survey is unaffected by the stance of the surveyors (as you seem to be saying in your reply above despite making it clear they are not neutrals) I cannot grasp what the purpose of the original paragraph is…could you clarify your point?

John M.,

The survey could be biased by the ethnic and religious affiliations of the interviewers without the interviewers being dishonest. This has been documented a hundred times.

For example, respondents could have reported what they thought the interviewers wanted to hear. Or the surveyors could have gotten things wrong because they didn’t speak the same language as the respondents. And so on.

For all the war apologists in this thread… do you ever get tired of being always wrong about everything? Or is this just the case that acknowledging reality at this point will be too psychically painful, and so denying reality is the only path to follow?

Seriously, go back to the arguments you were making back in 2002, 2003, 2004, and 2005. Notice any patterns?

I agree with Jack. So-called “scientists” are nothing but cogs in the apparatus of ideological oppression, who construct culture in accordance with their bourgeois “values”. This survey is yet more establishment propaganda designed to minimise the extent of capitalism’s crimes. The estimate is just obviously far too low. And I have never seen the data. Prove me wrong, please.

Didn’t we go though all of this last time over the Fallujah ommissions where they deliberately excluded the worst figures to keep the numbers down?

There is a a thing about that you may not be aware of.

According to calculations available on Tim Lambert’s blog, adding in the Fallujah data, while it increases the median prediction, also increases the variance sufficiently to make the 95% CI span zero.

Lambert dismisses that as an artefact, but I think it is somewhat interesting.

The estimate is just obviously far too low.

It wouldn’t surprise me if it indeed was far too low, simply because they must’ve known they would be accused of overcounting. Certainly they had to protect themselves by being overly conservative every step of the way.

John Emerson — you’re right that it may not make much moral difference whether the number dead is 100,000 or 600,000. But at the same time, notice how your fellow lefties are suddenly overwhelmingly convinced that the number has to be 600,000; that the study is unimpeachable and no criticisms are correct; and that they must — must — defend the study to the death. Why are they so emotionally attached to the study, if it makes no difference? (Answer: It’s not because they’re non-partisan scholars interested solely in the truth; it’s because they’re anti-war and they’re eager to have anything that will help their side of the argument.)

Overall: Why does everyone seem to believe that a gut-check is somehow irrational? I.e., “this study is an order of magnitude greater than every other statistical source; maybe their extrapolation from a few hundred confirmed deaths isn’t actually accurate.”

What’s wrong with that? That’s the way we all reason when calculations turn out to have results that seem improbable. You’re balancing the checkbook, and you get to the end, and you have 10 times as much money in the account as you had imagined. Do you throw up your hands and say, “Whoo-hoo! That’s what the math says, and math is never wrong! I’m having a party!” Or do you recheck your work?

“The survey could be biased by the ethnic and religious affiliations of the interviewers without the interviewers being dishonest. This has been documented a hundred times.”

Ragout, thanks for answering on Soru’s behalf, but are you asserting that this happened in this case? i.e. that the respondents deliberately lied because of who was asking the questions? If Soru wanted to make this point, why not just say so or is cryptic inference a more useful form of argument for some reason? Presumably on this basis all the polls discussed above telling us how delighted the Iraqis are with democracy will have suffered the same problem?

Is it really necessary to point out that asserting that something could have happened rather than offering evidence that it did happen does not actually undermine the survey? Of those hundred documented instances you refer to, what was the effect on said surveys? Were all the results wrong? Just some? What % did it effect? Were any of them similar studies, employing similar methodologies in similar circumstances? The determination of the survey’s critics to find fault – any fault – is leading to some really desperate ‘arguments’.

To that end, and in order to help, I would like to point out the entire study is discredited because I first heard about it on a Thursday.

If Soru wanted to make this point, why not just say so or is cryptic inference a more useful form of argument for some reason?

Sorry, I thought it was obvious based on the content of the point I was responding to, which said that the surveyors, as admitted in the report, would tend to avoid the most dangerous areas.

The same thing applies to reporters – the areas that are most dangerous to a western journalist are not necessarily those with the highest mortality rates.

“The same thing applies to reporters – the areas that are most dangerous to a western journalist are not necessarily those with the highest mortality rates.”

…except for the western journalist mortality rate, obviously. It’s an interesting approach you’re taking Soru but ultimately it is an opinion based one rather than statisically generated based on actual sampling.

Heckuva job, Brownie!

I appreciate Daniel providing a citation. However, that citation does not seem (to me) to support the claims that he is making. That article reports:

Iraqi authorities say morgue counts are more accurate than is generally thought. Iraqis prefer to bury their dead immediately, and hurry bodies of loved ones to plots near mosques or, in the case of Shiites, in sacred burial sites. Even so, they have strong incentives to register the death with a central morgue or hospital in order to obtain a death certificate, required at highway checkpoints, by cemetery workers, and for government pensions. Death certificates are counted in the statistics kept by morgues around the country.

Great. But that the issues that (reasonable) people have with the death certificate question. We all agree that death certificates are issued, that a family is better off with one than without one. Great. But I have my doubts about a) if the government is/was actually issuing certificates in the most violent parts of Iraq during large parts of the study and b) if death certificates get you benefits, are there any false ones is existence.

Recall that Daniel made a sweeping claim:

When someone dies, you get a death certificate from the hospital, morgue or coroner, in your hand. This bit of the death infrastructure is still working in Iraq.

I do not think that this claim was true in, say, Falluja, for 2004-2005. I suspect that it is not true for many other places.

This does not mean that the study is wrong or that people aren’t dead. I just don’t think that Daniel should be implying that death certificates are not an issue if the only information he has to go on is news reports which don’t address the issue.

Let me be polite (try, try….).

Without a death certificate you can’t transfer property like houses, farm land and any number of a hundred things. Without a death certificate your loved one’s obligations don’t die with him.

You will discover this when you have to get a hundred copies of a death certificate for someone near and dear to you. It is not one of life’s pleasures, but it is dealth’s necessities. Well, maybe not in Somalia.

I certainly hope these Lancet figures count if the Dems win and they do investigate the conduct of Bush’s dirty war.

Don’t get your hopes up. They’re a bunch of spineless eunuchs. And all too many of them were complicit in the decision to go to war in the first place.

Steve, while you are probably right that the Dems will act like they have acted, I’m not counting on their inherent morality, but on two things: the mounting toll of this insane war, and two, their self interest. Eventually, some politician is going to figure out that all the preliminaries to an anti-war politics are in place, and decide to enter that position.

Since the Dem leadership lives in D.C., I think they are insulated from the true emotional disgust with the war elsewhere because, frankly, the war is one of the best things ever to happen to D.C. The reason it has the hottest real estate market and the most booming economy east of Las Vegas is because the Bush administration has used the bogus war on terror to pour tens of billions of dollars into the local economy – via a plethora of war industry firms and all their hangers on. Dem honchos sit among all this wealth, have benefitted from it themselves, and naturally are hesitant about turning off that faucet. Thus, the triumph of Lieberman-ism. My hope is simply that the war is in the final bubble phase — the money has to run out sometime, and then D.C. will be left, overbuilt, with hundreds of disgruntled, laid off think tankers looking for another war to promote. Which will mean taking down this one.

Why does everyone seem to believe that a gut-check is somehow irrational? I.e., “this study is an order of magnitude greater than every other statistical source…

It’s a fair point, but these other sources are not statistical and they only track the number of reported casualty-style deaths, like when someone gets hit by a bullet and dies within hours (at most) after the incident.

There are no news reports that would say something like: “…remember a month ago we reported an incident where 10 people were killed and 40 wounded? Well, 25 of those wounded are dead now.”

And this paricular study is concerned with increase in the death rate, which is a completely different concept.

I thought I’d come back and make my own favorite point about all this–none of the prowar critics I’ve seen or heard have as yet called for an independent investigation to determine how many Iraqis have died. The critics are obviously very upset over this 600,000 dead allegation and wish to discredit it, but none of them are willing to call for a large scale statistical survey conducted by independent researchers with US government support (and if necessary, bodyguards) to determine the truth.

Two of these critics include George Bush and General Casey. Casey couldn’t even remember the source of the 50,000 dead figure that he cited when asked about the Lancet study in a press conference, which tells you how much he supposedly cares about the issue. Which is odd–one would think that a country intending to win over hearts and minds would be very interested in knowing how many are dying. Government bureaucrats love metrics and what metric could be more important?

I wouldn’t really want to give “gut feelings” a pass on this. When these people read “A Brief History of Time” did they throw it down in disgust at page one, because the figure given for the size of the universe is just too freakin’ big! Actually, I have a suspicion that Conservatives do essentially think like that, which is one reason why they are overrepresented among The Stupid.

Donald Johnson: I thought I’d come back and make my own favorite point about all this—none of the prowar critics I’ve seen or heard have as yet called for an independent investigation to determine how many Iraqis have died. The critics are obviously very upset over this 600,000 dead allegation and wish to discredit it, but none of them are willing to call for a large scale statistical survey conducted by independent researchers with US government support (and if necessary, bodyguards) to determine the truth.

I’ll happily call for it. Not sure anyone’d particularly listen to me, but anyway. I think it would be in the British government’s interests to obtain a more accurate figure (and probably indirectly in my interests too, since as a British citizen, I am a possible target of reprisal attacks motivated by such high figures).

David Kane,

Tim Lambert has posted a lengthy excerpt from a blog-post by Zeyad (an Iraqi dentist) where he says that death certificates are issued and officials have them. The problem is that “the Ministry of Health and the Baghdad Medico-legal Institute (Baghdad’s main mortuary) is under the control of Sadrists, who have prohibited access to medical records and morgue counts by the press, and who have an interest in manipulating numbers for their own political agendas….”

This differs from dsquared’s take on the situation but the upshot is much the same: the families get certificates but no reliable aggregate figures are available.

The entire post is worth reading, but Zeyad’s blog is a bit slow to load at present.

Kevin,

Thank you for the pointer. What about this part?

The Ministry of Health should have access to certificates issued throughout the country over the last 3 years. And both the Defense and Interior ministries have their own counts. Now why isn’t any independent body looking into that information?

This directly contradicts Daniel’s claim above (that everyone gets a death certificate but that the central government can’t access them).

Then the person who issued the death certificate is meant to send a copy to the central government records office where they collate them, tabulate them and collect the overall mortality statistics. This bit of the death infrastructure is not still working in Iraq.

Now, I don’t know if Daniel is right or Zeyad or both or neither. But I find it suspect that Daniel, on the basis of Western news reports, can claim to settle the issue of death certificates, which is why I asked him for a citation. So, I think that these questions are still open:

1) Has most (75%?, 95%?) every family with a death in Iraq received an official death certificate?

2) Does someone in the central government have access to these?

3) Does having just a certificate get you benefits that you would not get without one? If so, is there any/many fakes?

Again, for now I just want to understand better the reality in Iraq over the last 3 years with regards to death certificates. Once we understand those facts, we can connect them (or not) to the Lancet study. I think Daniel is guilty of claiming to understand these facts much more than he actually does.

it’s true that the bodycount in Vietnam turned out to be an incentive to commit mass murder

It might be worth remembering that incentives also apply to ‘the handful of foreign jihadists’.

It might be worth remembering that incentives also apply to ‘the handful of foreign jihadists’.

It’s not a handful, Soru & Donald J. There are actually ‘huge numbers’ of them – didn’t you read post #97?

So what’s soru’s and Lopakhin’s point? Yes, lots and lots of Iraqis died after “we” invaded Iraq, but a lot of the deaths are caused by jihadists anyway, so it’s not really “our” fault that “we” failed to make Iraq safe?

As Edroso said: “let the jaw-dropping commence!”

Daniel and Kevin:

I’d be interested in your thoughts on the apparent huge variance in the violent versus non-violent composition of the excess death toll from the first Lancet study to the second. The authors have made several statements that the 2006 study vindicates the 2004 survey, because the excess death figures are similar. The bottom line numbers are close, but they’re polar opposites in terms of composition.

In the interest of saving time, I’ve taken the liberty of cutting and pasting my argument from this morning at Tim Lambert’s blog. Hope that’s okay:

Tim, D Squared:

Hi, it’s been a while. Hope you’re both doing well and enjoying life. Since my main man Heiko hasn’t weighed in on this by now, I guess I’ll have to sub in.

Sortition beat me to it, but didn’t follow through. Everyone seems to be taking at face value the repeated assertions of the study authors that the 2006 survey corroborates the findings of the 2004 survey. To quote from the companion document to the 2006 study:

” That these two surveys were carried out in different locations and two years apart from each other yet yielded results that were very similar to each other, is strong validation of both surveys.”

The problem is, the two surveys didn’t yield ” very similar results.”

The authors are basing this supposed mutual corroboration entirely on the similar excess death figures derived from the studies, for the time frame covered by the 2004 survey (100,000 excess deaths in the 2004 survey and 112,000 from the 2006 survey).

However, the composition of the excess death tolls differs radically from study to study. As I argued here in 2005 when comparing Lancet 1 to the UNDP survey, deaths from various causes are not interchangeable when using one study’s bottom line to bolster the bottom line of another. In other words, one can’t simply say ” we’ll make up for a shortfall in coalition air strike deaths in one study with some heart attack deaths from the other.”

Recall that in the 2004 survey, the headline-grabbing 100,000 excess death figure owed much of its punch to the 40,000-plus excess deaths attributed to non-violent causes. The 2006 survey tells an exceedingly different tale of mortality in Iraq during the first 18 months of the war. Not only is all of the excess death toll in the 2006 survey the result of violence, it’s actually greater than the entire 112,000 excess death figure, because the death rate from non-violent causes is significantly less than the base line non-violent death rate. My rough math indicates an extrapolated violent death toll of more than 130,000 for the second survey, for the same time frame covered by Lancet 1.

As I recall from the 2004 discussions and debates, the study authors and their supporters here and elsewhere attributed the large non-violent excess death toll to the inevitable effect on infrastructure, health care, etc arising from the chaos of invasion and subsequent insurgency.

That sounded reasonable enough at the time, assuming one accepted the premise that the invasion and occupation had severely degraded and impaired Iraq’s domestic infrastructure. But now, in a complete contradiction of the defence of their earlier work, the study authors are telling us that said chaos of war had no effect on the overall post-invasion non-violent death rate until the beginning of 2006, and in fact the post-invasion non-violent mortality rate was actually lower than the pre-invasion rate for more than two years after regime change.

Obviously, there are two serious problems arising from the comparison of the data sets for the two studies. First, unless I’ve completely botched the math, the 2006 survey extrapolates approximately 75,000 more violent deaths and 63,000 fewer non-violent deaths than the first survey. This is a monstrous discrepancy, when considering the base line annual mortality figure is approximately 120,000 deaths.

Second, if there is a correlation between extreme violence in a society and an increase in the non-violent death rate, why does one study confirm the correlation, while the second rebuts it?

Posted by: Mike H | October 13, 2006 09:15 AM

Jack:

What approximate date would you give for the start of the ” civil war?”

I don’t think you understand my argument. The second Lancet study overlaps the first for the time period from March 2003 to September 2004. The studies reveal excess death figures of 112,000 and 98,000 respectively, for that time period. Because the non-violent death rate in the 2006 study is actually lower than the pre-invasion rate, the excess death toll for that study is entirely made up of violent deaths, and in fact is quite a bit higher than the 112,000 figure (somewhere in excess of 130,000).

The first study measured 57,000 deaths from violence, and more than 40,000 from non-violent causes. The second study removes all of these non-violent deaths, then subtracts more than 20,000 more non-violent deaths from the pre-invasion expected death toll over the 18 month time frame, and replaces all of them with violent deaths.

That’s some anomaly.

Recall that in the 2004 survey, the headline-grabbing 100,000 excess death figure owed much of its punch to the 40,000-plus excess deaths attributed to non-violent causes.

I don’t see that anywhere in the first study. In fact, it says,

“Violence accounted for most of the excess deaths and air strikes from coalition forces accounted for most violent deaths.”

According to the first study’s figures, the non-violent death rate only went up about 20% from the pre-invasion to the post-invasion period, while the violent death rate went up by 5800%; violence accounted for 2% of reported deaths before and 51% of reported deaths afterwards. So only about 15% of the excess reported deaths were non-violent, which agrees with the authors’ statement above.

The only place I could find “40,000 excess deaths” mentioned after a few minutes of googling was here, but they don’t give a specific quote. Perhaps that was computed by someone after discarding the Falluja cluster?

And while the second study does show a decrease in the non-violent mortality rate, the authors don’t consider it a significant one–in fact, the error bars on that number permit a slight increase on the order of what was claimed in the first study.

I don’t see any glaring contradiction here between the two studies.

Upon further number-dabbling, I think the issue here is whether the first study’s Falluja data is included or not. Without, you do get about 40% non-violent excess deaths. But the first study only ignored that data when estimating total mortality increase, to make sure that estimate was conservative. The rest of the time they took Falluja into account, so they never make the 40% non-violent claim.

Anton, I don’t think that the first study reported that 40% of the excess deaths were nonviolent, but it can (almost) be inferred from the study. (Actually, I recall the nonviolent excess deaths figure as being 2/3 of deaths, not 40%). Tim Lambert obtained some further unpublished figures directly from Roberts that nailed down this conclusion.

Tim says “Leaving the rest of the population with a crude death rate of 2.1%.

Less than half that of any country in the world.
Ever.”

Saudi Arabia’s death rate in 2006 was 2.58 according to the CIA, its birth rate was around 30, and its infant mortality 12.

I agree with Dan Simon that, as a population-reduction measure, we should continue to liberate the world one nation at a time.

Any other random mud that anyone want to throw? It might stick!

Tim, post 155. Please respond, don’t just continue your original assertion, when some facts are brought to your attention.

Daniel’s explanation might be right–I’m not sure.

But Mike isn’t giving the best argument for his case, IMO. The best argument is that the UNDP/ICLS survey found 24,000 violent deaths (19-28,000 is the CI) in the first 13 months and if I’m doing the approximate calculation correctly, the latest Lancet paper gives about 90,000. That’s based on one of the tables showing 45 violent deaths in the March 2003-April 2004 period and assuming each represents 2000 deaths. The 90,000 and 24,000 numbers aren’t for the same thing, because the 24,000 excludes criminal murders. But does someone want to argue that this explains the difference?

I doubt that it does. The first Lancet paper had criminal murders as 1/3 of the total violent death toll (excluding Fallujah). So if you tack on 12,000 deaths to the 24,000, you’d have 90,000 vs. 36,000 for the first 13 months.

Not that this disproves the paper, but it doesn’t do much to inspire confidence. Somebody’s statistical survey was affected by bias. Maybe the people in Iraq lied in the earlier survey if they had an Iraqi soldier or guerilla death in the family, as Daniel suggests. I think we’re better off just taking the IBC number and saying the true death toll is probably several times higher.

And then, of course, absolutely demanding that the US and British governments sponsor an independent investigation to determine the true death toll (with particular focus on deaths caused by the US and Britain–Iraq Body Count statistics on those are absurdly low for most months. The coalition killed 370 people in the third year of the occupation, if one goes by what the press has reported.)

The problem with the comparison to Russia is that a lot of people think the 1990’s excess mortality figure for that country had a lot to do with the economic policies pushed by the West. So maybe your comparison doesn’t prove what you want it to prove.

Tim: “Barry ,
post 158”

That’s been covered already; it isn’t out of line with neighboring countries.

bi:

You’re confusing the folks here, since you’re linking a comment from another blog, minus the link. My response to Robert at Deltoid is up, if you’d like to have a look.

“Mike H: The simplest explanation, and the one I in fact believe, is that in the 2004 study, survey respondents lied about the deaths of young men who had died while fighting in the insurgency and said that they had died of natural causes, but in the 2006 study they told the truth.”

I have some problems with that, Daniel. That’s a mighty big assumption you’re making. First off, it seems implausible to me that so many would lie, enough to account for all or most of a 75,000 increase in violent deaths over the 18 months.

Second, you’re now adopting an astronomically high KIA figure for insurgents in the first 18 months, and one that would probably roughly extrapolate to several hundred thousand insurgent deaths for the entire 40 months. I just can’t see that being accurate.

By the way, here’s Tim’s explanation for the variance:

“Mike H, given the wide CI for the first Lancet study, the close agreement (98k vs 110k) with the second study is surprisingly good. I’m expecting someone to seize on this and accuse them of cooking their statistics any time now. When you break it down into violent and non-violent deaths things get more uncertain. I’m pretty sure that if you did the test you’ll find no statistically significant differences in the violent and non-violent rates between the two studies.

There’s a 50% chance that the differences will have opposite signs, and partly cancel, which is what happened

Posted by: Tim Lambert | October 13, 2006 10:59 AM

I concede I don’t know what he means by this ” test.” I asked, but didn’t get a response. Having no statistical background, his explanation probably wouldn’t help me anyway.

Mike H: Pick up a statistical text book and read it yourself. Come back to criticize the study when you’ve finished.

Or are we obliged to allow you to yammer on things you clearly know zilch about?

Dan, you don’t, really. If Saddam had collapsed, there is no way the chaos that would have resulted would have equalled the number of Iraqis and Iranians murdered in that war. Not all collapses lead to mass death. In fact, in Northern Iraq in the 90s, there was a similar collapse, and it lead to something like 6-10,000 deaths.

What does multiply deaths is occupation. The U.S. occupation is the huge multiplier here. Compare the number of deaths from any other period in which an Iraqi leader was overthrown, and the deaths are way below what has happened since the U.S. decided to occupy Iraq.

Point, then, is simple. Your counterexample doesn’t hold water. The precedents for the social and political collapse of other regimes in Iraq point to this one being several magnitudes worse. The factor that made it worse was the U.S. Occupation. Getting the U.S. out of Iraq is not only what the Iraqis want, but it will be good for Iraq.

Nice dodge, bi.

One doesn’t have to be a statistician to seee the huge variance between the two studies, in terms of violent and non-violent deaths. I realize that makes you angry, and eager to avoid dealing with it.

Tough.

Anton Mates:

There is a huge discrepancy. You can’t include Falluja, because the 2004 authors excluded it from their results. One can’t try to reinsert it now, to try and explain away an uncomfortable variance from a new study. Even if you could, reintroducing the Falluja data makes your problem even worse in terms of study to study corroboration, because it adds about 200,000 additional violent deaths to the first study.

Dan Simon: So you’re lowering the standard of “success” from “make Iraq into a modern democracy where people can live and work safely” to “make Iraq into a country which is just better than if we allow Saddam’s regime to crumble by itself”.

I’m not “lowering the standard”–I never had such high expectations in the first place.

Again, I’ll ask: is the former Soviet Union better off–despite all the death, violence, chaos, economic decline, and resurgent authoritarianism–for being “former”? And if so, why isn’t the same true of the former Saddam-ruled Iraq?

If Saddam had collapsed, there is no way the chaos that would have resulted would have equalled the number of Iraqis and Iranians murdered in that war.

Quite possibly correct. But I’ve already noted that the US’ early support for Saddam Hussein was motivated by cynical self-interest, not humanitarian concern for saving lives.

What does multiply deaths is occupation. The U.S. occupation is the huge multiplier here.

Utter nonsense. Neither the Soviet Union nor Yugoslavia were occupied by anyone. They succeeded in generating huge numbers of excess deaths through economic collapse and civil strife all by themselves. If anything, the US occupation, by tamping down (however partially) the worst violence and assisting (however haltingly) in the reconstruction of a viable central government, has mitigated the horror.

Compare the number of deaths from any other period in which an Iraqi leader was overthrown, and the deaths are way below what has happened since the U.S. decided to occupy Iraq.

Apples and oranges. Previous regimes weren’t nearly as repressively totalitarian as Saddam’s, and therefore could be overthrown without the government disintegrating completely. The collapse of Saddam’s Ba’ath party, on the other hand, was bound to be more like the collapse of the Soviet or Titoist regimes–an internal breakdown leaving no civil society from which a successor could emerge to fill the vaccuum.

There is a huge discrepancy. You can’t include Falluja, because the 2004 authors excluded it from their results. One can’t try to reinsert it now, to try and explain away an uncomfortable variance from a new study.

Sure one can, because the authors didn’t exclude any clusters in their new study. If one study has an apparent outlier which it excludes, and the other one does not, there’s no “discrepancy” in their getting different results.

Even if you could, reintroducing the Falluja data makes your problem even worse in terms of study to study corroboration, because it adds about 200,000 additional violent deaths to the first study.

No, it doesn’t. It raises the excess violent death count to about 200,000, which is an increase of about 100,000 over the Falluja-excluded data. And it makes the error bars on that count so wide that the second study’s result doesn’t conflict with it at all.

Mike H,

I started reading your comments last night both here and at Deltoid. I’ve watched carefully today to see how the discussion has progressed on both blgos – just in case you had got something spectacularly wrong.

Well, it seems your logic is sound, so much so that I’ve cited you on HP (http://hurryupharry.bloghouse.net).

Nice work, fella.

Dan, well, you obviously have a different idea of fruits than I do. The obvious ‘collapse’ of the Baathist regime, in your world, has nothing to do with the disbanding of the army by the Americans, as well as their general disbanding of the government, concern with creating a state economically like Pinochet’s Chile while the ‘deadenders’ flared, and in general the concrete historical occurences of 2003-2005. On the other hand, it is nice that you think the Baathists took power after the revolution deposing the King.

As for the odd position that Iraq is now better off without Saddam, but in 1983 would have been better off with Saddam – or is it that 1983 would have been chaos because the Baathist party would have disbanded – or is it that the chaos has nothing to do with the occupation and is because of apples not being oranges – in any case, it seems that any moral points one scores for getting rid of Saddam (even if the method was rather in the way Mafioso are gotten rid of — by their confederates) are going to slip away unless one decides, for no apparent reason, that chaos like this is to be expected. Yes, you supported the invasion thinking, well, 600 000 Iraq deaths is a small price to pay.
Pity the war supporters weren’t talking that talk in 2003.

As for the odd position that Iraq is now better off without Saddam…

You mean the “odd” position held by a majority of Iraqis, according to the polls?

Who gets to decide whether Iraqi is “better off”, Roger? The people living there, or bloggers in Six Mile Bottom?

Wow, Brownie, if you analyse statistics like you parse sentences it would explain a lot.

Now, do try and read that sentence again. I’ll give you three more chances to figure out what it means. I’m not going to ask you to bring to this task any complicated rhetorical skills — don’t worry! No metaphors there to bite you, or irony. But watch out for the treacherous conjunctions! It is a skill you will need when you are older, say, and reading contracts, or signing up for cable tv. It is a harsh world out there for people who don’t understand main clause, modifying sub clause kinds of relations. Good luck! Hint: ‘but’ is one of those conjunctions like ‘and’, except often in a sentence, when you see it followed by a subjunctive modal, it is tricky, tricky tricky!

“No, it doesn’t. It raises the excess violent death count to about 200,000, which is an increase of about 100,000 over the Falluja-excluded data. And it makes the error bars on that count so wide that the second study’s result doesn’t conflict with it at all.”

You’re wrong, Anton. From the study itself, page 4:

” In our Falluja sample, we recorded 53 deaths when only 1.4 were expected under the national pre-war rate. This indicates a point estimate of about 200,000 excess deaths in the 3% of Iraq represented by this cluster.

D’oh! You’re right about the 200,000 thing. Nonetheless, the next sentence is, “However, the uncertainty in this value is substantial and implies additional deaths above those measured in the rest of the country.” So: error bars.

“Sure one can, because the authors didn’t exclude any clusters in their new study. If one study has an apparent outlier which it excludes, and the other one does not, there’s no “discrepancy” in their getting different results.”

Okay Anton, I give up. We’ll do it your way. That leaves us with these 3 violent death figures from the Lancet studies, for the first 18 months of the war:

2004 study, minus Falluja cluster – 57,000.

2004 study, with Falluja cluster – more than a quarter million.

2006 study – more than 130,000.

So 2004 figure ignoring Falluja data

And to make the first post a little more readable:

“No, it doesn’t. It raises the excess violent death count to about 200,000, which is an increase of about 100,000 over the Falluja-excluded data. And it makes the error bars on that count so wide that the second study’s result doesn’t conflict with it at all.”

You’re wrong, Anton. From the study itself, page 4:

” In our Falluja sample, we recorded 53 deaths when only 1.4 were expected under the national pre-war rate. This indicates a point estimate of about 200,000 excess deaths in the 3% of Iraq represented by this cluster.”

D’oh! You’re right about the 200,000 thing. Nonetheless, the next sentence is, “However, the uncertainty in this value is substantial and implies additional deaths above those measured in the rest of the country.”

Anton,

By my calculations, we have the following figures for the change in non-violent mortality rates:
First Study Change (After – Before) = 1.0 +/- 2.3
Second Study Change (1 Yr After – Before) = – 0.9 +/- 1.9
diff-in-diff of non-violent mortality = 1.9 +/- 3.0

So the first time round they said non-violent mortality went up by 1.0 per 1000, while the second time around they said it went down by 0.9 per 1000. The difference isn’t statistically significant, but it’s fairly close.

My point is that these calculations are based on the simplest possible statistics: comparing unadjusted means. Anything more complicated is going to reduce the confidence interval, and could well indicate a statistically significant difference. Examples of more complicated would be adjusting for region and age (which the Lancet studies never do) or basing the analysis on the log of cluster deaths (which is what they emphasized in the first study).

—– Details of Calculations —–

Mortality, excluding Falluja, old study
Before: 5.0 +/- 1.3, SE=0.65
After: 7.9 +/- 2.3, SE=1.15
change=2.9 +/- 2.6 SE= 1.3
= 71,000 +/- 63,000

Non-violent Mortality, excluding Falluja, old study
Before=45/(110,538/12) = 4.9 +/- 1.7
After = 69/(138,439/12) = 5.9 +/- 1.5
change=1.0 +/- 2.3
= 24,400 +/- 56,000
Note: SEs extrapolated from new study by multiplying by sqrt(47/33)

Non-violent mortality, new study
Before: 5.4 +/- 1.4
1 year After: 4.5 +/- 1.3
Change = – 0.9 +/- 1.9

diff-in-diff of non-violent mortality = 1.9 +/- 3.0

Ragout,

So the first time round they said non-violent mortality went up by 1.0 per 1000, while the second time around they said it went down by 0.9 per 1000. The difference isn’t statistically significant, but it’s fairly close.

I think the second number should probably be 0.8 rather than 0.9, since the first study’s postwar period includes about a third of the second study’s next interval (where the non-violent deathrate increased a bit, though not back up to prewar levels.) While that may still be “close to significant,” as long as it’s not significant I don’t think the authors can be charged with dishonesty for not mentioning it.

My point is that these calculations are based on the simplest possible statistics: comparing unadjusted means. Anything more complicated is going to reduce the confidence interval, and could well indicate a statistically significant difference.

Right, but if the authors didn’t choose the simplest possible statistics to compute their estimated excess deaths in either study (note that they say they used the log-linear model in the second study as well), then I don’t see this implying any incomparability between the two. I realize that in principle you can throw enough parameters into a model to cook up a significant difference no matter what the data looks like, and I’m not competent to judge whether the authors did that, but I think the paper’s reviewers probably were.

Roger,

Pitiful.

Tim, your link doesn’t work. But the 9.03 figure is true – see my blog for instance. Or go to the WHO pages and select Iraq and 2001.

Tim,
You write:
between 4 and 5 according to the WHO.
And it is a freak stat at that.
25% of the population are “guests” who generally dont stay around to die there.
You’re not arguing that Iraq had the lowest adult death rate in the world are you?

I’ll answer once more but to be honest with you I don’t think you are particularly interested in the answer. I’m not arguing that as you well know. If you look at Wikipedia’s crude death rates you’ll see the lowest proper countries are Kuwait, Saudi and Jordan. Then Brunei, Libya and Oman. Then the Gaza Strip. Then (after the Solomon Islands) the West Bank.

These are all much lower than the estimate for Iraq, but it’s hardly an odd company to find Iraq, is it? I think if Jordan can have a death rate of 2.63, Iraq can have 5.5 or whatever it was.

matthew,
the research claims a huge death toll for under ones in Iraq (10% infant mortality,3.4 deaths per thousand of the population)
but then it became the healthiest place in the world (2.1)

not credible.

Also, as mentioned several times, low crude death rate doesn’t = healthy anyway. Look at the CIA Factbook list of death rates, for example.

Or, if you prefer WHO figures from 2001–the same figures that gave a crude death rate for Iraq of 9.03–they give the UK a rate of 10.83. Does that mean Iraqis were healthier than Brits that year?

Additionally, if you can go to YouTube and watch a video of Horton venting his spleen at an anti-war rally, suspicions are going to be raised, justifiably or not. Horton has every right to do this and more, but if you are seriously looking to avoid the controversy that surrounds your work, it’s probably not a good idea. Grist to the mill, and all that.

The fact that the right wing is investing so much energy in ad hominem attacks on Horton indicates to me that they haven’t been able to find anything to attack the actual authors of the study. I suppose that possible political bias on the part of the editor might be a valid concern if the article were of flagrantly low quality and not up to scientific standards for the field of epidemiology. However, this is clearly not the case; the methodology is one that is standard and well-validated, and the authors do a good job of identifying and acknowledging factors that could potentially bias the conclusions.

Anton.
Two things here.
On infant mortality the WHO gives a figure of almost 10% for 2001.The Lancet claims that this has fallen to 2.9 percent just over a year later.
Not credible.

secondly,the WHO figure of 9.03 crude death rate if used in the research (rather than their 5.5 figure,based on analysis of a mere 82 deaths)would have changed the whole basis of the report.
They would have been left arguing that the invasion saved lives for the first two years.

Tim:

To sum up both your objections, WHO’s 2001 mortality estimates are very different from the Lancet studies’ 2002-2003 mortality estimates.

The WHO life tables are extrapolations of data from previous years–for instance, the table for 1999 relied on a variety of surveys, the latest of which was done in 1999. (See Life tables for 191 countries: Data, methods, and results for the gory details.)

So it’s possible that the WHO estimates are simply inaccurate for 2001, or at least less accurate than a post-2001 survey of that year’s mortality would have been. It’s also possible that mortality drastically improved between 2001 and early 2003.

To say the Lancet results aren’t “credible” based on conflict with WHO, you need to show that
a) WHO’s estimates are more likely to be accurate than the Lancet researchers’, even though only the latter used actual mortality data covering the time in question, and
b) mortality couldn’t drastically change in a year or two. Do you have evidence for that?

You do realise you sound like a three-year-old, don’t you? (Brownie, #78)

if you can go to YouTube and watch a video of Horton venting his spleen at an anti-war rally, suspicions are going to be raised (Brownie, #62)

Let me know when you want to debate like a grown-up. I’ll be right here. (Brownie, #206)

What exactly is “ad hominem” about my comment? (Brownie, #212)

ad hominem — an argument “against the man” or person. This is a device employed to attack not the issues but rather the one you are arguing with, especially on a personal level or basis. It is usually employed by those whose arguments are weak.

Brownie, just came back to applaud your call for an investigation into the death toll in Iraq. Yes, it surprises me. I haven’t seen too many war supporters wanting this. (Mike H wants it, though come to think of it I’m not sure he’s a war supporter now.)

“Brownie, just came back to applaud your call for an investigation into the death toll in Iraq. Yes, it surprises me. I haven’t seen too many war supporters wanting this. (Mike H wants it, though come to think of it I’m not sure he’s a war supporter now.)”

Hi Donald:

I abandoned this thread for fresher, greener debate pastures a couple days ago, but decided to pop back for a look at anything noteworthy.

You make an interesting observation about whether I’m a ” war supporter,” still. I confess I haven’t really asked myself that recently, at least not with the ” supporter ” moniker.

I’m disgusted with the way things have turned out. If I’d known that in advance, I wouldn’t have supported regime change, although I usually try to avoid the ” if I’d known in advance ” supposition because it’s essentially worthless when reevaluating one’s previous stance.

Most of my anger arises from the belief that this could have been done much better. I’m not prepared to say, knowing what we know now, that it could definitely have been done ” right,” but certainly better.

Bush and his inner circle deserve much of the blame, naturally. But the more *ahem* “partisan” of their critics have been unfair in apportioning blame. I see plenty of the more vocal Bush critics giving both the military and the Iraqi people complete passes on the blame game front. That’s ridiculous.

It’s ludicrous to absolve the American military of its share of responsibility, whether by inept high ranking commanders in the Pentagon and in the field, or the small percentage of psycopathic grunts on the ground who committed war crimes.

It’s equally ludicrous to allow the Iraqi people to emerge blameless. Given a chance at a fresh start, freed from a brutal dictator by the only means that was likely to accomplish that (foreign invasion), they’ve squandered it in a frenzy of sectarian, tribal hatred.

Sorry for digressing. Back to the question of my ” support, ” or lack thereof, for the war.

I support the continuing presence of coalition troops in Iraq. I think there’s a subtle difference between that, and supporting the war. It would be better if there had been no invasion, although that comes with the caveat that we don’t know what Saddam might have done to his people had he been in power still. But there was an invasion, and we are where we are. We can’t change that.

U.S. troops are preventing a much higher death toll from occurring. It’s interesting to note (and this seems to have been overlooked by many who want the U.S. out of Iraq), that it isn’t just the Shiite dominated Iraqi government that wants U.S. troops to remain. Since the Shiites started fighting back, tit for tat, on the sectarian front, we’re seeing prominent Sunnis stating that U.S. troops need to stay, to protect their communities.

You’ve seen enough of my posts here and at Deltoid to know I believe Lancet 2’s death toll is grossly inaccurate. If the Americans were to abruptly pull out, it won’t be for long.