Of course I didn't understand it, because it was arrant nonsense. That's why the WHO team took care to say, in their paper:
The IFHS results for trends and distribution of deaths according to province are consistent with what has been reported from the scanning of press reports for civilian casualties through the Iraq Body Count project. The estimated number of deathsin the IFHS is about three times as high as that reported by the Iraq Body Count. Both sources indicate that the 2006 study by Burnham et al. considerably overestimated the number of violent deaths. For instance, to reach the 925 violent deaths per day reported by Burnham et al. for June 2005 through June 2006, as many as 87% of violent deaths would have been missed in the IFHS and more than 90% in the Iraq Body Count. This level of underreporting is highly improbable, given the internal and external consistency of the data and the much larger sample size and quality-control measures taken in the implementation of the IFHS.
Given that almost all the deaths in their study resulted from violence, this is the same as saying that they overcounted.
Moreover, if I made a mistake, apparently every conflict epidemiologist I talked to made the same mistake, because they had trouble believing that I was quoting Les Roberts correctly. I worked with his statement in front of me, and asked whether there was any possibility that these studies agreed with each other on the overall level of deaths in Iraq. That's one point were everyone I talked to was unanimous: they didn't, and also, you couldn't compare the violent deaths figure to the overall deaths estimate that Kieran Healy tried to back out of the raw data. The people I asked included Olivier Degomme, one of the very few researchers who had access to both data sets.
That does not prove that Burnham was wrong, only that the two studies do not, in any sense recognized by the other conflict epidemiologists I talked to, agree. Iraq in 2006 was a terrible place to collect data, and even with a large data set it wouldn't be totally shocking if the WHO study were off. And though I think that the WHO is the most likely to be accurate, given its much larger sample size and better supervised research teams, that certainly doesn't prove that Burnham et al. did anything wrong--only that there was something wrong with their methodology, or that they simply hit the jackpot and got two outlying results. But the attempt to salvage the Burnham study by claiming it basically found the same thing as the WHO did is deeply silly.
As I said @12:49 in part in your other recent post on this subject:
The advantage of being able to say 'what goes on' is that, having established that, if it disagrees with Bush policy, then "we," as Americans, 'don't have friends around the world' which is the equivalent of 'our behavior is not approved of by opinion leaders,' thus we should have a change of behavior (or government). The fact that the study came out shortly before the 2004 election made the 'simply hit the jackpot and got outlying results' a double improbability. The argument that Burnham's results were essentially no different etc. is simply a rear guard defense of the social rank influence sought by Lancet.
The 400,000 excess deaths is within the confidence interval for excess deaths from the Lancet study. Certainly the numbers for violent deaths are in disagreement, but I can't imagine anyone telling you that the numbers for excess deaths are statistically significantly different.
One bias question not answered is if the "poll" was biased toward the Sunni community...
A second bias is that the Iraqi doctor had worked for the Saddam Iraqi government...again suggesting a "Sunni" bias.
A third problem was that if one calculates the incredibly short amount of time spent with each family in the poll, one wonders how it was done so quickly..
Asking delicate questions in time of war requires more than a quick hello goodbye approach in Arab societies...especially since the interviewers claimed to have seen death certificates to verify every death.
McArdle is playing a political game called "defend the Iraq war." Thus she will attempt to challenge any information that underscores the horror of this atrocity. Of course, McArdle lacks the honesty to defend the war directly. She expects us to believe that she challenges the Lancet study on technical grounds, but her motives are plainly ideological: defending predators against their prey.
Presumably this is why Johns Hopkins is investigating, too, xyz. It's all those evil Bushbots. They're everywhere.
"horror of this atrocity"
Yeah, a country that had a psychopathic dictator who peddled in international terrorism, illegally harbored known murderers and terrorists, and regularly sodomized and raped his own citizens on a daily basis whilst taking the international communities sympathy money and spending it on golden palaces gets removed from power with minimal civilian casualties, far less than what said Dictator was responsible for during his tenure and idiots like you call this....an atrocity?
This right after said country, now removed from the grip of said psychopathic dictator, just had ANOTHER round of democratic elections that were not possible under the previous dictator, thus improving the chances that Iraqis may one day live in peace after all these years.....an atrocity?
Screw you man. You don't understand what those words mean if this is how you use them.
Speaking of political games why are so many people suddenly defending the lancet study, which for all intents and purposes is recognized as worthless. Lancet has distanced themselves from the study and Burnham's employer has reprimanded him for not disclosing his methods(making the study impossible to replicate and scientifically impossible for use even as an outlier).
I'm calling BS on this. Megan initially supported the war. She then, very publicly on her blog, changed her mind and said she had been wrong to support the war. She has no reason to defend a war that she's been calling a mistake for more than two years.
Agreed, Hopkin's think an investigation [now] will restore their credibility with people like me. Add in all their anti-gun "studies", and its too little too late. Lancet = CBS Memos
What ticks me off is that I questioned their methodology as suspect when it first came out, while lefties were pounding the table with the bogus figures. But now, its kinda irrelevant if Hopkin's makes a correction. Their lies have become ingrained urban myth with moonbats [like xyz above]
There never seems to be a consequence to warn Hopkin's, Climate Change "scientists", and people like Mapes and Rather from doing it all over again. There should be.
xyz,
You are speaking of the scientific atrocity of a "researcher" just making sh*t up that was eagerly printed by Lancet in attempt to undermine a US election - right?
More telling than the conflict between the 2006 report and other people's studies is the conflict between Burnham et al 2006 study and their 2004 study.
(I did a detailed examination of the 2004 study at Chicagoboyz.)
In a nutshell, the 2004 study claimed that the majority of all deaths in Iraq were caused by air strikes. They based this claim on a single cluster in Falujia sampled outside the study's methodology. However, including that cluster pushed the actual reported death's from violence north of 250,000 with (IIRC) 180,000 deaths from air strikes in Anbar province alone.
The 2006 study found no such air strike massacre and concluded that, during the same time period covered by the 2004 study, almost all deaths from violence resulted from small arms fire.
So, not only does each study disagree with all other studies on the matter, studies conducted by the same researchers using the same methodology disagreed with each other in critical detail.
No such conflict would be tolerated in any study in any other context.
I do find it really amusing that Tim Lambert, a guy who made his reputation by attack John Lott, using research information he got from John Lott, is now willing to defend "researchers" who publicly refuse to give their data to anyone who might disagree with them.
Tells you everything you need to know about Tim Lambert. And not a bit of it is good.
Every bit of ground-breaking research overturns someone else's apple cart. If you publish a scientific paper challenging other people's assumptions, you expect that people whose assumptions you're challenging will demand copies of your data, and pour over it for the purpose of proving you wrong. Real scientists hand over the data, because they know they've done nothing wrong.
The John's Hopkins political whores refuse to give their data to those who will try to prove them wrong. A sure sign that they know they have done something wrong.
Dear Megan,
I invited you before to ask some questions about why AAPOR is condemning someone who's not part of their organization about a study published 3 years ago.
Tim Lambert did, and he was blown off. Perhaps you could try your hand at it?
The confidence level for the 650,000 "excess deaths" estimate was +/- 200,000. It was extrapolated from 647 actual deaths among 128,000 or so people.
The estimate is only as good as your sampling. Too many people were reporting this figure without the error bars. That's the first problem.
This was a very rough estimate paraded as being more-or-less exact.
Iraq Bounty Count tracks every single reported death in Iraq, meaning that their error bars are much smaller, becuase they have a much larger sample size.
The second problem is counting "excess deaths"--those who would not have died had there been no invasion of Iraq--and laying all these deaths at the feet of the US, as though Baathists and Al Qaeda and Sunni-Shiite fighting had no moral responsibility.
I wish someone had doen a study of "excess deaths" resulting from not overthrowing Saddam Hussein in 1990, but no one seems interested.
I do remember the "dead baby parades" where Saddam and his amen chorus in the West laid the deaths resulting from economic sanctions at the feet of the US as well. Saddam Hussein, it seems, was never responsible for anything. The same people who opposed the Iraq War in 2002 were in general the same as those who favored lifting the sanctions in 2000, and only rediscovered the effectiveness of sanctions and inspections when Bush was threatening to invade.
And Scott Ritter, who opposed the invasion, had this to say:
http://www.time.com/time/nation/article/0,8599,351165,00.html
"The prison in question is at the General Security Services headquarters, which was inspected by my team in Jan. 1998. It appeared to be a prison for children — toddlers up to pre-adolescents — whose only crime was to be the offspring of those who have spoken out politically against the regime of Saddam Hussein. It was a horrific scene. Actually I'm not going to describe what I saw there because what I saw was so horrible that it can be used by those who would want to promote war with Iraq, and right now I'm waging peace."
Prisons for toddlers are better than war. Well, genocide in Rwanda was better than war, and so in Darfur, and in the thirties concentration camps were better than war with Germany. There are worse things than war, I think.
This is exactly the mindset I have the problem with--that any horrible thing that goes on in the world is better than American intervention. And this is the mindset of those who carried out the Lancet study.
Is there one honest scientist left in the world?
Tim Lambert:
If I understand you, since 400,000 is within the confidence interval of which 650,000 is the mean, the (p=.05?) confidence interval has a width of at least 500,000, covering numbers from at least 400,000 to at least 900,000 deaths. Are you comfortable with a confidence interval that wide, relative to the magnitude of the mean?
If the study is accurate, where do you think all the bodies went that so few would show up in the body count? I can think of no answer that is nearly as plausible as that the Lancet study's sampling was badly flawed. Can you?
DannyK being fair you should note that Megan made a point, before she got into the specifics of their charges and her own work on the story
of, saying "I don't know what to make of this. Burnham is not a member of the AAPOR, so I'm not sure why they felt they had to censure him."
The title of the post you mention is called,
"Author of Iraq body count study censured by a group he doesn't belong to" for goodness sake.
This study should be given no credence because it was mostly funded by petroleum lobbyists.... wait, scratch that- it was funded by George Soros. In that case there is no conflict of interest and the conclusions should be accepted at on faith.
The Lancet has gotten itself into such trouble with these political articles that its reputation has been harmed. It also published the bogus study that began all the hysteria about autism and vaccines. That study has now been shown to have been funded by attorneys and included a total of six children studied. The authors, except for the lead author who is associated with attorneys, have retracted the conclusions in a major scandal. I'm afraid Lancet cannot be relied upon if there is any possibility of political influence.
I am surprised that Tim Lambert brings up the idea that 400,000 is within the confidence interval of the Lancet study that arrived at the 650,000 excess deaths. Since, 400,000 is about 60% of 650,000, this ends up with a huge confidence interval. The confidence interval is so high as to cast complete doubt on the study in the first place.
Rick
The serial lies employed by the Bush administration do not cause the least annoyance to the backers of the Iraq war, but they attack ferociously any hint of inaccuracy or exaggeration in casualty statistics.
They should be reminded that the Iraq war was not marketed to the American people as a mercy mission to rescue an oppressed people. It was cooked up by the PNAC boys as a geo-strategic power grab, propelled by the public hunger for revenge that followed 9/11, and "justified" on the basis of nonexistent "threats." Now, the last excuse with a shred of credibility, humanitarian intervention, is defended tenaciously by the war apologists.
The Iraq war was not launched as a humanitarian intervention, and the vast damage it has inflicted on the people of Iraq makes a mockery of that assertion.
Lancet published their first survey on 29 October 2004, and Les Roberts insisted that it be published before the 2004 Presidential election.
The second survey was published before 11 October 2006, just before the mid-term elections.
If the surge had not worked, I would not be surprised if there had been a third survey in October 2008.
This is the kind of blind bias that gives science a bad name.
This would be the Burnham who was just censured for refusing to validate his study or release his data for a third party to validate it. That sort of censure comes about because they can't get at the data to demonstrate other misconduct, but it's misconduct in itself. At least unless you're a climate researcher or politically convenient.
Lambert's not known for his willingness to see counter-arguments once he's decided who he wants to be right.
No, Tim, because violent deaths are not related to non-violent deaths. Non-violent deaths are randomly distributed throughout a population by completely different--and roughly stable--forces. I am telling you that professionals who do this stuff for a living and had seen the data sets specifically said that the two data sets did not agree about the total number of people who had died in Iraq. People who are professionals who had seen the papers said they did not agree about the number of people who had died in Iraq. In fact, the only conflict epidemiologist that I am aware of who believes the papers agree is . . . Les Roberts. There is some overlap between violent deaths and non-violent deaths, as things like war degrade hospital infrastructure, but not enough to explain either the size or the distribution of the Burnham results. Moreover, in particulars, such as the distribution of deaths by province, Burnham et al. was again the outlier, something that the authors of the study made quite clear.
And of course, you are comparing the upper bound of an estimate the WHO researchers didn't even calculate (the number of total excess deaths), with the very lowest value of the Burnham paper. With a confidence interval this wide, that does not constitute substantive agreement.
Now, maybe you and Les Roberts are right, and everyone else I interviewed about the subject is wrong. But they were pretty convincing.
How many studies of Iraqi civilian casualties did the Bush administration publish? What better source of authoritative statistics that the military occupation that was directly producing many of the casualties?
Not only did the Bush gang not publish such data, it actively interfered with Iraqi authorities attempting to collect casualty statistics. The Orwellian propaganda machine assembled by Rove worked incessantly to keep blood and coffins off America's TV screens as part of their psyops campaign directed against the American public.
This allowed Bush partisans to believe that the invasion caused "minimal" casualties, when it produced vast human suffering as it destabilized a nation, hurling it into the abyss of ethnic cleansing and the collapse of civil order.
The Bush administration used every available trick to conceal and obfuscate the extent of damage done to Iraq, yet backers of this criminal regime clamor loudly for "accuracy" in tallying the death and destruction. What disgusting hypocrisy!
The defenders of this dishonesty are giving us a great example of modern liberalism: tell lies under the guise of "science" or a "survey", obfuscate and refuse to allow any substantive peer review of your methods or your data, then refuse to engage your critics substantively by going ad hominem. Everything about this "study" (aka political hatchet job) stunk to high heaven. Its timing, the defensiveness of its authors, its source of funding, etc. The confidence interval was absurd as well. I believe the lower bound of the 95% confidence interval was 6,000 deaths. The study told us nothing other than liberals are liars. Not just the authors, but the idiots who reported on it and used it as a supposed piece of evidence against the Bush Administration and the Iraq war. No matter how hateful and angry Soros, Lambert, Burnham, Roberts et al. are, they can't deal with the fact that Iraq is better off now than it was with Saddam in power. I remember when liberals were apoplectic about John Burns's estimate of pre-war Iraqi deaths. But that was no more of a guess than this piece of Soros funded propaganda from Bush hating liberals. Just keep defending it. None of the defenders of this nonsensical, pretend "survey" have any credibility left to lose. Their continued defense of this obvious fraud is proof of their lack of intelligence. Either liberals are dumb enough to believe this nonsense or they are dumb enough to believe that we believe it.
xyz, are you seriously proposing that the World Health Organization is a tool of the Bush administration?
Megan,
No, I am proposing that you are a tool of the toxic plutocratic residue of the Bush administration. Not content with voting for this war criminal twice, you now persist in a pathetic rear-guard action to defend his domestic economic voodoo (tax cuts and trickle down) and his foreign policy crimes (Halliburton brings you freedom and some offers you can't refuse).
For some reason, you make no reference to the recent independent study that supports the Burnham estimates:
ORB survey of Iraq War casualties
From Wikipedia, the free encyclopedia
On Friday, September 14, 2007, ORB (Opinion Research Business), an independent polling agency located in London, published estimates of the total war casualties in Iraq since the US-led invasion of Iraq in 2003.[1] At over 1.2 million deaths (1,220,580), this estimate is the highest number published so far. From the poll margin of error of +/-2.5% ORB calculated a range of 733,158 to 1,446,063 deaths. The ORB estimate was performed by a random survey of 1,720 adults aged 18+, out of which 1,499 responded, in fifteen of the eighteen governorates within Iraq, between August 12 and August 19, 2007.[2][3] In comparison, the 2006 Lancet survey suggested almost half this number (654,965 deaths) through the end of June 2006. The Lancet authors calculated a range of 392,979 to 942,636 deaths.
On 28 January 2008, ORB published an update based on additional work carried out in rural areas of Iraq. Some 600 additional interviews were undertaken and as a result of this the death estimate was revised to 1,033,000 with a given range of 946,000 to 1,120,000.[4] opinion in Iraq since 2005."[1]
xyz: No, I am proposing that you are a tool of the toxic plutocratic residue of the Bush administration.
LOL, you are too funny. I am guessing that you are one of those tolerant and inclusive people respectful of diversity.
The ORB study is not a reliable study. It's methodology has not been validated, and what sketchy details are available indicate that it's worthless.
xyz,
"the Iraq war was not marketed to the American people as a mercy mission to rescue an oppressed people. It was cooked up by the PNAC boys as a geo-strategic power grab, propelled by the public hunger for revenge that followed 9/11, and "justified" on the basis of nonexistent "threats."
Instead of buying your Oxford-fellating rhetoric, one can simply read the Iraq War Resolution to see exactly what was sold to the American people, and then unanimously approved by the legislature after deliberation.
http://en.wikipedia.org/wiki/Iraq_Resolution
"Now, the last excuse with a shred of credibility, humanitarian intervention, is defended tenaciously by the war apologists."
See, the problem is none of the things in the above mentioned Iraqi war resolution were really ever proven untrue. So in order for your point to be merited you should at least show which points listed in the Iraqi War Resolution.
I would suggest you prime yourself with this page prior to your rebuttal just to save you the time:
www.husseinandterror.com
"This allowed Bush partisans to believe that the invasion caused "minimal" casualties, when it produced vast human suffering as it destabilized a nation, hurling it into the abyss of ethnic cleansing and the collapse of civil order."
Any single violent death is regrettable. But you can't produce anything worse in terms of "vast human suffering" or "an abyss of ethnic cleansing" than what existed in Iraq prior to the removal of Saddam Hussein. It isn't Vermont, but considering what Saddam did you can't honestly claim that Iraqi's were "better off", can you? Unless the US gassed a few dozen Iraqi villages that I'm not aware of, you insult the honor and integrity of the coalition forces and post-Saddam Iraqi army and security forces by claiming that is at work here.
A lot of good people fought and died for Iraqi's to have a better life than they used to have living under the bootheel of Saddam. And none of the people who opposed the Iraqi war in the beginning gave a crap about the Iraqi people enough to come up with an alternative for removing Saddam.
You stood on the wrong side of history then, and you continue to stand there now. It will only get lonelier over there as time marches on.
"but her motives are plainly ideological.."
As opposed to the motives of the study's authors? It is a little odd to decry someone's supposedly ideological biases when they are pointing out the egregious and politically-motivated errors of the study's authors rather than attacking the authors themselves. When it comes to plain ideological biases, you would be wise to remember the expression about a pot calling a kettle black.
And to further prove my point, you cite one widely mocked study rather than the multiple researchers who have already called it and The Lancet study a bunch of garbage.
Of course anyone who suggests The WHO is a tool of the "plutocratic Bush regime" is best met with mockery and derision rather than reasoned debate.
The ORB study is not a reliable study. It's methodology has not been validated, and what sketchy details are available indicate that it's worthless.
What an odd coincidence that a "worthless" study reaches a count so close to the Lancet total that you can't quite seem to undermine. How inconvenient for you. But, then again, you haven't any credentials at all in this area, do you? You are just trying to discredit high casualty estimates in Iraq.
Yet the organization with the greatest motive to authoritatively discredit the Lancet study, THE UNITED STATES GOVERNMENT, did not choose to do so. Rather, the Bush administration deliberately put obstacles in the path of those who sought to calculate the damage.
You have no explanation for why the Bush administration has failed to produce accurate casualty statistics, but you have plenty of sniping criticism for those who have attempted this feat WITH ZERO COOPERATION from your "compassionate conservative" friends.
Your tendentious hypocrisy is conspicuous in every post. When it comes to defending the rich and powerful, your rules are elastic. When it comes to attacking those who question the rich and powerful, you demand exacting rigor and accountability. How do you sleep at night?
This seems like an odd argument to get into. Does the fact that 400,000 versus 600,000 people died in Iraq make the war more or less justifiable? Further, no matter how many people died, its difficult to pin how many deaths are the responsibility of US military action versus how many people would have died anyway in the ensuing civil war. After all, a lot of people died after the first gulf war and up to few years prior to the Iraq invasion in a related civil war. No one can say for sure how many. Record keeping didn't improve when the Americans came but it certainly didn't get worse.
Regardless, invading Iraq was the stupidest goddamn decision made by a US president in the history of our country. There's been - and likely will be - no benefit to the US taxpayer while the cost has been way, way - WAY too high. That said, Iraq itself probably came out ahead because all of the Republican arguments about how Sadaam was such an asshole are completely true. A shame they couldn't kick him out themselves.
The ORB study is not a reliable study. It's methodology has not been validated, and what sketchy details are available indicate that it's worthless.
What an odd coincidence that a "worthless" study reaches a count so close to the Lancet total that you can't quite seem to undermine. How inconvenient for you. But, then again, you haven't any credentials at all in this area, do you? You are just trying to discredit high casualty estimates in Iraq.
Yet the organization with the greatest motive to authoritatively discredit the Lancet study, THE UNITED STATES GOVERNMENT, did not choose to do so. Rather, the Bush administration deliberately put obstacles in the path of those who sought to calculate the damage.
You have no explanation for why the Bush administration has failed to produce accurate casualty statistics, but you have plenty of sniping criticism for those who have attempted this feat WITH ZERO COOPERATION from your "compassionate conservative" friends.
Your tendentious hypocrisy is conspicuous in every post. When it comes to defending the rich and powerful, your rules are elastic. When it comes to attacking those who question the rich and powerful, you demand exacting rigor and accountability. How do you sleep at night?
See, the problem is none of the things in the above mentioned Iraqi war resolution were really ever proven untrue.
Got it. When you want to launch a war of choice -- a war that's bound to kill many thousands of human beings -- the burden ain't on you to prove that "the things" you're basing your war on are true. The burden of proof is on the side of the war's opponents. Cute. And incredibly pathetic.
Hey Jasper,
I have good news. The Tonkin Gulf incident reports of an attack by North Vietnamese patrol boats on US destroyers have been proven to be untrue. Now we have proof that the Vietnam War was just a big mistake.
Maybe, in 20 or 30 years, we will determine conclusively that the Iraq war was based on deliberate lies. Until then, you can be confident in your righteous support of this war and all the damage it has inflicted on the people of Iraq.
You see, if you believe in something that is not conclusively falsifiable, then you can cling to that belief forever, no matter how much the balance of evidence tilts against it. That's the great thing about being an ideologue: you fit the facts to the policy, not the other way around.
You and Megan want to believe that America did much less damage to Iraq than Saddam. Unless someone produces overwhelming, iron-clad, indisputable evidence that this is not so, you will keep describing the invasion and trashing of Iraq as a good deal for the Iraqi people. It takes a long time to produce such evidence, but by then, you hope people will have lost interest.
The stains of Iraqi blood will never wash off America's hands, no matter how much apologetic detergent you and Megan apply.
The 400,000 excess deaths is within the confidence interval for excess deaths from the Lancet study
Sheesh Tim, you're going to cite the confidence interval?
You know what else is within the confidence interval? 6,000. If a study comes out that says there were only 6,000 excess deaths in Iraq in 2006, will you support that? I didn't think so.
I always prefer to use raw data collected for analysis rather than rely on someone else who has an agenda analyzing data. And I was always suspicious when Lancet published a report of 100,000 Iraqis killed in late Sept, 2004 and another report of 600,000 also in late Sept, 2006. There seemed to be an attempt to influence the US election. And many bought it, hook, line, and sinker.
Meanwhile, raw data collected by the UN and reported in World Almanacs seemed to always list the death rate in Iraq from 8 to 14 per 1,000 per year during the period from 1980 to 2002. For some reason, during the last 4 years, the rate has been 5.6, 5,2, 5.1, 5.14. Odd. The death rate went down? This is all based on the same source UN data. Also, I noticed that the infant mortality rate seemed to go up steadily from 18 per 1,000 in 1980 to 65 in 2002. And this is going down over the past 4 years. Based on the birth rate, it would seem that 10,000 infants are being saved per year. Also, I looked at the 2002 population rate and rate of increase and I can't figure out why the Iraq population went up so much over the past 5 years if so many were being killed in combat, let alone natural deaths.
Oh well. In the future someone may be able to estimate whether the Iraq war saved lives or costs lives. Saddam was rather bad for Iraqis.
Unless someone produces overwhelming, iron-clad, indisputable evidence that this is not so, you will keep describing the invasion and trashing of Iraq as a good deal for the Iraqi people
There's all kinds of evidence: GDP has doubled (tripled buy some estimates), access to basic services like water, sewer, and electricity has doubled, there are hundreds of free newspapers, radio, TV stations -- oh yeah, and they hold elections now. This is all available in the Brookings Institute's Iraq Index.
Here's the biggest one though: even in 2007, large majorities of Shia and Kurds said the war was the right decision. Adjusting for their weight in the population, about 62% of Iraqis overall think the war was the right decision.
And again, it has to be pointed out, the innocent Iraqis who died (i.e. those were not part of the Baathist torture/rape machine) generally died at the hands of terrorists in spite of our troops' best efforts, not because of them.
Saying our invasion of Iraq made things worse for Iraqis is as silly as saying the Normandy invasion and the attacks on Japan in the Pacific Rim made things worse for Europeans and Asians.
Can anyone point me to where the 400,000 figure is coming from? I looked through the IFHS report and didn't see it.
One last note on Iraq: it's hard to imagine any other way we could have armed half a million Muslims and put them to work fighting Islamic extremism.
And Iraqis made it very clear at the polls that they appreciate the continuing gains in security and prosperity, and aren't interested in taking orders from extremists in Iran.
I write a weekly column in the Red Bluff Daily News, "News and Views," and my name is searchable: Don Polson
From my Sept 7, 2007 column:
Since the whole issue of “dead Iraqis” has become part of the latest permutation of the “Bush lied, people died” tripe, my July 30, 2005, ‘News and Views’ is particularly relevant. At that time, a British medical journal, ‘The Lancet’, was prominently quoted by war opponents because it published the calculations of scholars (anti-war, of course) from Johns Hopkins University, estimating Iraqi civilian deaths.
The “100,000 civilians died” myth was thus born, based on shabby methodology that even I could poke holes in. You see, their actual figure of “98,000 extra deaths in the post-invasion period (through July, 2005)” was arrived at—you’re not going to believe this—by surveying Iraqis for anecdotal (word of mouth) reports.
I translated their formula so readers could see that they were “95 percent confident” that between 8,000 and 194,000 civilians died during that period. I wrote “This pathetic attempt at quantification is what passes for objective fact to those on the left. Iraq’s own Health Ministry put civilian fatalities at less than 4,000. The Project on Defense Alternatives estimated 11,000 to 15,000 Iraqis were killed in the war, only 3,200 to 4,300 of those being civilians.”
(Continued in next post)
xyz,
Thanks for noting that the Gulf of Tonkin incident was fake. Contrary to many who call Iraq the biggest mistake in US history and the first war based on lies and the first pre-emptive stike we have a long history of such mistakes. The war of 1812 included the invasion of Canada in which we assumed that we would be treated as liberators. The War of 1812 was mainly about subjugating native Americans, not repelling the British. The Mexican war was about some incident that may or may not have happened in what may or may not be the US. (Lincoln thought the logic was bogus and tried to pass what is know as the Spot Resolution.) But the US stole a bunch of land. We still don't know what the Civil War was about as Lincoln told some the in NE one thing and the the Midwest something else. Of course, the Spanish American war started when the USS Maine blew up for some unknown reason. The US got involved in WW1 basically when the US intercepted a communication from Germany to Mexico urging Mexico to attack the US so Wilson did his pre-emptive attack on Germany. The American people weren't told.
Of course, WWII was started by Japan but Roosevelt really wanted to go against Germany. There was no reason to attack Germany but Germany did declare war on the US, given Roosevelt the legal basis for what he wanted to do anyway.
And Korea was only a "UN sanctioned" war because Russia abstained and China was not on the Security Council then. The list goes on.
The Iraq war seems like an historical blip compared to those other lies. Lies that led to many more deaths in these other wars, and the companion decrease in civil liberties during these other wars, make the anti-Bush criticism somewhat hysterical.
(continued on Iraq deaths):
The Iraq Body Count reported as many as 25,000 civilian deaths through that July. They used news media reports that were tallied by anti-war organizations with an obvious bias to maximize the deaths. Moreover, their own analysis said that U.S.-led forces accounted for a little more than one-third of the deaths, with crime/terrorism/insurgent activities (Iraqi-on-Iraqi) accounting for the rest.
I just checked the latest report at the iraqbodycount.org website and, through July 17 this year, they total between 71,302 and 77,852 “reported civilian deaths resulting from the US-led military intervention in Iraq” which by now is overwhelmingly “Iraqi-on-Iraqi”. So, less than 80,000 through July 17—remember that as you read on.
I then wrote, “A much larger point is the utter hypocrisy of the anti-war crowd. In the 12 years of sanctions, imposed by the U.N. to pressure Saddam on weapons compliance, up to one million Iraqi civilians died, many of them infants. That figures out to about 83,000 dead per year while Saddam used money from approved oil sales to build palaces and starve his people (of food and medicine).
“Can the anti-war left show how they expressed appropriate outrage, at the time, over the hundreds of thousands of deaths Saddam caused? I think not!”
In light of all this, I say with 100 percent confidence, that President Bush, by removing Saddam and thereby ending the sanctions, is responsible for almost 300,000 more Iraqis being alive, mostly children, than would have been under Saddam. Do the math! AND QUIT WITH THE BUSH-BASHING, “SOPHOMORIC, DISINGENUOUS” CHEAPSHOTS!
Current total (as of now, Feb 2009) at Iraqbodycount.org is just shy of 100,000. The 83,000 deaths per year under Saddam would total about 500,000 (x's 6 years if he stayed in power). So mathematically, there are now about 400,000 more Iraqi's alive because Saddam is gone, than if he had remained in power.
Don Polson
Adjusting for their weight in the population, about 62% of Iraqis overall think the war was the right decision.
Why are you war lovers so consistently dishonest? On page 4, the report you cite plainly states the following:
Views of the United States, while still broadly negative, have moderated in some respects. Just shy of half, 49 percent, now say it was right for the U.S.-led coalition to have invaded, up by 12 points from August; the previous high was 48 percent in the
first ABC News poll in Iraq in February 2004.
The report goes on to say that 40% still favor armed attacks against US forces. Yes, indeed, the Iraqis really love us. The Iraqis are so delighted with the US presence that their government has forced us to agree to get our occupation army out of their country in two years. Of course, the Iraqi dead, and those who have fled the country, have no say in the matter, and are thus under-represented in polls.
xyz is a big fan of prisons for toddlers. He doesn't care how many brown people die as long as white hands stay clean.
xyz is a big fan of prisons for toddlers
Child atrocity stories are popular with the war-lovers. Remember the one about Iraqi soldiers pulling babies out of hospital incubators in Kuwait? Nothing like a good atrocity to justify anything you do to your enemies. It doesn't even have to be true if it is repeated often enough.
The funny thing about the unspeakably evil Saddam is that he was our good buddy while he was murdering and torturing his people. There is a famous picture of Don Rumsfeld shaking his hand before Gulf War I. Don was on a diplomatic mission to assure Saddam that the US would not protest his use of chemical weapons against Iran.
Let this sink in, war lovers: the United States government was effectively allied with Saddam Hussein at the height of his vicious tyranny - because he was waging war against Iran, a nation that had just deposed our puppet, the Shah. We backed Saddam. We armed Saddam. We ignored Saddam's atrocities. So spare me your outrage over Saddam. It is just apologist bull$hit you are using to cover Bush's war crimes.
"That's the great thing about being an ideologue..."
Thanks, xyz. It's good to get tips from an insider, and if anyone should know, it's you.
Are you saying that, if you were an Iraqi, you'd rather bet your children's futures on Saddam, Uday and Qusay, rather than on the democracy that Iraq has now where people just recently got a chance to vote once again on their own futures?
Megan, the number I used for excess deaths from the IFHS survey comes from one of the authors. I quote from a November 2008 story by Robin Meija:
Are you contending that Ali is a "lunatic" as well?
Nor is it true that I am comparing the upper end of the WHO confidence interval with the lowe end of the Lancet confidence interval. The WHO central estimate is inside the Lancet confidence interval. That means that the difference is not statistically significant, even before allowing for uncertainty in the WHO number.
For the millionth time, I never want to hear another lefty prattling on about how their side is all about science, and "reality-based" and whatnot. Here we have a bunch of lefties -- best exemplified by xyz -- basically saying: "Screw science, ethics, openness, proper data-gathering technique and all that crap. We'll stick with our dogma, thanks."
xyz is the lefty equivalent of some crazed Appalachian snake handler. He's the nutty relative that everyone avoids at Thanksgiving for fear of being buttonholed in a corner of the dining room and being forced to listen to a wild-eyed diatribe concerning the latest Rovian Bu$Hitler plot to take over the world. There are doctors for this sort of thing, you know.
You just don't know how to count. According to Pelosi, we are losing 50 million jobs per month. Using Pelosi's methodology, Burham might have undercounted big time.
When someone is killed it leaves a body. As murderers discover, a human body is not easy to dispose of, let alone to do it without a trace.
The Lancet study has about a half-million missing bodies. Where did they go? Based on the physical evidence, the findings of the Lancet study are simply and obviously preposterous.
Mohamed Ali is missing about 200,000 bodies. Unless he can tell us where they went, his numbers have no credibility, either.
There are numerical estimates, and then there are objects in the physical world. A weird sort of Platonism has some people more willing to believe in the former than the latter, as if the only real world is a spirit world of numbers. If the physical world does not seem to contain hundreds of thousands of the human bodies those numbers are supposed to represent, then too bad for the physical world. What nonsense.
Tim Lambert says, "That means that the difference is not statistically significant"
I have two studies, one says 600,000 people died in car accidents, and 50,000 from old age. This has been corroborated by a new study that says 150,000 people died in car accidents and contains info that could also be used to derive an unpublished number of 250,000 who died of old age. They also disagree on where these deaths occurred, and disagree on when they occurred over the period. But what's important is that the numbers - 400 and 650 - fit barely into a wide error range, therefore these studies agree. And after all, what more could we need to know from a public health perspective?
Lunatic is about right, Megan.
Also, i recall Tim Lambert in 2005-2006 was all in a tizzy about Iraq Body Count because, he said, Lancet 1 proved it was wildly low and something had to be urgently done to promote this. But the low end of Lancet 1's confidence interval was 8,000. And Iraq Body Count was well inside this interval, yet Lambert seemed to be acting as if they disagreed. Now we know they did not. This was irrelevant for Tim in 2005, but has suddenly become the most crucial of facts now (when it's WHO and Lancet 2). Too bad Iraq Body Count didn't realize they could have just told Tim that their numbers fell inside the lower end of the CI of Lancet 1 (fell somewhere on its dartboard, if you will). I guess Tim didn't realize this at the time as it could have saved him from all kinds of outrage and all his effort spent trying to get others outraged over a disagreement between two sources that were "not statistically significantly different".
Two crap numbers with huge error bars still average to one crap number with a slightly less huge error bar. I can't believe that anyone who takes statistics seriously is arguing, with a straight face, that because two crap estimates overlap we have to believe them.
Do 50 studies and get 50 crap numbers, and then I will start to believe in the Lancet number.
But the most careful studies, that involve counting tens of thousands of deaths, get--tens of thousands of deaths. For some reason, though, we are expected to take the 647 deaths extrapolated to 600,000 +/- 600,000 as gold.
xyz-Scott Ritter saw the children's prisons with his own eyes. He opposed the war in Iraq, as you would know if you read the quote, so he wasn't lying about atrocities to get us into war. These prisons are well documented by multiple sources, such as Human Rights Watch and Amnesty International.
You saying that I love war is just as fair as my saying you love prisons for toddlers. But if it weren't for the war those prisons would still be there, and still be full. And you think American soldiers are worse than people who put toddlers in prison.
Raymond, your recollection of what I wrote in 2005-2006 bears no relation to what I actually wrote. The IBC and the Lancet studies measured different things.
Tim, again, if I am making a mistake, it is a mistake that conflict epidemiologists seem to be making over and over. Now, perhaps you know how to do their jobs better than they do. But they repeatedly said THAT THE DIFFERING ESTIMATES OF VIOLENT DEATHS MEANT THAT THE STUDY COULD NOT BE SAID TO AGREE WITH THE STUDY PUBLISHED IN THE LANCET. They weren't making an impassioned argument--they frankly couldn't seem to believe they were explaining it at all, because the claim was ludicrous on its face.
xyz, no, it's not a coincidence; the shoddier your methodology, the more likely you are to come up with an extreme number. The ORB estimate is the highest estimate, and does not agree with Burnham et al either, as it's wildly higher.
I tangled with Tim Lambert re Lancet 1 at the time of its release.
I pointed out that the sampling was based on an estimate of the size and distribution of the Iraqi population, which had an unknown error.
He replied that nevertheless the research produced a statistically significant result. He did not seem to understand that you cannot get around a measurement error by statistical analysis of the measurements, especially when the error is completely unknown.
Lambert has a lot of dishonesty chips invested in Lancet so it is pointless debating him and using any form of reason, as a lot of people have found out.
The reasonable and fair minded editor of Online Opinion (and author) had this to say about Lambert in a essay, titled "The ABC broadcast bullying and science hooliganism problem"
Lambert was one of the bullies and hooligans cited in this middle of the road political discussion blog with the accusation that Lambert is intellectually dishonest.
Lambert, through his blog Deltoid promulgates whatever the current orthodoxy happens to be, but he does not restrict himself to his blog, frequently diving into comment threads on other online publications. And once you have Lambert on your thread, he sticks closer than a tick, hoping to suck the lifeblood out of the argument until you give up.
http://www.onlineopinion.com.au/view.asp?article=7373&page=3
A few months from now he'll come back hoping everything has been forgotten about this episode and continue propagandizing Lancet in the same way he's done over the years.
So you know, Lambert is an academic and has not written one single research paper in a decade he's been teaching IT in his so-called area of expertise. Not a single one is listed at his public university site, so don't think you are dealing with a well credentialed academic here.
Lambert is a political activist with his own objectives where truth and honesty don't really matter.
Great, fun with statistics and "confiedence" intervals. One of my favourite games.
What is the confidence interval to be placed around a random outcome of 400,000 to allow it to encompass an alternative random outcome of 650,000?
Assuming symmetry here of course, then at a "95% confidence" level it would have to be 400,000 +/- 250,000 (at least).
So we say we are 95% confident that the true value is something in the range of 150,000 to 650,000.
150,000, 650,000, take your pick.
xyz,
Why are you war lovers so consistently dishonest? On page 4, the report you cite plainly states the following: Views of the United States, while still broadly negative, have moderated in some respects. Just shy of half, 49 percent, now say it was right for the U.S.-led coalition to have invaded
Sigh, reading comprehension FTW. Go back and read what I said: Adjusting for their weights, 62% of Iraqis support the decision to invade.
You're citing the unweighted response; the polls wildly overrepresents Sunni Arabs, who were the privileged minority under Saddam's racist, sectarian regime. Unless you believe Sunni Arabs are 30% of the population, which is about twice what most estimates put them at, you need to re-weight.
So, the poll says 5% of Sunnis answered “Yes” to this question, 65% of Shia and 87% of Kurds. This means of 100 Sunnis, 5 would answer yes, as would 65 of 100 Shia, and 87 of 100 Kurds. Weighting by the D3 poll’s breakdown, for 100 Iraqis overall we get 1.5 “yes" from Sunni, 16.5 from Kurds, and 33 from Shia for a total of 51 per 100 or 51%, (slightly higher than the D3 result due to rounding in their numbers). Re-weighting by the real numbers, however, gives a very different result. We get new weighted numbers of .8 Sunni, 22 Kurd, and 39 Shia for a total of about 62%.
But even ignoring all that and using your incorrect unweighted number, how do explain that half the population still welcome this war you claim was so awful for them?
The report goes on to say that 40% still favor armed attacks against US forces.
Besides being irrelevant to the question of whether Iraqis believe the war was the right decision, it doesn't say they "favor" them, it says such attacks are "sometimes acceptable." That could mean a lot of things, including "acceptable in self-defense."
More details on the re-weighting process here.
The CIA Factbook puts the total Sunni population at 32-37%, but remember the Kurds are also Sunni, and are 20% of the total population. That leaves about 15% for Sunni Arabs.
Hey xyz, come on, let it all out. You know you want to blame those neocon Jews. Switch off the Alex Jones for a minute and call a spade a spade.
"The IBC and the Lancet studies measured different things."
As opposed to L2 and IFHS?
Does anyone disagree that L2 reported 4 times as many violent deaths as IFHS?
"Does anyone disagree that L2 reported 4 times as many violent deaths as IFHS?"
I do. If you control for the fact that IFHS made several upwards adjustments which L2 did not, then the real difference is more like 8 times. Full details here.
There is no way that both IFHS and L2 can both be right about the number of violent deaths in Iraq. Tim, smart as he is, can not produce a quote from a single expert (other than Roberts) to the contrary.
David, apparently by "control for" you mean "fail to make the adjustments that the study's authors felt were needed for accuracy".
And neither I nor Les Roberts have said that L2 and IFHS agree on violent deaths. Why the straw man?
"And neither I nor Les Roberts have said that L2 and IFHS agree on violent deaths. Why the straw man?"
Isn't the excess deaths number the straw man? Considering the IFHS didn't publish that number?
Tim, tell me this- if we agree that L2 and IFHS disagree by a factor of 4 on violent deaths, doesn't that already prove something is wrong with the agreement of the surveys?
Violent deaths are a component of excess deaths, right? How can you admit that a major component of the surveys are in huge disagreement, but their totals are supposedly closer so its nothing to worry about? Wouldn't the non-violent components have to be even MORE unreliable to make up the difference so the totals kinda sorta fall in the same margin of error?
Megan, it seems likely that you have misunderstood what epidemiologists have told you, so let's take this one step at a time, so you can tell me what you disagree with.
1. Mohamed Ali, one of the members of the WHO team estimated that there about 397,000 excess deaths in Iraq up to Sep 2006. Do you dispute this? Do you claim that he is a lunatic?
2. The 95% confidence interval for excess deaths from Lancet2 was 393,000 to 943,000, with a central estimate of 655,000.
3. 397,000 lies inside this confidence interval, so even ignoring the uncertainty in the 397,000 estimate, the difference between 397,000 and 655,000 is not statistically significant.
Which of these steps do you think is lunacy?
Tim Lambert, what kind of scientist puts three digits on his error estimate? Don't you know that the confidence interval is itself a statistical estimate, with its own error, or did you skip the day they taught statistics in statistics class? It's not lunacy, it's cargo cult.
A more honest way to cite the Lancet number is 700,000 +/- 300,000. Seriously, take a class, or at least read the Wikipedia article.
There is only one real number of people who died from the Iraq War. Only God knows it.
To take an example from my own research, if you wanted to estimate the value of pi, then you could draw a circle, fill it with tiny squares, and count them. Call this the Iraq Body Count method.
Another, equally valid way, would be to draw a square around the circle. Throw pennies at random in to the square and the fraction landing the circle estimates pi. Call this the Lancet method.
So the Lancet gets pi = 6 +/- 3. WHO gets 4 +/- 2. Both are perfectly reasonable if they did their sampling right. But it would take a special kind of mind-Tim Lambert's mind-to say that Iraq Body Count must be wrong when they get 3.1 +/- 1, because the other two estimates "overlapped".
One method is inherently more accurate. The number of deaths IraqBodyCount gets is not going to be off by a factor of ten.
xyz:
You're so used to invective you don't realize it makes people distrust you.
I think the truth is more complex: Bush was not the person who should have led the US effort. But war with Saddam was probably not avoidable. A wiser leader would have delayed the war a year or so, trying to avoid it while building support, but, once in war, would then would have better managed it.
Bush painted us into a corner by toppling Saddam. But this didn't leave us the option of putting Saddam back in power or letting al-Queda run Iraq. Let's say the Anbar Awakening never happened and the bloody war was still going on full bore now in 2009. Even if we hung Bush for treason, we'd still have to defeat the insurgency in Iraq.
Why? One reason is that we can't let al-Queda get the hundreds of billions of dollars in oil revenue that comes into Iraq each year. And we can't let them have the population to recruit into a war machine.
Well, Tim, I have them on tape, so I'll go back and listen. But I'm fairly confident I didn't misunderstand, because our fact checkers also got the recordings when I handed in the article. Nor, in fact, do you have any reason to believe otherwise.
The WHO survey said that the Burnham survey overestimated the number of violent deaths. You're talking about it as if the number of violent deaths and non-violent deaths were a ratio, so that the important number is the excess deaths, not the distribution between them.
The conflict epidemiologists very sensibly treat the numbers as additive. Excess deaths is the sum of the non violent deaths, and the violent deaths. If the violent deaths are off, it doesn't mean that you misallocated deaths between violent and non-violent; it means that your excess deaths figure is off by the amount of the error in the violent deaths figure. No one except you and Les Roberts appear to think that this is the slightest bit controversial.
There will be, of course, a little overlap between them--say a gunshot to the lung who gets pneumonia and dies a year later. But the size of the difference is not possible to reconcile with the WHO study or the original UN study, and again, people who had seen the data sets were very clear that they did. not. agree. about. how. many. people. had. been. killed. in. Iraq. by. the. war.
I was asking about excess deaths, not violent deaths. Put it this way: it's that a survey that found 100,000 homicides against young males backs up a survey that found 20,000 homocides in young males, and 40,000 heart attack, cancer, & stroke deaths. The distribution of the two kinds of deaths is at best weakly linked: they occur in different populations from completely separate causes.
What's the confidence interval for the IFHS total excess deaths estimate? What matters for whether two studies agree is if the estimates both fall within the confidence intervals of both studies.
If the estimates aren't in each others confidence intervals, the "agreement" can occur simply because one of the studies isn't very powerful.
I just don't understand how a central component of the studies (violent deaths), I would argue THE central component of the studies can be acknowledged to be off by a factor of 4 in the Lancet study, but the total can still be assumed to be accurate.
Doesn't the violent deaths issue call into question the accuracy of the conclusions? How can it not?
Sorry Megan, but do you dispute any of my three points?
If you want to know how many people we killed by invading Iraq, then the important statistic is the number of excess deaths.
"If you want to know how many people we killed by invading Iraq, then the important statistic is the number of excess deaths."
And we all know that is nonsense. We can get a fair estimate on the number "we killed" by examining the violent deaths. Excess deaths can't be laid at our door that easily- correlation does not imply causation.
That's the truly fundamental flaw in this entire argument (quite aside from the lack of accurate data in pre-war death rates, the question marks on methodology, and the current problem that no other study matches up well with L2 numbers. Oh and rushing out a Soros funded study in time for an election).
You can never prove that US action or inaction caused excess deaths in Iraq (at least aside from violent deaths, and the majority of those were inflicted by the other side anyway).
Iraqi deaths have been spiking since Lawrence v. Texas made sodomy laws unconstitutional. That doesn't imply that a lack of sodomy laws are killing Iraqis. I understand that in a sense we've all come onto that playing field with the Lancet study, but that doesn't make it logical. And that's particularly silly when L2 is so far off in violent deaths- the conclusion would have to be that US presence is somehow killing even MORE Iraqis of natural causes than we even assumed last week. Thats absurd, and that's why the studies don't match up.
then the important statistic is the number of excess deaths...
Which you have demonstrated over and over, Mr Lambert, that you do not understand.
I can say with 100% certainty that the number of excess deaths is between 1 and 10 million, but that's not a very useful number is it?
And neither is 700,000 +/- 300,000, and if you knew anything about numbers you would know this (and you wouldn't write scientific illiteracies like 655,000 +/ 298,000). Especially considering that there are people who are trying to count EVERY death who don't come out anywhere near that number.
Should we elect presidents using polls instead of counted votes? That's what you are arguing--that the extrapolated estimate from a random sample is somehow more accurate than the counted number. And if the votes don't match the polls, you want to throw out the votes!
Lambert -
What's your source for your claim that Mohamed Ali said the WHO survey found 397,000 had been killed? You didn't provide a link to any primary source, and all the instances I find in the internet searches come up sourced from the non-scientific leftist political site Mother Jones.
The official WHO press release on the completed result states:
That press release is consistent with the actual survey results.
Do you have any support for your contradictory claim other than the Mother Jones political magazine?
"If you want to know how many people we killed by invading Iraq, then the important statistic is the number of excess deaths."
Actually, the number of excess deaths is most likely negative by now. In January, Iraq's violent death rate fell to 8/100K, which is within shouting distance of the U.S. rate and far, far below any conceivable pre-war rate.
Given that the Lancet study team won't let anyone see the evidence, I'm guessing there was significant curbstoning. Censure was a good start.
Oh, and since we're taking about excess deaths, the fact that access to potable water, sewer service, and electricity have doubled along with GDP means deaths from other causes have probably dropped significantly as well.
One of us is confused, Tim. As I understand it, the IFHS team did not report excess deaths; they reported total deaths.
This being what everyone has been saying to you for over a year now.
You can also listen to Dr. Ali discussing the survey.
I have no source for the 400,000 figure except a MedPage editor who seems to have pulled it out of thin air, and one article in Mother Jones that does not indicate whether that is the central estimate or the outer bound, excess or total, or what the confidence intervals are. But as of the publication of his article last year, Ali--like the other researchers I interviewed--thought that the high prevalance of violent deaths in Burnham's results meant that the studies were not in agreement.
Tim,
You are quick to demand answers from Megan. How about returning the favor by answering a question from me?
This is, as you know, a trap since you don't know the answer. (Nor do I.) Either, you admit that you don't even know the most basic fact about the sampling scheme. Or you have to take a guess, at which point it is easy for me to show that your answer is inconsistent with public statements made by the Lancet authors and/or with common sense.
But, good news! Since you are in contact with Les Roberts, it should be easy for you to clarify exactly what the sampling scheme used in L2 was. How about asking him?
David:
Lambert never answers questions unless he's pinned under a truck and that's is his only way out before the tires go flat.
Lambert has too much dishonesty invested in this to let it go without a fight to the death.
He'll go away, remain quiet for a few months then declare victory if a supportive piece comes out in Mother Jones.
Megan, can I assume that you don't dispute my points 2 and 3? The Mother Jones article makes it perfectly clear that the 397,000 is the central estimate for excess deaths. I suppose it is possible that the reporter got it wrong, but the number is about the same as what I calculate from the numbers in their NEJM paper. But feel free to contact Ali.
And can I remind you, yet again, that saying that the studies agree about excess deaths is not the same as saying that they agree about everything.
Lambert
So we can assume "that you don't dispute" David's point?
Answer his question or STFU
David Kane, I created a new thread at Deltoid to discuss that issue, and you try to distract the discussion here from the point at issue.
You organized the seminar at which Ali presented the IFHS estimate for excess deaths. Perhaps you could confirm this for Megan?
Tim,
1) You created a thread at Deltoid for me to discuss the Johnson et al paper. Thanks! I think we are having a productive discussion.
2) But it would also be nice to have a different thread at Deltoid discussing my specific question:
Now, it is true that this issue touches on the Johnson et al paper, but only tangentially. The exact sampling method used by the Lancet authors has been a subject of some dispute, but that discussion is scattered over several years and many blog posts. If you started a new thread, we could gather all that information together and have a productive discussion.
3) I will check with Dr. Ali about that estimate.
He won't answer the question, David. LOL
Don't fall the his trick of luring you over there hoping his sycophantic followers will change the subject.
Lambert uses every single trick in the book to avoid getting pinned down on answering any specific question that would pin him down.
Lambert please answer David's question and stop changing the subject, you big pussy.
1. Lambert, your point 1 is incorrect and unsupported. Despite being asked to source this claim, you've failed to do so. It is in dispute. Mohamed Ali is not the lunatic, here.
Until you can support your contention that the published results of the IFHS study are wrong and should be ignored in favor of the unsourced assertion in an opinion piece published by the non-scientific, anti-Iraq-war political magazine Mother Jones, your premise for this point is invalid
2. Sure, we can stipulate that the Lancet2 study, which refused to release their their exact survey, validation, or methodology techniques, or even their raw data for independent review, published the numbers you cite.
3. See point 1. Again, here's the actual results from the IFHS survey:
Your conclusion that a difference of 258,000 deaths (which can only be achieved by using your specious 397,000 figure for the IFHS survey), which exceeds the upper bound of the published results of the IFHS survey, is not statistically significant is laughable.
Also, you might want to look at the WHO's FAQ for this survey to clarify a few other points you seem confused on:
Tim,
1) I can't comment on a tenative number that has not been published. What was the confidence interval?
2) You seem to think that if the confidence intervals overlap at all, we cannot reject the null hypothesis that the studies agree. This is not correct, for all that it is frequently done, even by people who ought to know better. It is especially not correct when you have two very different sample sizes, are aggregating a master number from two smaller surveyed figures, the confidence intervals are extremely wide, and the area of overlap is not (AFAICT) large.
3) The authors of the paper, as well as numerous other experts, one of whom was in possession of both data sets, have also flatly rejected the possibility that the two data sets agree, again, because violent deaths are additive, not (1-NVd)(Ed)
Say a study surveys the US population and finds that homicides have gone up thirtyfold, that something like 5% of the prime aged male population in the country have been murdered in the last year, and as a result 1 million people have died.
Another study surveys the US population and finds that homicides have increased modestly, killing another 100,000 people a year, while infant mortality and cancer deaths killed 900,000 extra people. Do those studies agree? No.
The finding of violent deaths is statistically significant, which means that the authors are affirmatively claiming that an extra 600,000 people, plus or minus 200,000, have been murdered in Iraq. You cannot say the study agrees because it is barely possible, by adding up the much lower violent deaths to an unpublished estimate, that they agree. Total deaths=Nonviolent + violent. If you screwed up the violent count--as Ali says, as everyone except you and Les Roberts say--you screwed up the count.
The authors were not counting deaths, and then screwing up the allocation. Some number of people have died in Iraq who wouldn't have died absent the war. Some of them were shot, some of them died in childbirth, some of them drowned, some of them got cancer, some slit their own wrists, etc.
Now, there is some allocation error among deaths anywhere, because if someone has diabetes, maybe his kidneys go. Do you call it death from kidney failure, heart failure, or diabetes? But you don't accidentally think that your daughter who died in childbirth was your son who got killed by an IED. Nor is there some identity where a guy who doesn't get shot this month will die of cancer a few months hence. That kind of slippage does happen--presumably a few people who have been killed by car bombings were going to die soon anyway--but not in anything like the size needed to produce this anomaly.
Just as a thought experiment, let's assume that we are 99.5% confident that Ali et. al. hit it right on the nose--a heroic assumption, given the hamfisted upward adjustments they had to make to their numbers. What does that tell you?
It tells you that something went desperately wrong with the Burnham et. al. count, since after all, we are told they got death certificates for virtually all of their corpses. They *couldn't*, according to their own claims, have misallocated 90% of the deaths.
Studies can't just be said to agree because you can cherry pick one number that might be vaguely similar--particularly if that number is an aggregate of smaller numbers that don't overlap--as I say, 25 year old men don't automatically die of cancer because they didn't get shot this month. If the WHO is right, Burnham et al's sample was somehow massively unrepresentative, which means that its number is garbage, or the interviewers were so sloppy as to confuse the gender, age, and cause of death of most of the people who were killed, which again makes the number garbage. Once you have claimed that a sub-figure is statistically significant, you're stuck with it.
To sum up, the fact that an unpublished tenative number might fall within one confidence interval does not tell us that the studies are not different at a statistically significant level. THe "overlapping confidence intervals" test is not a good test, for all that it is often used in at-a-glance analysis. You have to whip out the stats software and start a-calculatin', compensating for things like variant sample size.
The fact that the studies disagree in virtually every other particular, however, also leads them independently to reject the belief that they agree.
I am happy to concede that Robin Mejia heard and published a figure of 397,000. With our faith in the ability of journalists restored, will you now concede that I heard multiple conflict epidemiologists, who were working with the data sets in question, rejected the possibility that these studies could possibly be said to agree, including Mohamed Ali as published in the New England Journal of Medicine?
If what you want me to concede is that it is highly probable that a whole number of people have died in Iraq as a result of the invasion, then yes, I think you're right--though I think these precise numbers give a dangerous illusion of accuracy in the counts we have, which are all highly uncertain. But I can't concede that these studies agree. They don't say the same thing at all. If Ali's numbers are right, then Burnham et. al. could have at best included the actual number in their confidence intervals by accident.
Is this anything like being right 'by accident' about Saddam's nuclear capabilities?
Oh, you're quite right about overlapping confidence intervals, but not for the reasons you think: increase the levels of certainty and the confidence intervals perforce widen. At a 99.99% confidence interval, the possibility of overlaps become quite large. At a 20% confidence interval, those possibilities become rather smaller.
But for better or worse, we're stuck with the 95% confidence interval (lord knows statisticians have railed against this one for a long time) as the standard and the statement remains more or less accurate. This comment seems rather odd though, because way back in the day you maintained:
But now you you've switched to:
Note that you never performed any such analysis before stating the point I quoted first from last January.
Has anyone ever determined if ScentOfViolets and Daniel Davies are the same person? They seem to share the same positions, passive-aggressive writing style, and lack of knowledge concerning statistical methodology. Just wonderin'.
ScentOfViolets, since you're dim enough to argue that the NBA determines the salaries of all of its players based on some precise stat-based formula, you're probably not tall enough to ride this ride.
If Lancet 2 says that about 90% of their reported deaths were verified by death certificate, and they truly had a representative sample, would you not expect that organizations like Iraq Body Count, who track death certificates would have a similar death total number to Lancet2 ?? Either the death certificate statistic is a fraud, or the sample was not representative of the population. I don't see how you can have it both ways.
See David, Lambert will not answer the question you asked.
The reason is so obvious.
Megan, it is disappointing that after all this back and forth, you still don't understand that Roberts did say that the studies agreed, but rather that the agreed about EXCESS DEATHS. Yes, epidemiologists said that didn't agree and that Roberts couldn't have said that, because he didn't.
So you're not disputing my point 1. I don't know what the confidence interval was. Fortunately we don't need to know it to show that the two estimates for excess deaths are not significantly different.I do not think anything of the sort and did not say anything of the sort. I said that the central estimate for the IFHS for excess deaths (397,000) was inside the confidence interval for Lancet 2. That means we cannot reject the null hypothesis that the number of excess deaths is 397,000, ie the two estimates for excess deaths are not statistically significantly different.
I did not say anything about the confidence intervals overlapping and do not even know what the confidence interval for the IFHS is. But if you are interested in how much they overlap, because the midpoint of the IFHS confidence interval is inside the confidence interval for the Lancet, more than half of the IFHS confidence interval must overlap with the Lancet confidence interval.
Megan, it is disappointing that after all this back and forth, you still don't understand that Roberts did say that the studies agreed, but rather that the agreed about EXCESS DEATHS.
Which is a good part of the dispute. lol
Lambert defending lancet
http://www.ebaumsworld.com/video/watch/924879/
Lambert -
Why do you keep lying and say that no one is denying your point #1?
You have provided a wholly unsourced number whose only originating reference is from a fluff piece in a anti-Iraq-war political magazine. You have been unable to substantiate that number in any way.
That fabricated number does not match any of the published results for the survey under discussion. EVERYONE who has responded to you on this thread has pointed out that the number you are trying to use is wrong.
Yet you continue to pretend that no one is disputing it.
Are you that stupid, or that dishonest?
Nice work Megan, David, hawkeye, Yahoos, et al.
This turkey has needed roasting for a long time.