The Lancet study asserting 650,000 war-caused deaths in Iraq from the US invasion to mid-2006 is now strongly suspected to be fiction – the correct figure is closer to 151,000 according to a study that relies on more than 1,000 clusters rather than 47 as in the Lancet study. It is also interesting to note that the Lancet study was financed by George Soros with at least one of its authors now claimed to be a Saddam apologist and with claims the study was released to influence election outcomes. The figure of 650,000 would have made the Iraq conflict one of the bloodiest in recent military history – it remains a nasty conflict but the actual death toll is apparently one quarter that reported on in Lancet.
On the local scene those promoting the value of the Lancet figures (here*, here) have either not responded or responded by discussing the new results but not admitting error. Perhaps fair enough – the issues involve complexities and already there have been denials of ulterior motives and bias.
I reported the results here and stated that the sampling methodology seemed sound – it obviously was not. In a sense the dispute is a grim insult to the dead – 151,000 is still a lot of human misery. But strong views were expressed stridently criticising those who disbelieved the Lancet figures.
Perhaps there is a lesson for us all here.
*Update: John Quiggin has now responded to the claims made here both in the comments below and on his blog site.
Saturday, January 12, 2008
Subscribe to:
Post Comments (Atom)
Posts
34 comments:
It shows you that trying to collect messy data which can can be interpreted many ways means you end up with a lot of error, and henceforth you shouldn't be using a word like "correct", otherwise you'll just look like the fool when it undoubtedly happens to be revised again.
I deleted an initial negative comment that criticised a colleague. I'll delete all further such comments.
To amplify Conrad's comment, it's obviously wrong to compare point estimates. Also the two numbers are estimates of different things - violent deaths vs all excess deaths.
If the confidence intervals for the same variables overlap, then there's no conflict on standard assumptions.
But even if they don't there's no reason to automatically prefer the study with larger sampling size. Non-sampling error can go either way, and any study in a war zone is subject to it.
All we can say with high confidence (bearing in mind that this study covered only the first three years of the war) is that the death toll is almost certainly greater than 100 000, and probably much higher.
To restate what I said here
http://johnquiggin.com/index.php/archives/2006/04/30/linkfest-2/
"Apart from ... continuing statistical disputes over how many tens of thousands have died, there’s nothing much on Iraq where we seem to have run out of things to say."
I guess the only change required a couple of years later is to substitute "hundreds" for "tens"
Again I received a post which presented certain arguments (which had some interest) but which contained extremely insulting remarks about a colleague of mind.
The comment like many of this type was anonymous.
Blogger does not give me the opportunity to selectively edit comments so I have deleted the lot.
I like strong arguments and criticism but the points must deal with the issues and not be overly personal.
Again a comment deleted - I will not allow this blog to be used by people interested in pursuing personal vendettas. There are other blogs where you can do that but not here.
I won't bother responding to anonymous - he's drunk too much of the Kool-Aid. The conspiracy theories needed to deny reality are getting loonier and loonier (I'm surprised anonymous hasn't mentioned that Soros is one "those people").
Surely the 150k estimate is actually reasonably consistent with the 650k estimate because they're measuring different things. The former is the number of violent deaths, the latter the number of excess deaths (eg it includes those who've died from disease due to destroyed infrastructure).
The persistent attempts by even otherwise reasonable people like Harry to put lipstick on the very, very ugly Iraqi pig still amazes me. They can't admit to themselves they've been partly responsible for a holocaust.
Derrida derrida,
The original Lancet study claimed total deaths of 654,965 and deaths due to violence of 601,027. The New England study estimates total deaths due to violence of 151,000. The claim you make that the latter figure is 'actually reasonably consistent' with the earlier Lancet figure seems false.
And I agree with anonymous that it was the anti-war group that sought to use the Lancet figures to criticise the war.
I tire of attempts to brand me as a Nazi who supports a holocaust. This is deeply offensive and false. I have criticised the conduct of the war but do not believe that Iraq should now be abandoned to terrorists particularly when now even Sunni groups are actively fighting al Qaeda.
The vast bulk of casualties in Iraq occur as the result of terrorist actions that I revile.
If you read the last paragraph in the post you will see that I don't seek to put 'lipstick' on any 'Iraqi pig' (your words).
I do however believe it is important to be truthful about such things as casualty counts and that was the point of the post.
"and deaths due to violence"
Actually, I think it should be reduction in average life expectancies. If you destroy all the infrastructure and people die 20 years early, its clear you are responsible for the deaths. If this also causes violence (and refugees, which I'm sure we'll see), you are still responsible.
"it was the anti-war group that sought to use the Lancet figures to criticise the war"
What has that got to do with the credibility of the study?
Anonymous: Take a stress pill and a first year statistics course. Larger studies are not necessarily better than bigger ones, and nor do they necessarily have smaller confidence intervals. In addition, even if what you were saying was completely true, it wouldn't be the worst scientific fraud of the decade -- its easy to think of worse things, especially given that it makes no difference to the perpetrators, whom it seems absolutely nothing will happen to.
Conrad and John Quiggin also need to net out all those deaths caused directly and indirectly by UN food for oil sanctions, now that we know the ongoing fallout they caused and would still be causing if their prescription were still being applied
http://news.yahoo.com/s/afp/20080103/wl_mideast_afp/iraqunrestcorruption
observa
Actually here's the gist of the link-
'Iraq's culture of corruption stems from the actions of the international community and the controversial UN oil-for-food scheme, the deputy prime minister Barham Saleh said on Thursday.
Speaking at a new anti-corruption forum in Baghdad, Saleh said that the programme, run between 1996 and 2003 while Iraq was under UN sanctions, and what he charged was the body's wasteful use of money were to blame for the rampant corruption that bedevils Iraq.
"A large responsibility for the outbreak of corruption in Iraq lies on the international community," said Saleh.
"The scandals of food-for-oil and the wasting of public riches by the UN... is evidence of the serious damage that has deepened the problem in the country."
For seven years before the US-led invasion of Iraq in 2003, the oil-for-food programme allowed Baghdad -- which was under a punishing trade embargo since its invasion of Kuwait in 1990 -- to sell oil in return for cash to buy food and medicine.
In 2000, then president Saddam Hussein began making the right to purchase its oil under the UN programme conditional on the buyer's willingness to pay kickbacks.
The regime diverted about 1.8 billion dollars (1.2 billion euros), an independent probe found in 2005.
"The Iraqi people want a clear statement on how this money was administered in the past, to take back the stolen money, and to chase and call into account those who played with the public's money," said Saleh.
An analysis by the Berlin-based Transparency International in September found that Iraq was one of the three most corrupt countries in the world alongside Somalia and Myanmar'
observa
Conrad and John would readily recall all those reliable sources of theirs at the time-
'Are "a million innocent children...dying at this time...in Iraq" because of U.S. sanctions, as Osama bin Laden claimed in his October 7 videotaped message to the world? Has the United Nations Children's Emergency Fund (UNICEF) discovered that "at least 200 children are dying every day...as a direct result of sanctions," as advocacy journalist John Pilger maintains on his Web site? Is it official U.N. belief that 5,000 Iraqi children under the age of 5 are dying each month due to its own policy, as writers of letters to virtually every U.S. newspaper have stated repeatedly during the past three years?
The short answer to all of these questions is no. The sanctions, first imposed in 1990 after Iraq's invasion of Kuwait, are administered by the U.N., not the U.S. They were first imposed on all exports from Iraq and occupied Kuwait, and all non-humanitarian imports, in an effort to persuade Saddam Hussein to retreat within his own borders. After the Gulf War, they were broadened to include a dismantling of Iraq's biological, chemical, nuclear, and missile-based weapons systems, out of fear that Hussein would otherwise lash out again. Estimates of sanctions-era "excess" child deaths -- the number above the normal mortality rate -- vary widely due to politics and inadequate data, especially concerning children older than 5. The dictatorial Iraqi government, which has blamed nearly every civilian funeral since 1991 on sanctions, claims there have been more than 600,000 deaths of under-5-year-olds these past 11 years (4,500 per month) and 1.5 million deaths overall.'
Time now to net out all those reliable estimates eh chaps and what would that leave, even with your similarly reliable Lancet figures?
observa
A further string of 5 anonymous comments have been deleted on the grounds that they directed personal derogatory comments.
Comments need to keep to the issue under discussion. The comments have caused offense and damage the credibility of my blog.
I don't like personal attacks generally but when they are made anonymously they are completely intolerable.
Observa, the excess death measure takes account of deaths caused by sanctions. The reason the effect is small is that the Oil for Food program, despite the corruption it engendered, ended the massive loss of life caused by the earlier sanctions regime.
In any case, earlier commenters are right to say that these sanctions were imposed at the behest of the US - the UN decision to relax them was made despite US opposition.
So it's doubly wrong to suggest that the US invasion can be justified becuase it brought an end to the suffering caused by US sanctions.
More deletions - this is getting tedious.
Personal attacks by anonymous posters are not wanted.
Derrida derrida, I inadvertantly deleted your comment. I appreciated your point.
My apologies - it was completely unintentional but it seems I cannot recover the comment.
It's a neat argument splitting the outcomes of economic sanctions into sanctions A and sanctions B John. I concede the point and you'll recall the Iraqi Transitional Government was the government of Iraq from May 3rd, 2005. After that, they were responsible for security outcomes in Iraq, albeit it was awfully nice of some of the COW to stay on with their consent, to help out and try and improve security and reduce the death toll they were experiencing amongst the populace. Adjust the casualty counts accordingly.
observa
Harry the defintion between violent and excess deaths sems to me to be problematic.
Having read at CT about the various surveys I can't see what all the hoohaa is about statistically
Harry the defintion between violent and excess deaths sems to me to be problematic.
You obviously haven't followed the debate which isn't surprising.
Lancet says they tabulated off the death certificates which would carry the cause and circumstances of death. So we could compare.
However Lancet isn't "able" to give the raw data out for L1. Also it isn't able to furnish some deatils of the raw data in L2 which would allow reviewers to see if there was any fraud committed and if the survey was actually done in a random manner.
Mark U says:
Harry,
It puzzles me why you allow people to post comments anonymously. It is confusing when readers of comments cannot tell whether anonymous comments are coming from one or a number of people. Surely it is reasonable for people to at least provide a nick name and reveal their URL to you. Then you could avoid the problems that you have had here.
(Having said that, I just ried to post under the Nickname option and it kept on saying my URL had an illegal character, so I have had to post unde anonymous.)
Sorry about that Mark but, yes, I cannot always understand why people will not use a consistent pseudonym.
Sometimes those who do so are seeking to use my blog (and other blogs) to attack people. They even manage to use different IP addresses. It isn't going to happen.
Ultimately they risk prosecution in the courts because some of the things they say are defamatory.
I will be ruthless with anonymous
insulting posts and will delete them.
cannot always understand why people will not use a consistent pseudonym.
Because your silly system doesn't seem to support it, harry. I'd be happy to identify myself but at times the identity box doesn't work and kicks out the comment. It's very user unfriendly, so I'm buggered if I'm going waste more time worrying about identifying myself if you can't be bothered doing something about. Obviously it hasn't been important to you.
Mark U says:
Anonymous at 10.58 pm.
You could always say who you are in your comment.
If you can post anonymous you can identify yourself - at a pinch as Mark says - by typing your name at the end of the comments.
death certificates were used in L2 not L1 and the latest survey did not survey in the areas of most casulties and merely extrapolated figures from IBC. Not a promising way for a statistican to use.
Post a Comment