New doubts raised about Lancet’s Iraq death toll study
posted at 10:33 am on January 4, 2008 by Allahpundit
Send to a Friend |
Share on Facebook | printer-friendly
You know the one, merrily cited on every Democratic politician and nutroots blog in Nutrootsville as proof that the number of Iraqis killed since 2003 in fact exceeds 600,000, fully an order of magnitude greater than the most trusted body count numbers. People have been challenging it since 2006, but read this National Journal report today pulling the various strands of criticism together. There’s not a whole lot new here that I can see but it’s impressive to see it all in one place and presented against the backdrop of the investigators’ political biases. To wit:
No matter whether a latent desire to feed the American public’s opposition to the war might have shaped these studies, another audience was paying close attention: jihadists who used this research as a justification for killing Americans. Roberts already believed that jihadi attacks were, in part, driven by the international image of the United States. “The greatest threat to U.S. national security [is] the image that the United States is a violator of international laws and order and that there is no means other than violence to curb it,” Roberts wrote in a July 2005 article for Tirman’s center. When NJ asked Roberts about the risk that his estimate would incite more violence, his confidence seemed to waver for the only time during the interview. “This area of study is a minefield,” he said. “The people you are talking about are the same kind of people who deny the Holocaust.” Does it give him qualms that some of those people use his study to recruit suicide bombers? “It does,” he replied after a pause. “My guess is that I’ve provided data that can be narrowly cited to incite hatred. On the other hand, I think it’s worse to have our leaders downplaying the level of violence.”
This is the same guy who ran for Congress in the Democratic primary in 2006 and told an interviewer “I consider myself an advocate.” In fact, he initiated an earlier death toll study that found 100,000 dead by 2004. One of the co-authors of that one told the Journal last month that he now estimates the overall toll to be in the neighborhood of 250,000 — which of course is less than one half the estimate of what Lancet says.
If you don’t have time for all of it, start halfway down with the “Potential Problems” section. And don’t miss the tidbit about Cluster 33. Can one car bombing account for 76,000 deaths? Yep, sort of.
You must be logged in to post a comment.
Blowback
Note from Hot Air management: This section is for comments from Hot Air's community of registered readers. Please don't assume that Hot Air management agrees with or otherwise endorses any particular comment just because we let it stand. A reminder: Anyone who fails to comply with our terms of use may lose their posting privilege.
Trackbacks/Pings
Trackback URL
Comments
I thought this was disproved early last year.
rmgraha on January 4, 2008 at 10:35 AM
Maybe so, but I still see liberals quoting the numbers, so it didn’t stick.
James on January 4, 2008 at 10:39 AM
Yep, folks. According to this “study”, U.S. interdiction in Iraq has led to a death toll higher than the combined toll of the conventional and nuclear bombings of Tokyo, Hiroshima, Nagasaki, Dresden and Hamburg.
Hoodlumman on January 4, 2008 at 10:42 AM
Can we please not confuse the issue with facts!
TugboatPhil on January 4, 2008 at 10:49 AM
If only we could turn things around in Iraq, there might be fewer deaths.
You know, like a surge or something.
What’s that? Oh, did anyone tell the Democrats?
fogw on January 4, 2008 at 10:53 AM
Little sloppy on the count. It’s 6,000,000 now and thats just the little girls raped with bayonets, their eyes gouged out and eaten (or sent home as earrings to vicious women in the US) who were killed by having their throats torn open by human teeth.
BL@KBIRD on January 4, 2008 at 10:54 AM
The jihadis routinely violate international law by dressing as civilians, hiding behind civilians, deliberately targeting civilians, using mosques (i.e. buildings for religious use) to store weapons, firing from mosques, and torturing their prisoners. (And I mean real torture, much more severe than waterboarding.) Roberts needs to stop looking at the splinter in the eye of the U.S. and instead notice that giant Sequoia in the eyes of the jihadists.
Bigfoot on January 4, 2008 at 10:54 AM
Reminds me of a quote from Seinfeld:
CP on January 4, 2008 at 11:01 AM
If I get the time, I will most likely be able to debunk the 2004 Lancet mortality study this year. Quite simple really - using the UNDP’s 2004 data from their Iraq Living Conditions Survey, I will show that the pairings of Iraq’s regions were based on a false assumption. The data for the survey was collected and analyzed by an organization here in Norway called FAFO. All I have to do is get a hold of the data, and I’ve already been communicating with those who have it.
The main guy behind the ILCS survey from FAFO has often criticized Les Roberts’ Lancet study, that it greatly exaggerates the death numbers and is a generally poor study methodically.
Seixon on January 4, 2008 at 11:02 AM
I hit a bunny once with my car. By the “science” of this survey, I’ve killed 400,000 bunnies and 3,000 ducks. As well as a wooly mammoth, which is sort of weird. But the science proves that I did it so I must have.
mjk on January 4, 2008 at 11:04 AM
I asked Burnham to see the data as soon as the thing was published (I live less than an hour from Hopkins) and he politely refused, although he did quickly clarify some other questions I asked (e.g. one would get the impression that since they didn’t take any personal information outside the households the teams didn’t cross-reference death certificate information across households, which might lead to double-counting of multiply claimed persons, although they claimed a protocol to deal with this). He wouldn’t share the data processing routines with me at the time either, which I thought was downright odd. Those are programs, not data. No privacy/safety issues at stake.
I’m an econometrician — and the cluster sampling methodology, at least as described in the paper, is a valid method. Let’s not knock sampling theory or we’re going to have Tim Lambert’s people over here saying we’ve declared war on Mathematics or something.
The main issues are (1) why not share the data and programs, (2) did the use of old population data for the districts skew results (as BR suggested it might), and (3) was the field data collection work actually conducted as described — was the protocol adhered to?
DrSteve on January 4, 2008 at 11:06 AM
I wonder how many people that report killed. Things like this only made the situation worse.
tomas on January 4, 2008 at 11:23 AM
‘And it’s NOT the truth if I don’t believe it!’
N. O'Brain on January 4, 2008 at 11:42 AM
Many people defend The Lancet study by saying its been peer reviewed. How much rewiew could there be if they aren’t allowing this information to be released?
BohicaTwentyTwo on January 4, 2008 at 12:13 PM
I live in a suburban city - pop 26k. A little over a year ago a neighbor of mine was murdered by her teenage son (allegedly) and stuffed in a freezer. It was the first murder here in about 4 years. However, if you were to use Lancet’s survey methods you would likely get an annual death toll in the thousands because everybody here heard about it.
If a car bomb killed the grocer who sold the best pomegranates in the arab world, everyone who went out of their way to get their pomegranates from him would know about it and the statistical “death toll” would skew upward dramatically as a result.
In summary - DUH!
I tried to explain this once on Kos and it got me banned.
Wingo on January 4, 2008 at 12:20 PM
This is why the Lancet ultimately must shoulder the blame. It wasn’t peer reviewed. It went right to press without being peer reviewed on whimsical statements that the very lives of the researchers depended on its publication?
The Lancet took a ridiculous study and put it on the map. It was never reviewed by any peers.
Oh yeah, and George Soros funded the study and it’s not like he hasn’t expressed interest in manufacturing controversy to support his political objectives.
Recall, George Soros, despite being Jewish, was invited to Saudi Arabia(Jeddah) of all places to give a speech advocating the Euro become the petro dollar in order to check the alleged aggression of the United States. That was in 2005.
Remember, George Soros has no influence whatsoever on the value of currency if you ignore how he made his billions with Jim Rogers in the 70s and that brief, but fruitful, bilking of the Bank of England in the 90s.
gabriel sutherland on January 4, 2008 at 12:39 PM
I’d like to refer you to a Tech Central Station article, Thoughts from a Lancet Skeptic where they had an article and discussion on the Lancet study.
The comments are very telling. I’m the same Mazztek in them as well.
I quote myself here, in response to a libturd who takes the Lancet as gospel:
Mazztek on January 4, 2008 at 12:56 PM
I argued this extensively way back when the first study was released. The huge confidence interval alone was enough to dismiss the study as meaningless (and recent analysis shows including Fallujah made it so large the srudy could not even say with 90% confidence that more than ZERO people had died as a result of the war), but there are some serious new nuggets here.
— Lafta, who did the actual survey work, was also employed by the Hussein regime to puff up some numbers to make the sanctions look bad. Remember, this was the regime that employed Baghdad “American troops are not in the city!” Bob. Lafta tellingly refuses to talk about his methods.
— The Sadr City data, where 12 of the 20 households surveyed improbably have exactly 1 death each from a car boming that struck a market used by tens of thousands, is almost certainly fraudulent.
— The Nineveh “heaping” problem indicates fraud was probably perpetrated throughout the data.
— A car bomb that happened after the surveying was supposedly over was also included, again indicating pervasive fraud.
This is about as discredited as a survey can be.
TallDave on January 4, 2008 at 1:28 PM
There’s peer review, and there’s peer review. When I do studies for the Feds, I have to give them everything they need to take the raw data, process it, and analyze it themselves (using my code) if they want to. In my code I nest sampling commands in other processing steps so that the lists of sampled elements are used immediately and aren’t left lying about for possible messing-around-with. I try to eliminate any opportunity for anyone to question my methods.
I also forgot to add: The irony of the conditions the authors placed on the “release” of data last year — including a no-critics-need-apply rule — is so rich you need a wide-open insulin drip to read them:
http://www.jhsph.edu/refugee/research/iraq/
No. 1 is my personal favorite. Roberts’ own views are illegitimate subjects for conversation, but woe be to any on-the-record skeptic who wants to see the numbers (such as they are).
DrSteve on January 4, 2008 at 1:35 PM
There’s this bizarre mythology that the “scientific community” is somehow less prone to corruption or idiocy than any other community, and it just ain’t so. There are just as many liars, frauds, and creeps among scientists as there are among plumbers, and a brief review of science news over the last few years will give good indications of that.
Numberwatch.co.uk, (enter, number of the month, start reading).
That site focuses mainly on Britain, but that’s just a primer.
The peer review system is officially broken, in the sense that it isn’t even remotely doing what it’s supposed to do.
Merovign on January 4, 2008 at 2:18 PM
DrSteve,
The methodology in the 2006 study was more sound than the 2004 study, where they used unproven assumptions to cut corners - ones that I intend to prove invalidate the entire study (2004 edition). Read the assumptions used to pair clusters in the 2004 study. I have a fundamental understanding of statistics, and that methodology only flies if the underlying assumptions being made are correct. But there was no evidence provided, other than “belief”, that the assumptions being used were valid.
Seixon on January 4, 2008 at 2:21 PM
In fact, a Norwegian scientist and a South Korean scientist have been busted for falsifying data and creating fraudulent research reports over the past few years. Their work was also peer reviewed. It doesn’t help that they’re peer reviewed because peer review most often does not uncover fraudulent data. They rely on good faith and the previous credibility of the scientist who conducted the work.
Seixon on January 4, 2008 at 2:23 PM
I knew the Lancent Study was garbage from day one. The only people who take the Lancent Study seriously are the nutroots. According to icasualties.org, 40,004 Iraqi Civilians have been killed since the start of the war. 40,004 is still a lot, but it’s far less than the numbers many liberals would like you too believe.
SoulGlo on January 4, 2008 at 4:32 PM
A case study in bad science.
Zorro on January 4, 2008 at 6:10 PM