[Psudo]

The cover story for the National Journal (a Washington insider magazine) entitled Data Bomb has articulated the criticism of the famed Lancet study that suggested that 655,000 Iraqis had died under US watch.

The National Journal wrote:

NJ has identified potential problems with the research that fall under three broad headings: 1) possible flaws in the design and execution of the study; 2) a lack of transparency in the data, which has raised suspicions of fraud; and 3) political preferences held by the authors and the funders, which include George Soros's Open Society Institute.
[. . .]
In 2004, several of the same authors had done a preliminary Iraq study using this method. Also published in The Lancet (and also deliberately timed, by the authors' admission, to appear just before a U.S. election), that article reported at least 98,000 "excess" Iraqi deaths. Perhaps because that estimate contrasted sharply with the observations of embedded reporters, human-rights activists, and others on the ground in Iraq, the media gave it limited coverage.

It's a rather long article, analyzing in great detail the clear bias and methodological failings of the report.

[romanP]

Of course it lacks credibility now, there's probably over a million deaths by now.

[Psudo]

Yes, ignore the fact that it towers over even human rights organizations' estimates and Iraqi Body Count and other organizations sharply critical of the war, the methodological exaggerations of casualty rates, it's funding by a wealthy dove Democrat activist, and it's admitted use to alter election results away from the Republicans and assume it's fundamentally right solely because it agrees with your views. Your rejection of all evidence in favor of your own biases is clearly the right thing to do.

If you'd read any of the article or offered more than a knee-jerk reaction to evidence opposing your faith, you'd know they're criticizing it's estimates at the time they were made, not as modern estimates.

[romanP]

Why do you think that you know what I'm thinking? Your clairvoyant capacity is rather low.

The thing about this report is that you don't need it to prove that the war Iraq has little purpose other than to take Iraqi oil and kill every last Iraqi man, woman and child. There is no measure of success anymore (if there ever was, since the people who started this war could never seem to figure out or properly state exactly what their goal was), all it is is ongoing low-intensity terrorism.

Debating over body counts at this point is like counting grains of sand at the beach to figure out if there's a lot of it.

[Psudo]

Terrifically interesting analysis, except that it has nothing to do with the topic of the thread.

If nothing else, the Lancet study is relevant because it was "featured in 25 news shows and 188 articles in U.S. newspapers and magazines" "Within a week" of it's publication. It has been touted as relevant so it's discrediting is equally relevant, regardless of your opinion of the war at large.

[BartSimpson]

romanP wrote:

Of course it lacks credibility now, there's probably over a million deaths by now.

From what I've heard at least eighteen billion innocent Iraqis have died so far.


[/sarcasm]

[Zipperfish]

I don't think the Lancet study ever had much credibility in the US. It goes against their self-image, therefore it must be rejected regardless of its merits. I note that a main crticism the article offers is that a primary researcher "operated with little American supervision." Kind of sums it all up right there, doesn't it? Obviously, any endeavour undertaken without American supervision is supect.

No one will, of course, ever know for certain what the death toll was because of the US invasion of Iraq. Certianly the US government ("We don't do body counts") couldn't be bothered to even put in the effort. The researchers in the Lancet study actually went door-to-door in a war zone and conducted a statitsical analysis to extrapolate the number of deaths (as opposed to, say, Iraq Body Count folks who sit in their office and read newspaper articles). Until another study uses a similarly direct method and refutes the findings that way, this study stands on its merits, I think. (Though I do accept that there is a wide bound of uncertainty in the estimates, given the realtively small sample size.)

The study estimates may well be too high; or it may be that, for the first time, science allows us to see that the real death count in this war--and, perhaps, in any war--far exceeds the numbers found by using estimation methods of the past. War does not just kill directly, through blllets and bombs. It degardes society, creates lawlessness, gives sadists and brutes free rein, creates deadly penury and kills in many, many other ways.

[Scape]

IBC is the verified body count. As there is no way to truly verify results in a war zone the Lancet is an estimate only which means the total could actually be twice or more then what they came up with.

[Psudo]

I respectfully disagree with your assessment, Zipperfish.

I don't think criticisms set forth in the article can be summarized in the phrase "operated with little American supervision" when there are so many other arguments made (several of which I previously quoted, and several of which I will mention below).

the National Journal wrote:

The key person involved in collecting the data -- Lafta, the researcher who assembled the survey teams, deployed them throughout Iraq, and assembled the results -- has refused to answer questions about his methods.

Some of these questions could be resolved if other researchers had access to the surveyors' original field reports and response forms. The authors have released files of collated survey results but not the original survey reports

Thus, the Lancet study only claims to have gone door to door in a statistically reasonable way but fails to show any evidence that they did so.

Further, "'cluster' sampling is a relatively new methodology that attempts to replicate the logic of public opinion polling in Third World locales that lack a telecommunications infrastructure." The statistical analysis they're using is experimental and, in this instance, returns results 10 times as great as the more traditional methodology used in other investigations.

It's further odd that their 2006 estimate is approximately ten times that of their 2004 estimate; no other analysis shows 90% of deaths occurring during that two-year period and only 10% occurring in the previous year and a half. Not only is the scale of deaths suggested by Lancet out of sync with all other studies, but the shape over time is absolutely unique as well.

The study's unique results ensure that it is either uniquely true or uniquely false. Why believe either? If they physically went door-to-door and if their statistical extrapolation is sound, there'd be a good reason to believe them; however, there is no reason to believe those conditions except their own, unverified word. They have admitted to their attempt to alter the election outcomes in both 2004 and 2006, thus admitting to political bias by the operators. So why trust their claim to have used valid methodology based on their unverified, admittedly biased word?

I cannot prove Lancet is wrong, but I can prove it's possible it's wrong and there is motivation to bias it inherent to it's release. You can't prove Lancet is right, only claim it's possible that it's the first accurate study of civilian causalities ever. That possibility, that overturning of a hundred years of mathematics and history for a mishmash of experimental math and unverified claims, is not sufficient to gain my trust. I assert that it should not be sufficient to gain the trust of any reasonable mind, and that only wishful thinking is capable of extracting trust for this fundamentally untrustworthy study.

[Zipperfish]

Psudo wrote:

I respectfully disagree with your assessment, Zipperfish.

Quote:

I don't think criticisms set forth in the article can be summarized in the phrase "operated with little American supervision" when there are so many other arguments made (several of which I previously quoted, and several of which I will mention below).

I agree. I just thought it was a singularly odd comment to make, and betrays, to my mind, the mindset of the writer of this article. That is to say "If this study was not supervised by Americans it is susupect on that basis alone." It could be I'm reading too much into it, though.

Psudo wrote:

Some of these questions could be resolved if other researchers had access to the surveyors' original field reports and response forms. The authors have released files of collated survey results but not the original survey reports

Thus, the Lancet study only claims to have gone door to door in a statistically reasonable way but fails to show any evidence that they did so.

Further, "'cluster' sampling is a relatively new methodology that attempts to replicate the logic of public opinion polling in Third World locales that lack a telecommunications infrastructure." The statistical analysis they're using is experimental and, in this instance, returns results 10 times as great as the more traditional methodology used in other investigations.

It's further odd that their 2006 estimate is approximately ten times that of their 2004 estimate; no other analysis shows 90% of deaths occurring during that two-year period and only 10% occurring in the previous year and a half. Not only is the scale of deaths suggested by Lancet out of sync with all other studies, but the shape over time is absolutely unique as well.

The study's unique results ensure that it is either uniquely true or uniquely false. Why believe either? If they physically went door-to-door and if their statistical extrapolation is sound, there'd be a good reason to believe them; however, there is no reason to believe those conditions except their own, unverified word. They have admitted to their attempt to alter the election outcomes in both 2004 and 2006, thus admitting to political bias by the operators. So why trust their claim to have used valid methodology based on their unverified, admittedly biased word?

But what would constitute verification? American supervision? I think verification would be repeatability of resutls, but unfortunately, no one has conducted a similar study. The researchers apparently have a "leftist" bias; but most people criticizing the study probably have a "rightist" bias. Who to believe? I'll go with the science. Flawed science, perhaps, but a lot better than pronouncement from partisan bodies.

Quote:

I cannot prove Lancet is wrong, but I can prove it's possible it's wrong and there is motivation to bias it inherent to it's release. You can't prove Lancet is right, only claim it's possible that it's the first accurate study of civilian causalities ever. That possibility, that overturning of a hundred years of mathematics and history for a mishmash of experimental math and unverified claims, is not sufficient to gain my trust. I assert that it should not be sufficient to gain the trust of any reasonable mind, and that only wishful thinking is capable of extracting trust for this fundamentally untrustworthy study.

I guess I disagree that the study overturns anything. I'm not sure that anyone has done a similar study in the past, except for these researchers. Body counts are notoriously controversial anyways--look at the six million and the holocaust--that became so controversial that the number was simply legislated as true in some countries.

[CanadianJeff]

Even if the body count is higher or lower that's really not my concern. the concern should be if the region can be stabilized right? I mean if the US leaves the region is still unstable then the total body count due to the removal of Saddam could climb fairly quickly.

I think that's the problem with the meida. They focus too much on body count when they could be exerting pressure to reform the region or praising what advances have been made. They hold a powerfull card in where to put focus in the war. Personally I don't think they are playing that card at all to the advantage of the people or the Country.

[Scape]

Why would it be the media when it is the Iraqi people dying? The media would only have influence on the western coalition and has zero sway with the will of the Iraqi people.

[Psudo]

This study cannot be considered "science" of any caliber until Lafta's data can be verified without relying upon his word. Science is observational, and none but Lafta observed his data collection. If it is valid, he has failed to offer any evidence of that. No amount of scientifically rigorous statistical analysis can validate invalid data, and we have no reason to believe Lafta's data is valid except his own word; the word of an utter stranger who refuses oversight and offers no proof.

Don't read to much into the use of the word "American" by an American newspaper. They mean "we". Canadians, Brazilians, Kawaitis, Iranians, Russians, Tibetians, French, Kenyans, German, and Inuit are all equally excluded from any verification of Lafta's means of collecting data. The Journal was merely writing to it's readers.

Quote:

So if the body count was half the estimate would it make it better?

It would make the numbers more trustworthy, yes. Obviously it wouldn't mitigate the actual suffering, but only prove to be a better measure of it. Knowing a truth doesn't change the truth, it just makes your reaction more rational. In this case, that means more likely to treat the war the right way to minimize death, suffering, and oppression. Accurate numbers are truly important in that sense, just as knowledge ever is power.

CanadianJeff wrote:

I think that's the problem with the meida. They focus too much on body count when they could be exerting pressure to reform the region or praising what advances have been made. They hold a powerfull card in where to put focus in the war. Personally I don't think they are playing that card at all to the advantage of the people or the Country.

I agree as much as it is possible for one man to agree with another.

[xerxes]

Half the problem with the original study was that people looked at the high-end estimate (~900,000) and automatically dismiss it. Yet no one looked at the range of estimates which weren't so alarmingly high with the low end being around 400,000 estimated deaths.