That's a very interesting article, wbull. 

I definively recomend anyone interested in the subject read that, along with
- The survey itself:    i.a.cnn.net/cnn/2006/images/10/11/human.cost.of.war.pdf  (cut & paste into browser)
- It's appendix and charts:   i.a.cnn.net/cnn/2006/images/10/11/appendices.human.cost.of.war.pdf  , and
- The relevant CIA figures (above, posted by Shek)

Coming back to the article -
David Frum is unabashedly partisan, in addition to being a profesional political hack. Crafting partisan arguments is what he does best - that actually is his line of work. You'd expect that if anyone out there would be able to provide a compelling summary of the arguments againt the Lancet study, that would be Mr Frum.

And of course the authors of the Lancet report are NOTpartisan in any why, altruistically publishing their "findings", so they are immediately credible, despite the fact that we have shown on these pages their motives for releasing their report? Cant have it both ways -  they are either credible on the merits, or you make a partisan ad hominem attack on David Frum for being partsisan, in a feeble effort to discredit his analysis before it even gets started. The bottom line is that it is up to the Lancet people to defend their methodology and findings, which to this point they have NOT done, other than to publish the report.

He does show very good technique - for intance, he refers to the Survey as "the Bloomberg study" throughout his article.  That's inaccurate; at least four institutions (the Jhons Hopkins School of Medicine, the U of Bagdhad medical school, MIT's Center for International Studies, and the Lancet's scientific review board) were directly involved in the production and vetting of this survey. 

Not only that, the term "the Bloomberg study" is clearly aimed at undermining the credibility of the research - who the heck is this "Bloomberg team" anyway?.  Though the reader may be willing to grant a certain amount of credibility to some of the world's best scientific institutions (like Jhons Hopkins, the Lancet or MIT), it's much easier to poo-poo "the Bloomberg team"

However, on the substantive facts, Mr Frum doesnt do nearly as well:

Point 1 - "Most Iraqui experts" estimate the casualty rate as 30-60,000 dead, while"the Bloomberg team" estimates 655,000.  Hence "the Bloomberg team" is not credible.

"Phooey!" to that.

Nice concise English, btw. About as much as your "experts" on the Lancet report.

For one, the Lancet study involved far more "Iraqi experts" than almost all the unnamed competing studies cite by Mr. Frum. That included 10 Iraqi MDs doing months of field reseach in every single province of Iraq, in addition to the Medical School of the University of Bagdhad activally working alongside both Jhons Hopkins and MIT.
 
For another, the studies reporting 30-60,000 dead are based on "secondary source" aggregation methodologies (adding up deaths reported by press reports and morgue records).  That's an *absurd* approach

Please explain that to us poor statisticians - the obviousness eludes us, since it is attempting to define the universe, which in many ways is a sounder approach

Re: the Lancet study, Shek's recent points were why they would spend time in their report comparing their findings to what we know about mortality in other wars (Congo, East Timor, Vietnam, etc.) and why they only used 14 months of pre-invasion data as their benchmark for the expected death rate.

As I see it, the answer to the first one is quite straightforward; it was an external 'sanity check' to see if their findings were reasonable.  The question there was "are the results we are obtained consistent with what we've seen in previous conflicts?", and the aswer to that was a resounding "yes" - 2.5% of the population dead over 3+ years of armed conflict seems to be highly consistent with what we have seen elsewhere.  If anything, it's unusualy low.

There are two issues here.

First, they attempt to benchmark their final numbers, but they hardly spend time developing this. For example, they cite Vietnam and 3 million casualties. Is that over 25 years? 15 years? 8 years? What about the causality of these deaths? Agent orange? Air strikes? Indiscriminate carpet bombing? The blockade of North Vietnam? Besides making zero effort to give details on Vietnam, they make no effort to motivate that comparing Vietnam and Iraq is a valid comparison.

For East Timor, they cite a number provided from CNN. Do you find this very scholarly? Do you expect researchers well versed in statistics to rely on a citation from CNN? They criticize the IBC method, and yet they engage in using a news source. Coincidentally, their number is the highest number out there, 200K, published by Amnesty International. While I didn’t do an exhaustive search, I couldn’t find any scientific methodology behind the 200,000K number that Amnesty International cites. However, why do they not cite the following? 

link

This study cites 103K as the number of excess deaths over the 25 year period. This would equate of half of the excess death rate claimed by Iraq.  Once again, they provide no details and make no attempt at motivating how East Timor is a good benchmark for Iraq. 

Lastly, they cite the Congo as a benchmark, once again without giving many details that could then be used to motivate why it is a good benchmark for Iraq. As an aside, when you read the study they cite, you find that the methodology used for Congo is different. Furthermore, the study they cite doesn’t even talk about the methodology in computing the wartime deaths. Instead, the study they cite looks at the peacetime (post-war) death rates and talks about how there had been 500K of deaths in the 16 months following the peace, with 0.1% of the deaths resulting from violence. The bulk of the deaths resulted from HIV/AIDS, lack of vaccinations, and malaria. Those are certainly problems in Iraq, not! Either their scholarly research capabilities are not worthy of the doctorates they have, or else they purpose chose black box figures. 

The second issue is that they don’t even touch benchmarking for their pre-war numbers (other than citing their original study, which is slightly incestuous). Given their attempt to benchmark their overall results, don’t you find it disingenuous that there is no benchmarking discussion over their pre-war results, for which there are benchmarks to compare against?

Bottomline, their benchmarking efforts are poor and not sufficient. 

The implication (which they did not state) is that numbers previously presented (such as Bush's claim last December that around 30,000 Iraqis had died) are extremely unusual.  0.125% of the population dead over 3 years of fighting is on the face o