Sunday, October 15, 2006
Well, not exactly. But almost.
By now, most of you know that Jack Murtha wrote an op-ed piece for this morning's Washington Post in which he reiterated his call for an essentially immediate withdrawal from Iraq. It is not entirely obvious why he thought it was smart politics to title his essay "Confessions of a 'Defeatocrat'", but he does, in fact, argue that it is absolutely patriotic to argue that we can't win in Iraq. Well, maybe. I note, however, that Murtha identifies civilian casualties among the various bad things about Iraq:
In the past two months, 6,000 Iraqis died, more than in the first year of the war.
If Murtha is right -- and I imagine he is -- he is obviously disagreeing with The Lancet, which published papers in October 2004 and October 2006 that claimed, respectively, that 100,000 and 600,000 Iraqi civilians had died because of the war. By The Lancet's reckoning excess civilian deaths have been running at the rate of 700 per day, every day, in the last two years, or more than seven times Murtha's rate. I wonder if any reporter will ask Murtha whether he supports the findings of The Lancet studies? I'm not really holding my breath, but the political consequences of his answer would certainly be interesting.
MORE: It isn't just Jack Murtha that is inadvertently calling into question the objectivity behind The Lancet's publication of the Iraq casualty surveys. Behold, its editor-in-chief:
CWCID: Megan McArdle, who writes "The editor of The Lancet isn't going to stand idly by while people sow doubts about his judgement in publishing the Iraq mortality study. He's going to go right out and sow some himself."
This isn't disagreeing with The Lancet, but with the study by Johns Hopkins U. which was PUBLISHED in The Lancet.
That study IS the largest by far number yet given. The U.S. has deliberately not counted civilian dead, so all such counts are estimates. Iraq Body Count has been working with confirmed burials, but this leaves out many who are never identified. All other studies are based on estimates and we have to check on methodology.
But while Bush and others have been quick to dismiss the Johns Hopkins study, no one has yet said what is wrong with their methodology or proposed an alternative method that might be more accurate. Bush says that his generals give him different numbers, but this is problematic for several reasons: 1)Rumsfeld ordered no civilian body counts from the beginning, so the generals cannot have accurate info. 2)Bush hates bad news and this is well known so it is very unlikely that the generals would give him unwelcome info.
"no one has yet said what is wrong with their methodology or proposed an alternative method that might be more accurate" is either ignorant or a lie. Take your choice. Statisticians who vehemently disagree with US Iraq policy have denounced the bogus 'study' by the Johns Hopkins bunch. It's not hard to find those who show what is wrong with the methodology (check out junkscience.com for a few links).
Projecting such cluster studies (which are difficult enough to do well in the US) is a little shaky when tracking infectious diseases. It is *worse* when trying to track 'clusters' of non-infectious problems like cancer or the human equivalent of mad-cow disease because the 'clusters' tend to *not* be representative of the general universe.
Finally, using cluster analysis to 'track' and project 'excess deaths' in a battle zone is either insane or driven by ideology. Even assuming that it would work in a neutral setting, having even a few subjects lie about the number of dead in the household would incredibly bias the projection.
Perhaps the authors and Lancet could have helped their cause by holding the study until after the US elections. Nah. That *is* their cause -- affecting the elections.
And there is not even a sense of shame for using such tactics.
"no one has yet said what is wrong with their methodology or proposed an alternative method"
NO ONE AT ALL? Are you serious? Ok, here is one I heard: they extrapolated the final number by asking about 500 houses in the middle of the most violent area of Iraq and applied the figure to everywhere in the country. That makes their conclusion undeniably wrong. Unless you think every province of Iraq is being torn apart by death and destruction (nope).
"no one has yet said what is wrong with their methodology"
Not to be confused with other responses to this point -- the authors themselves pointed out a number of ways in which the results could be wrong. Read the last section of the paper. Now, those potential errors cut both ways, but some are more serious than the others -- quite possibly the most serious is that internal migration from high-mortality to low-mortality areas since the population statistics were last developed for the sampled clusters significantly overstates the resulting counts.
It doesn't relate directly to the credibility of the report given the *very* difficult security situation in Iraq, but the research team isn't releasing its raw data, interview disposition counts (refusals, etc.) or even the log files of the SPSS and Stata sessions used to draw samples and calculate results. That last item is a bit odd, since they're just sitting on a hard drive somewhere. I am sympathetic to their not wanting to get their surveyors and respondents killed, but eventually these data *will* have to be released (even the GIS data can be transformed to globally anonymize them but retain key spatial information like distances between points).
The point is not that Lancet’s numbers are inflated (however true that may be), rather that its editor, by indulging in his vein-popping, screaming leftist rant, has exposed his publication’s lack of objectivity.
But let us not quibble about numbers. Lancet has arrived at a remarkable medical finding: war causes mass death. Thank goodness for the geniuses at Lancet. Who would have suspected such a thing? And now from its editor we receive the rest of the syllogism: England and America have waged war; Therefore, the Axis of Anglo-American Imperialism desires to kill children instead of building hospitals and schools.
If you don’t quite follow his logic, you must admit that the guys seems very sincere about his feelings.
BTW, This poor fellow’s loss of composure makes one wonder if he has ever read his publication’s studies on the dangers of high blood pressure.
"here is one I heard: they extrapolated the final number by asking about 500 houses in the middle of the most violent area of Iraq and applied the figure to everywhere in the country."
I don't doubt that that is one you heard.
But no, they didn't.
"As a first stage of sampling, 50 clusters were selected systematically by Governorate with a population proportional to size approach, on the basis of the 2004 UNDP/Iraqi Ministry of Planning population estimates (table 1). At the second stage of sampling, the Governorate's constituent administrative units were listed by population or estimated population, and location(s) were selected randomly proportionate to population size. The third stage consisted of random selection of a main street within the administrative unit from a list of all main streets. A residential street was then randomly selected from a list of residential streets crossing the main street. On the residential street, houses were numbered and a start household was randomly selected. From this start household, the team proceeded to the adjacent residence until 40 households were surveyed. For this study, a household was defined as a unit that ate together, and had a separate entrance from the street or a separate apartment entrance."
(I copied this from the article. I apologize if it violates copyright, but I'm pretty sure it doesn't.)