Via commenter AMac:
Tim Lambert has a new post that responds to some of Shannon’s arguments and takes me to task.
AMac himself posted this comment on Tim Lambert’s site, and I think summarizes well most of the important concerns about the Lancet piece. AMac is more cautious in his inferences than I would be, but I may be wrong and he may be right. He is generally thoughtful and fair-minded, and his contributions to the comment threads on this topic have been very helpful. I suspect that the data used in the Lancet study are of such low quality as to be of little practical use, but additional scrutiny of those data can’t hurt and should suggest ways in which future surveys could be improved.
UPDATE: AMac helpfully forwards some additional links in his comment below.
UPDATE 2: AMac provides this link to his latest and greatest post at WOC. Worth reading.
Thanks for the kind words, Jonathan, and the push into the limelight of notoriety. Since comment-specific permalinks don’t seem to work reliably, some curious readers may be put off by your link to my “peer review” at Deltoid. Tim Lambert’s post ‘Lancet Links’ is here; scroll waaay down to timestamp (28/3/2005 13:48) for the text. It begins, “sorry–long comment.”
Also see Heiko Gerhauser’s comment immediately following (28/3/2005 21:40) for a suggestion of how Roberts could have handled Fallujah.
Kevin Donoghue (29/3/2005 04:51) posts a dissent from a pro-Roberts perspective. It highlights how thoughtful people can have very different standards on what constitutes good conduct in scientific publications.
dsquared (29/3/2005 09:16) expands on how a cohort study arrives at excess-death estimates.
Finally, Joe Katzman has graciously dedicated a thread to the “peer review” at Winds of Change.
It’s kind of funny to read dsquared’s summary of the Lancet study. So, he thinks it shows that, probably, violent death rates have gone up a lot, and consequently overall death rates have, probably, also gone up. Which of course, he thinks, shows that things have gotten “worse” and the Iraq war is therefore a mistake.
I completely agree that violent death rates have probably gone up a lot, and consequently overall death rates have, probably, also gone up.
However, there is plenty of evidence, much harder evidence to boot than the Lancet study, that indicates that violent death rates have gone up a lot (sufficiently so that a drop in overall death rates is unlikely when comparing 2002 with March 2003 to March 2005).
So, that’s a not exactly a new finding. Nor do I believe that dsquared’s conclusion that the Iraq war was a “mistake” crucially depends on the finding (how about the effects on the region, on Iraq in future years, “international law”, the value of democracy and free speech in Iraq …).
More importantly, that finding is not what’s controversial about the study. What’s controversial is the impression given by the authors that of the order of a 100,000 civilians have been killed by coalition bombing. Careful reading of the key sentence alone shows that it might be much less (most by violence could be just over half and only part of that is bombing, but most could also be 95%, and most of that could also be 95%). On the other hand they say their estimate is “conservative”, easily implying that more than 100,000 coalition bombing deaths is not all that unlikely.
Tim and dsquared know that the study doesn’t support such a finding. For them this “misinterpratibility” of the summary is, hmm, a minor error that hardly weighs in the balance when compared with how terribly, terribly misunderstood and denigrated the Lancet study is by its mostly “pro-war”, and mostly clueless, critics.
Heiko,
While I found it striking that so many reports confused variants of “100,000 killed by airstrikes,” with “98,000 excess deaths from all causes, CI 8,000-194,000”, I don’t think either Tim Lambert or dsquared have made this error. Not that I have seen in recent threads on “Deltoid” or “Crooked Timber” or this latest exchange at Winds of Change.
Nor do I believe that dsquared’s conclusion that the Iraq war was a “mistake”
Heiko, you’ve put the word “mistake” in quotes there and I didn’t use it. It so happens that I do believe that the Iraq war was a mistake, but I don’t believe that the Lancet study proves that it was a mistake and I’ve never said that it does. I think that the Lancet study shows that something has gone very wrong with the Iraq invasion, because it ought to have been possible to carry out the Iraqi invasion in a way which did not increase the death rate.
I am having to be very particular about not letting people put words in my mouth, because Shannon Love appears to be now claiming that I have accused US forces of war crimes. This is an outright lie and I consider it to be a very offensive one indeed. So as I say, I’m having to be even more persnickety than usual at the moment.
Hi Amac,
what I wrote is:
“Tim and dsquared know that the study doesn’t support such a finding. For them this “misinterpratibility” of the summary is, hmm, a minor error”
I didn’t say that either of them actually made that error. And they haven’t, to my knowledge, at least not in the crude form of outright equating the figure for excess deaths with bombing deaths.