A misleading sidebar on this BBC story on Iraq only presents the IBC count of civilians killed in Iraq (which is guaranteed to be a significant undercount) and omits to mention the Lancet estimate of roughly 100,000 excess deaths.
Gabriele Zamparini wrote to the BBC seeking an explanation. The BBC's Steve Hermann tries to justify it:
The Lancet study is a snapshot taken more than 18 months ago and though the methodology has been widely acknowledged as standard, there has been argument about whether the sampling method is the most appropriate for this kind of survey.
Alas, it's another case of a reporter who doesn't understand statistics. You just loudly repeat your arguments and the reporters become reluctant to cite the study because they lack the knowledge to see that the arguments are rubbish. Yes, the sampling method (cluster sampling) does have some problems in measuring deaths in a place like Iraq. It will tend to underestimate them.
The BBC also gets it from pro-war bloggers who don't understand statistics. In this post one Scott Callahan belabours BBC reporter Paul Reynolds with emails about Kaplan's bogus dartboard argument. He also cites a "debunking" by George Gooding. You may know Gooding better as Seixon.
And remember: the total number of extra deaths as a result of the Iraq war is very roughly 200,000 once you include the increase in disease and accidents since the invasion. This number is more likely to be too low than too high since it comes from doubling the 100,000 estimate from the Lancet study (which just covered the first eighteen months) and violence has worsened since then.
Update: Callahan objected to being described as a war blogger so I changed it to "pro-war blogger".
Tim,
Your reference to my post is in error on a number of fronts. First, I'm not sure what a "war blogger" is, but I'm pretty sure I am not one. I blog primarily about the media in the UK, with a particular focus on its coverage of the US.
Second, I neither cited nor linked to Gooding or "Seixon". I linked to David Kaspar and cited Fred Kaplan. (The Kaspar piece itself included a link to Seixon, although it was not discussed at all in detail.)
Third, I did not "belabor" Paul Reynolds about Kaplan's arguments. Indeed, in my correspondence with him, as detailed in my post, neither Kaplan nor his critique were ever mentioned. Not once. Instead, my exchange with him was focused primarily on the journalistic thought process, or lack thereof, by which he came to include the Lancet study in his article. It was not a discussion of the study itself.
Regards,
Scott Callahan
The American Expatriate
Scott,
I fail to see how Tim is 'wrong' on a number of fronts. It appears his opinion is different than yours. That is all. It doesn't make it wrong in a nonbinary nonmonochrome world.
After reading your tightly-wound anti-beeb blog, however, if you don't mind me sharing Scott, when I see stuff like what I see in your banner, the tone you used in your widdle correspondence and the type of comments in the comment section, the image thus created can easily be interpreted as belonging to the personality type category, IMHO, in which warbloggers are placed.
Jus' sayin'.
HTH,
D
Indeed, in my correspondence with him, as detailed in my post, neither Kaplan nor his critique were ever mentioned.
Kaplan's critique wasn't mentioned? I took this for a reference to Kaplan's critique (a.k.a. Kaplan's Fallacy):
Yes, it is true to say that the correspondence was not a discussion of the study - the idea was to rubbish the study without putting up any worthwhile argument against it.
Scott, I looked at your blog, and you sure seem to be a partisan supporter of the Iraq war. If you can point to a post where you say the war was a mistake I'll be happy to make a correction.
I did find this review of "State of Fear":
See RealClimate on Crichton's science howlers.
And as Kevin has pointed out, you did use Kaplan's critique.
Ha. You claim a reporter "doesn't understand statistics" (surely you meant geostatistics? as all area studies involve geography... or don't you know much about geostats?) because they don't cite a flakey statistical study which extrapolates from 988 "Iraqi households" located in only 33 "neighbourhoods" to the entire country of Iraq, and then you pull a number (200,000) out of your jacksie simply because you think it looks nice (those big round zeros) and fits your theme!
Do you think the Australian Bureau of Statistics will be offering you a job any time soon?
Aside from "the statistics", why don't you read up on the Multiple Areal Unit Problem (Openshaw 84 from memory).
Related
=======
Here's a rep for Human Rights Watch commenting soon after the Lancet study, well known chumps of the Chimpler:
"The methods that they used are certainly prone to inflation due to overcounting," said Marc E. Garlasco, senior military analyst for Human Rights Watch, which investigated the number of civilian deaths that occurred during the invasion. "These numbers seem to be inflated."
http://www.washingtonpost.com/wp-dyn/articles/A7967-2004Oct28.html
And here is Galalsco retracting
As for the rest, if you have some specific criticism, share it with us. Vague references ot Openshaw 84 don't constitute a refutation.
So a war blogger, by your definition, is a blogger who doesnt oppose the war? WTF-ever.
Oh, and the fact that he made a reference to Rand means exactly what? Obviously it means you will use any irrelevant bullshit to avoid admitting you were mistaken. Why didnt you just come out and say, "well what would you know about it you war-blogger Randian facsist!"?
FFS, what they say about you really isnt exaggerated is it? I thought they were just being unkind.
Kimble, you are a small minded abusive prick.
That post differs from yours, how, exactly?
I called lambert on using irrelevancies as insults in place of any counter argument. I called him on his false logic. These are things he has been accused of in many places. I didnt automatically believe them, but with my first exposure to his writing all he did is confirm them.
stop being a sheep david
Everybody: please be civil, or I will delete your comments.
Kimble: I wrote "partisan supporter of the Iraq war". This is not the same as someone who doesn't oppose the war.
I didn't say anything about his Rand reference. What I did do was point to a detailed explanation of how Crichton got his science badly wrong. Did you read the linked post? Do you have any specific objections to it?
On his first point,
"First, I'm not sure what a "war blogger" is, but I'm pretty sure I am not one."
To which you responded, "If you can point to a post where you say the war was a mistake I'll be happy to make a correction."
You say he is a war blogger, he says he isnt, he says he blogs on media, you say you will consider him a war blogger until he can prove he thinks the war was a mistake.
Actually, what you do is say he is a partisan supporter of the Iraq war. As if that was what he was objecting to!
On his second point,
"I neither cited nor linked to Gooding or "Seixon"."
To which you responded, "Oops sorry, my bad :) :) I will change the post to remove that reference. Thanks for pointing it out."
BAHAHAHAA! Actually you responded, "I did find this review of "State of Fear": ... See RealClimate on Crichton's science howlers."
Exactly WHY you did this, when it has absolutely nothing to do with his comment, is yet to be determined for certain. But I have little doubt you were trying to point out that you think he has made some other mistakes as far as scientific studies go. As if his possible failings in this area somehow erases you own, more recent and obvious, errors.
On his third point,
"I did not "belabor" Paul Reynolds about Kaplan's arguments. Indeed, in my correspondence with him, as detailed in my post, neither Kaplan nor his critique were ever mentioned."
He did not mention Kaplan in his correspondence, but he did use the critique. Though, to be fair it is not really an idea that can be solely attributed to Kaplan. So the term Kaplans critique COULD have been misinterpreted.
"What I did do was point to a detailed explanation of how Crichton got his science badly wrong. Did you read the linked post? Do you have any specific objections to it?"
Nope, I didnt need to read it. I dont really care about Crichton's book or any errors it may contain. Why should I? In the context of this thread and what I have said, why would it matter? Why did you bring it up?
I neither cited nor linked to Gooding or "Seixon".
Actually, the link is to David Kaspar, who presents no substantive critique of his own but writes:
Maybe it's relevant that Kaspar had updated his post, adding Ray D in "support"; but Scott Callahan's claim that he wasn't acually citing Gooding is no more than a quibble.
It's amazing how the same tactics are employed on all fronts of the war on science. Like the Discovery Institute attacking biology, the Lancet critics (with the exception of a tiny numerate group) are just trying to create the appearance of a serious controversy. Then they go hounding the media to "report the controversy". Scott Callahan evidently doesn't care to find out whether the critique he (deniably) cites has any merit at all.
Tim,
I was a supporter of the invasion of Iraq and I remain in favor of a US military presence there. Still, the "war-blogger" label seems deceptive to me. The war is neither the main purpose of my blog nor is it the primary focus of discussion on it. I do comment on coverage of the war, just as I comment on coverage of all kinds of things related to the US. If I am a "war-blogger", then I suppose I am also an abortion blogger, a Supreme Court blogger, an evolution blogger, a religion blogger, and a global-warming blogger, among many others. In short, your characterization is, at best, greatly misleading.
Regarding Kaplan, as I said, I did cite his Slate article in my post, but I did so to supply context to my readers for my questioning of Reynolds. My "belaboring" of Reynolds, however, had nothing to do with Kaplan, and, again, he wasn't mentioned a single time in my e-mail exchanges. My exchange with Reynolds concerned his journalistic decision making process which led to the inclusion of the Lancet study in his article. Had he considered the credibility of the study before reporting it? The answer: No. Why hadn't he mentioned the controversy surrounding the study? Answer: Because he assumed his readers already knew all about it. If he assumed they already knew about the controversy, why did he then illogically assume they didn't know about the Lancet study itself? Answer: Silence.
In other words, the focus of my post was not the study itself, but was rather Reynolds' journalist practices and thinking. You may think that the study is highly credible, and therefore could be cited without any qualifications. But Reynolds, by his own admission, did not. Hence the question - why did he then include it without qualification?
I don't expect you to be able to answer that, but clearly I was not "belabouring Reynolds about Kaplan's bogus dartboard argument."
Regards,
Scott
Kevin,
You quote from my post:
"Do you think the fact that the study had a 95% confidence interval of 8,000 to 194,000 is irrelevant when trying to judge the usefulness and meaning of the 100,000 figure?"
...and say that you "took this for a reference to Kaplan." You should have taken it as a reference to the study itself, since it was the study, not Kaplan, that reported the 95% confidence interval of 8,000 to 194,000.
And, as far as I am aware, the notion that the confidence interval might be relevant when trying to interpret the usefulness and meaning of a given statistical extrapolation cannot be attributed to Kaplan. If I am wrong, and he is the first to assert that it is so, then please do correct me.
Regards,
Scott
Kevin,
"Maybe it's relevant that Kaspar had updated his post, adding Ray D in "support"
Indeed. It was that to which I had intended to direct my readers.
"Scott Callahan evidently doesn't care to find out whether the critique he (deniably) cites has any merit at all."
Evidently you don't care to find out whether your assumptions are correct before you make such charges. As I said, my link was intended to direct people to Kaspar's post of Ray D's analysis, not Siexon.
In any event, I had read the Siexon post (although I did not cite it) along with the argument that ensued in the comments section and continued on Deltoid. The sole criticism of Siexon was that he was wrong about the sample not being properly random. However, since not all of the criticism of the Lancet study rests on this point, and indeed none of the criticisms in my post does, it is not clear to me why you think it "evident" that I don't care whether the criticisms I actually do cite have any merit.
Evidently you are under the mistaken impression that, if a particular criticism has no merit, no criticism does.
Regards,
Scott Callahan
Scott supports US aggression that is in flagrant violation of the UN Charter and the Nuremberg Code. Glad we got that little factoid made clear. He also supports the US occupation of Iraq. He obviously doesn't give a hoot what the Iraqi people want - that is irrelevant. I am sure that the Germans felt that same way in World War Two, in occupying many European countries. They probably argued that their occupation couldn't end until they had put in place de facto forces - like the Vichy in France - that were capable of providing 'security'. To heck with what the vast majority of the populations of France, Holland, Belgium, Poland etc. etc. wanted, or what the Iraqi people want now. Its only the desires of the super powerful and super violent states that matter.
Finally, rule one of any aggressor nation or nation committing state terror: don't count the number of your victims. This enables any estimated or extrapolated count to be dismissed as 'mere propoaganda', thus enabling an entire suite of atrocities to be sent down the memory hole. The global body count for US aggresson/state terror (directly or by delegation via proxies) probably stands at some ten million since the end of WW II. Latin America alone is littered with the victims of US sponsored mass murder: tens of thousands dead in Nicaragua in the 1980's, hundreds of thousands dead in Guatemala, even an estimated 3,000 dead in Panana during "Operation Just Cause", where US forces removed a regional thug (Noriega) who committed most of his crimes while on the CIA payroll. And how many US victims were there in Indochina during the Korean and Vietamese wars? Two million? Five million? Did the US make any attempt to tally up the slaughter? Of course not. As long as the figure remains conjectural then the apologists for US agression can forever argue that the crimes were either overestimated or did not even occur.
Fast forward to Iraq. There's no doubt that US forces have committed crimes against humanity on a large scale in this country too, and that they are serial violaters of Geneva Conventions. Dahr Jamail's web site details US crimes on an almost daily basis. But these crimes are self-effacing. The well-oiled western state-corporate propaganda apparatus ignores them, and thus, they didn't happen. Just like previous US crimes 'didn't happen'. They don't become a part of history because they never entered history.
Jeff,
My presence here is not to argue the merits of having invaded Iraq or remaining there for the time being, (although if it were, you wouldn't pose much of a challenge, based on your most recent contribution). Instead I am only here to correct Tim's characterization of my blog and the post to which he linked.
But thanks for your, er, interesting thoughts.
Regards,
Scott
Scott,
Kaplan's Fallacy is not original to Kaplan, as Daniel Davies (d-squared) acknowledged when he gave it that name (in this context; elsewhere he calls it percentile fetishism) :
Now, look at what you are doing when you put questions like this to Reynolds :
I take this to mean that you consider the claim incredible. In reality the methodology used had a 50% chance of underestimating the mortality rate, but nobody reading your post is likely to be left with the impression that you think the true figure might be quite a bit higher. Your presentation is misleading in the same way as Kaplan's: using the fact that the study's estimate is imprecise, you seek to argue that it is too high.
... my link was intended to direct people to Kaspar's post of Ray D's analysis, not Siexon.
Ray D does no analysis to speak of. He trots out the Garlasco comment, which was retracted as Tim points out above. He asserts that the sample used in the study "was clearly far too small to produce reliable, conclusive data". He presents no statistical reasoning in support of this claim (and never did so when he advanced it in comments here). Then he tries the Andrew Bolt line: "Where are all the funerals? Where are the pictures?" Apparently an Iraqi can't dig a grave without hordes of photographers descending on the scene. He mentions the infant mortality rate (short response: it's a longitudinal study and infant deaths didn't figure largely pre- or post-war) and he concludes with the fact that the period prior to the invasion was relatively peaceful (true, but irrelevant to any evaluation of the study).
Evidently you are under the mistaken impression that, if a particular criticism has no merit, no criticism does.
On the contrary, as my previous comment notes, there are numerate Lancet critics. Some of them commented here quite regularly when the study was fresh, and they certainly helped to make this blog interesting and sometimes educational so far as I was concerned. AMac and Ragout especially come to mind. But you don't see them describing that the study as "infamous", "discredited" or "well-debunked". That's not the language of someone who wants to raise serious questions about the work of reputable researchers, nor for that matter, of someone who cares about impartial reporting.
Kevin,
I wasn't talking about Kaplan's "fallacy". I was talking about whether or not the notion of "confidence interval" is relevant to understanding the usefulness and meaning of a given statistic. I assume you agree it is relevant. Therefore I assume you also agree that there is a meaningful difference between claiming (falsely) that the Lancet determined a death figure of 100,000 and claiming that (as in fact was the case) the Lancet determined,, with a 95% confidence level, a death figure of somewhere between 8,000 and 194,000.
"I take this to mean that you consider the claim incredible."
Whether or not I do (and I do) is neither here nor there for the purposes of my questions to Reynolds. What was relevant was whether he considered it credible or not. He claimed to have no opinion. Which led me to my further questions.
Further, whether or not I personally think the claim is too high or too low (I suspect it is too high), I was not suggesting that Reynolds should have used a lower number. I was suggesting that Reynolds should have properly characterized the study by providing its confidence interval, thus allowing his readers to make up their own minds on whether a study with such a wide band of confidence provided any useful information.
"...but nobody reading your post is likely to be left with the impression that you think the true figure might be quite a bit higher."
They shouldn't be left with that impression, because I don't think it is higher. However, I did provide the confidence interval figures, thus providing the relevant information about the study by which they could draw their own conclusions. Reynolds did not. He force fed the misleading 100,000 figure to his audience, even without having determined in his own mind how credible the figure was. Which was precisely the point of my criticism of him.
"[Ray D] asserts that the sample used in the study "was clearly far too small to produce reliable, conclusive data". He presents no statistical reasoning in support of this claim..."
Isn't that precisely what the wide confidence interval tells us? A larger sample would have produced a smaller 95% band, and hence would have, by definition, provided more reliable, conclusive data, would it not? Perhaps you think that a study which concludes that the number of deaths was 98,000 (er, give or take 90,000) is reliable and conclusive. If you do, it would seem you have a different definition of those words than I do.
"On the contrary..."
Well, then, I have no idea why you came to the conclusion about me that you did.
Regards,
Scott
Scott, I am only stating the obvious. The trend should be abundantly clear: whenever a government is involved in some kind of military activity, it makes little or no attempt to count the victims of its actions. In fact, the US and UK governments have gone out of their way to block several attempts (e.g. by NGO's) to count the number of victims of the Iraq war, just as they didn't exactly want to co-operate over the effects of sanctions on the Iraq popluation between 1991 and 2003.
In contrast, our governments and the media that amplify their rhetoric certainly make a big deal out of the victims of officially designated enemies. The Lacet study was a brave attempt to provide some background to the human costs of the 50,000 plus bombs that the coalition dropped on Iraq. As expected, the state-corporate media swung into high gear, along with the apologists (people like Seixon and yourself) in decrying the study as flawed and thus invalid.
As for the merits of invading Iraq, shouldn't we leave that up to the Iraqi people to decide? What the heck do you know about it, sitting in the comfort of your home in the UK? A poll last year in Iraq suggested that 1% thought that the Americans had invaded to bring democracy whereas 62% thought that the Americans were there to control the countries resources. More than 70% thought that, even were the Americans to allow a democracy there, they would ensure that it was one that they could 'control' - much like the pseudo-democracies the US supports in Central America with abhorrent human rights records.
Finally, if you want to debate these issues with me, do so. But making pedantic remarks like "My presence here is not to argue the merits of having invaded Iraq or remaining there for the time being, (although if it were, you wouldn't pose much of a challenge, based on your most recent contribution)" are hollow and sanctimonious. I have found such vacuous remarks to be typical of those who supported the aggression against Iraq: by dememaning your opponents with baseless insults you think you win the argument. Based on your earlier posts and on your blog site, you would't stand a chance against someone with even a sliver of historical knowledge.
Kevin/Tim,
I am on the verge of a two week vacation (holiday in Brit-speak), and will be unable to continue this discussion much longer
Accordingly, I'll leave you with the last word.
Regards,
Scott
Scott, your objections to my post are the tiniest of quibbles. Since you don't like being described as a war blogger, I've changed it to pro-war blogger. Seixon's stuff is the first thing mentioned in your link, so I don't see how you can claim that you weren't referring to it. And it's actually better than Ray D's stuff. Kevin's already pointed the problems with Ray D's work, but if you wish I can go over it more thoroughly in another post. And you claim that the confidence interval makes the result meaningless citing Kaplan in support and then criticise Reynolds for not reporting the confidence interval. That is the Kaplan argument. It doesn't matter whether or not you cited Kaplan in your emails to Reynolds.
Finally, you raised a significant issue:
Citing it with qualifications is misleading, since the objections you and Ray D make to the study are without merit. Not citing it all is even more misleading, since it implies that the study should be ignored. Reynolds' point to you was that if he did either of those things he would (quite rightly) get beaten up by Zamparini. And there wasn't room in his story for a 20,000 exposition of all the arguments for and against. If you want to review those arguments there are 78 posts and a couple of thousand comments on this blog alone.
Tim,
Just so I understand, is it your position that the confidence interval is irrelevant to an understanding of what the study means? Or that, even worse, citing the interval is "misleading"? If so, why do statisticians include them when reporting their work?
Regards,
Scott
Scott,
Of course confidence intervals are relevant, that's why the Lancet expects authors to provide them. Mostly such things are used for hypothesis testing. If, for example, the 95% confidence interval straddled zero, we would have to entertain the possibility that the invasion might have saved lives. Some proponents of the invasion stressed that possibility beforehand. You may be sure that they wouldn't have dismissed a study which estimated a saving of 100,000 lives just because there was a 2.5% chance that the true number of lives saved was less than 8,000; nor should they be expected to dismiss it.
Now, if the 95% band straddled zero, it would be fair to say that the sample size was too small to be meaningful. As it turned out however, the effects of the invasion were sufficiently drastic to enable us to make pretty robust inferences from the study: mortality rose, the increase was very substantial, the main cause of the increase was violence, particularly the use of heavy weaponry by coalition forces. Many a larger study has yielded a lot less empirical fruit without being described as "useless", to mention just one of the pejorative terms you fling around.
Maybe someone is "claiming (falsely) that the Lancet determined a death figure of 100,000" but the BBC is not, as your quotation shows:
...I did provide the confidence interval figures, thus providing the relevant information about the study by which they could draw their own conclusions.
Any reader who took your account at face value would have been misled. You wrote:
That statement, like Kaplan's, gives the impression that the distribution is uniform. In reality, as no doubt you know, the true figure is a lot more likely to be in the range 80k-120k than 8k- 48k, or 156k - 196k. You also ignored the fact that the 100,000 figure is obtained by excluding the most violent cluster studied, so in that sense it is a conservative estimate. I'm not suggesting you could have done justice to the study in a few lines since I don't pretend to be able to do that myself. But you weren't even trying to be fair. That's normal enough in controversy, but it sits uncomfortably with your demands for greater fairness from the BBC.
Have a good vacation, holiday, whatever.
Regards, Kevin
Scott,
That might look as if I was answering on behalf of Tim - as the timing makes clear, I hadn't seen your short comment when I posted - I was dealing with your question to me.
Once again, Happy vacation.
Well, if somebody doesn't believe the Lancet study, they could just ask the coalition forces ... oh no wait, they don't do body counts do they.
Right or wrong?
The Lancet figure of 100,000 is NOT significantly different from 10,000 at a 95% confidence level?
In other words, it would fail a hypothesis test on whether the estimate was significantly different from any figure between 8,001 to 193,999 at 95% confidence.
The only thing not stradling zero means is that the Lancet study showed that some people died. Well done. War kills people. Someone alert the Nobel prize committee, we have a nominee for the 'no shit sherlock' award.
Even though the chance of the true figure being BELOW 8,000 is only 2.5% (which is still one in forty), the true figure has over a one in six chance of being under 70,000 or over 130,000.
Cluster sampling MAY be more likely to underestimate the death toll, but it MAY also overestimate it. There is no proof that the Lancet study UNDERestimated the death toll.
Removing the outlier makes the estimate MORE accurate.
At best, the Lancet study only succeeds in proving some people died, and doesnt help in anyway to determine a useful estimate of the number.
Kimble:
No, it shows there is a greater than 95% chance that excess deaths occurred after the invasion.
What do you mean by useful? Come to think of it, what does Scott mean by "reliable". I've never heard of a confidence interval being used as a measure of the "reliability" of a statistical study. From what I remember of my statistics, "reliability" has a particular meaning in statistics, but I suspect Scott has a different meaning.
Kimble:
So there's a 2/3 chance of the number being between 70,000 and 130,000.
I'd call that a pretty good chance that there have being large numbers of excess deaths.
If, for example, the 95% confidence interval straddled zero, we would have to entertain the possibility that the invasion might have saved lives.
As has been pointed out before on this blog, when all the data is included, it does. Only the arbitrary exclusion of the Fallujah cluster drops the variance sufficiently to raise the bottom end of the confidence interval above zero.
Artifacts of the data like that are what make responsible journalists use large-sample all-province survey data, instead of the one survey that produces the highest headline number.
There is a one in three chance that it isnt between 70,000 and 130,000. And you are OK with that, Kirby? You really WANT to believe it all, dont you?
Is that so Soru? I did not know that, but it makes sense because it would affect absolute variation.
'
[insert any comment from Kimble here]
What was that Tim title again...hmmm...what was that title...oh, yes:
Flypaper for innumerates.
Best,
D
Just in case anyone cares about my efforts to replicate the results of the Lancet study, alas, I have not succeeded as of yet. Tim, whom I have been trying to keep in the loop, was kind enough to post the data that the Lancet team provided a few months ago. Unfortunately, I can't find the thread here, perhaps because of the blog's recent move. A reposting might be helpful.
To date, the Lancet team has (reasonably) declined to provide data at the level which would allow someone to identify individual respondants. But they have also declined to provide any further supporting data. For example, they have so far declined to provide the population totals which they used for the sampling scheme. I believe that their position is that a replication is a waste of time.
So, to date, there is no way for any outsider to be confident that the Lancet results are correct. As far as I know (although I have not confirmed this) even the Lancet peer reviewers did not have access to the underlying data.
The problem here is not so much fraud as it is simple mistakes. The sampling scheme, whatever its other virtues, is non-trivial. Anyone who has coded up complex statistical routines knows that mistakes are all too easy to make, even for the best intentioned and most experienced among us.
Until an outside researcher is able to replicate the results of this study, I will remain less than fully confident in its conclusions.
David:
The thread with data you got from Roberts is here. Thanks, once again, for doing that -- being able to work with the data, even abridged, was quite convincing.
Robert's link was to his data-plot. The thread itself, with a link to the spreadsheet, is here:
http://timlambert.org/2005/12/lancet-study/
Now that I click on Robert's link a second time I get the thread, not the data-plot. Weird. Tim, did you modify the link or has my computer been drinking again?
I didn't modify the link.
Kimble:
Belief does not come into it. It's simply these are the sort of odds I would take if I were a betting man! It is you, Kimble, that wants to believe in spite of the odds.
Tim:
Sorry about the superfluous comments; delete them if you want. (Evidently I clicked on Robert's name instead of his link.)
Soru:
In the linked thread Robert dealt with the claim that without the exclusion of the Fallujah cluster the CI would extend below zero.
A war blogger is surely someone who blogs from a warzone. Like Michael Yon.
Nor would I cast Scott Callaghan as a pro-war blogger, that is a ridiculously narrow description of what his blog covers. A kind of smear. Just as I would not call this blog an anti-war blog if Iraq is only part of the subject material.
If there is any general description of Callaghan's blog it would be "Anti-media-bias-especially at the BBC" blog. Lots of his posts criticise ignorant or heavily biased reporting by the BBC. Particularly the endless criticisms of the US. This thread deals with a Paul Reynolds/BBC article. Reynolds - just like John Simpson - seems to regard Juan Cole as an "academic expert" on Middle East affairs, when Cole has been discredited on many occasions as being heavily biased. The BBC never describes Cole as "anti-war" - but is very ready to describe the other side as "right-wing" or "pro-war". And it was totally wrong of Reynolds to omit any proper reference to the wide criticism of the Lancet death figures and the methodology supporting them. Par for the course for the BBC, however - their reporting is far more sloppy these days, and far more reflective of a pacifist groupthink at the BBC.
dumbisco, warblogger is the term that is generally used to refer to bloggers who are pro-war. They refer to themselves that way. Whether you like it or not, Juan Cole is an academic expert on the Middle East. It is misleading for a report to refer to the criticism of the Lancet study without going into the details of what is wrong with the criticism. Similarly, I would not expect a report that mentions evolution to have to refer to criticism of evolution from Creationists.
Kevin,
My final word (I promise) on the subject.
I am glad we agree that confidence intervals are relevant. Therefore, I assume you agree that reports, such as the BBC's, which fail to include that fairly basic piece of information, are not presenting an accurate picture of the study.
I do find it interesting that you conclude that, all things being equal, one could fairly say the sample size was too small to be meaningful only if the 95% band straddled zero. As you say, if the confidence interval straddled zero, that would suggest some probability that the invasion had the net effect of saving lives rather than costing lives. So, it seems to me that you have decided, without even looking at the study, which results would be meaningful and which wouldn't. That is, if the result shows some probability that lives were saved, then it isn't meaningful. But if it shows no probability of lives being saved, then it is meaningful. That doesn't strike me as a particularly objective, scientific way of approaching the data.
It is also odd that you say the BBC did not claim, as I put it, "that the Lancet determined a death figure of 100,000", but as proof you quote the BBC as claiming precisely that: "The British medical journal The Lancet suggested a figure of about 100,000 back in October 2004." Unless you are quibbling about the difference between "determining" and "suggesting", your claim is simply inexplicable.
Also inexplicable is your claim that "Any reader who took your account at face value would have been misled." As you (accurately) quoted me, I said "In fact [100,000] is simply the approximate mean between an upper limit and a lower limit which bracketed the study's 95% confidence level." There is nothing at all about this statement that "gives the impression that the distribution is uniform." Nothing at all. And it does, at the very least, imply that there is a distribution of some sort, something that is distinctly lacking from the BBC characterization to which I objected. And yet you strangely think the BBC characterization is somehow less misleading than mine.
You further claim that the "true" figure is "more likely" to be in the range of 80k-120k than 8k-48k or 156k-196k. Asssuming there are no other problems with the study (like ability to verify that data collected is reliable), that is certainly true enough. However, "more likely" does not mean "likely". There is a reason that statisticians report figures with a 95% confidence interval rather than an 80%, or a 60%, or a 40% confidence interval. You may, if you want, cite such figures, but at the expense of reducing the usefulness and meaning of the figures even further. The fact remains that the Lancet study, in accord with normal practice, reported a 95% confidence interval of 8,000 to 194,000. If you regard such a wide interval to be providing a meaningful, useful number, fine, but I suspect many others will not, and I don't quite understand your objection to me trying to get the BBC to report it accurately. Unless, that is, you would prefer that people be provided less rather than more accurate information about the study.
Finally, you say that:
"the effects of the invasion were sufficiently drastic to enable us to make pretty robust inferences from the study: mortality rose, the increase was very substantial, the main cause of the increase was violence, particularly the use of heavy weaponry by coalition forces."
First, I think that common sense ought to tell us that, during a war, mortality will likely rise. We don't need a study to confirm that. What was notable about the Lancet study, and what garnered all the headlines, was the reported degree of the rise. But that is precisely what we don't know, and the study hardly helps us out, despite your claim that we can draw "robust" conclusions about it. Again, perhaps your understanding of the meaning of words differs from mine, but showing that the increase in mortality was somewhere between 8,000 and 194,000 lives, (which despite your protests is all the study allows us to conclude even if we accept that there are no other problems with it) is quite a far bit away from a "robust" conclusion.
Thanks for the good holiday wishes. Again, this will be my final word on the subject.
Regards,
Scott
Mr Lambert,
Callahan is not a war-blogger in the accepted sense:
Your evolution-creationism example is ludicrous. Creationists accept on faith that evolution is incorrect. The Lancet Iraq study is doubted because it is merely, as you say, an estimate, albeit a sophisticated one: it should not be cited without qualification; to do so implies it is accepted as factual. Similarly, climate models are approximations that should not be cited without qualification because their results are open to question.
Dumbcisco,
I suggest that you log into the MediaLens website which reveals how pro-war the BBC was and is. Or the latest book by David Edwards and David Cromwell (from MLens), "Guardians of Power: The Myth of the Liberal Media". I have personally written to the BBC on several occasions complaining about the crap they propound that masquerades as news. The Glasgow University Media Group examined the major media outlets leading up to the US-UK aggression and found that only 2% of BBC sources and interviewees were 'anti-war'. This was by far the worst record - even the US television networks, which hardly challenged a single one of their governments serial lies leading up to invasion, weren't as bad. ABC, which was second worst, came in with 7% of its sources as being critical of the invasion.
From Media Lens, February 22:
http://www.zmag.org/content/showarticle.cfm?ItemID=9784
Since the illegal invasion and occupation of Iraq, BBC reporters have bent over backward to legitimize it. It doesn't matter whether its Matt Frei, or Andrew Marr, or Nicholas Witchell or Paul Wood or Ben Brown, time and again BBC reporters have accepted British and American assertions at face value. Some or their reports have been almost nauseatingly sycophantic. If Scott Callahan's web site aims to redress BBC bias, then he could start by challenging all of the BBC-ITN propaganda which normalizes US-UK atrocities and marginalizes dissenting views. He could also start by learning a bit about the propaganda model outlined by Chomsky and Herman in their groundbreaking book, "Manufacturing Consent". Mark Curtis, in his outstanding book, "Web of Deceit", says it clearly when he states that the corporate-state media aims at the "Mass production of Ignorance".
medialens is itself biased. Just look at their targets. Anyone quoting it should mention their bias.
Likewise Juan Cole has a severe anti-Coalition slant - and anti-Israel slant. Anyone quoting him as an "expert" should mention this slant. And for a so-called "expert" he has made some very serious howlers of fact.
Also, right now there is NOT an occupation of Iraq. There is a legitimate government - legitimised by the UN. Overseas troops there are acting in support of the civil power.
And anyone saying that the BBC is coalition-friendly really is nuts. How many times have they looped the Abu Ghraib stuff ? Why did Newsnight this week give 25 minutes to a bunch of ex-soldiers opposed to the war, but would never give airtimne to the far greater numbers of ex-soldiers who support what is being done in Iraq ? Why does John Humphrys act like an attack-dog every time he interviews anyone who is not anti-war ?
I have listened to and watched the BBC for 50 years. I have never seen it so biased, filtering and warping the news.
[Comment deleted. No personal attacks on other commentors, please. Tim]
Scott:
What do you mean by "useful" or "meaningful"? These are words that you are throwing around without thinking about what they mean. The "usefulness" of the information depends on the question being asked. If the question is, how many excess deaths have occurred, then the study is useful (especially given the lack of other studies/information). It says there is a 5/6 chance that the number of excess deaths in the first 18 months was in excess of 70,000.
Statistican use the 95% because of the academic preference for this level. Academics (in their capacity as academics) are in the business of conclusive refutation of hypotheses (amongst other things). Outside the ivory tower, most of the time, it comes down to playing the odds. By your standard GWB should have waited for a study to reject the null hypothesis that Saddam did not have weapons of mass distruction at the 5% level before invading Iraq, which as far as I know, he did not.
Hmmm, so the UNDP study, which seems to say:
The number of deaths of civilians and military personnel in Iraq in the aftermath of the 2003 invasion is another set of figures that has raised controversy. The Living Conditions Survey data indicates 24,000 deaths, with a 95 percent confidence interval from 18,000 to 29,000 deaths. According to the survey data, children aged below 18 years comprise 12% percent of the deaths due to warfare.
is not relevant to this debate, I guess.
Iraq Living Conditions Survey 2004
I continue to search for the actual report.
Oh, the UNDP study is quite relevant,as explained here.
"I just finished reading Michael Crichton's State of Fear, and duly recommend it to anyone and everyone. It was not only a great thriller but also a powerful manifesto debunking one of the great phantoms of our our era - catastrophic man-made global warming."
Perhaps we can now expect a well-reasoned rebuttal of current enviroNazi psuedoscientific propaganda, i.e. Ice Age 2? http://www.iceagemovie.com/
"The methods that they used are certainly prone to inflation due to overcounting," said Marc E. Garlasco, senior military analyst for Human Rights Watch, which investigated the number of civilian deaths that occurred during the invasion. "These numbers seem to be inflated."
This was well and truly hashed over at the time, resulting in:
"Mr. Garlasco says now that he hadn't read the paper at the time and calls his quote in the Post "really unfortunate." He says he told the reporter, "I haven't read it. I haven't seen it. I don't know anything about it, so I shouldn't comment on it." But, Mr. Garlasco continues, "Like any good journalist, he got me to."
"Mr. Garlasco says he misunderstood the reporter's description of the paper's results. He didn't understand that the paper's estimate includes deaths caused not only directly by violence but also by its offshoots: chaos leading to lack of sanitation and medical care. "
http://chronicle.com/free/v51/i22/22a01001.htm
"He didn't understand that the paper's estimate includes deaths caused not only directly by violence but also by its offshoots: chaos leading to lack of sanitation and medical care." Describes about 80% of those who object to the Hopkins study, seems to me.
"Mr. Garlasco says now that he hadn't read the paper at the time and calls his quote in the Post "really unfortunate." He says he told the reporter, "I haven't read it. I haven't seen it. I don't know anything about it, so I shouldn't comment on it." But, Mr. Garlasco continues, "Like any good journalist, he got me to."
"Mr. Garlasco says he misunderstood the reporter's description of the paper's results. He didn't understand that the paper's estimate includes deaths caused not only directly by violence but also by its offshoots: chaos leading to lack of sanitation and medical care. "
Another example, by the way, of how the well known liberal slant of the Washington Post causes it to twist the news to fit its antiBush agenda!!
"Isn't that precisely what the wide confidence interval tells us? A larger sample would have produced a smaller 95% band, and hence would have, by definition, provided more reliable, conclusive data, would it not? "
By definition (more or less) if you provide the relevant CI95 and not just the point estimate, then reliable and conclusive no longer enter into the picture. A CI95 of 0-100 is not more reliable or conclusive than a CI95 of 25-75, it's just more precise. Just increasing n is a study without changing anything else makes it neither more reliable (bias remains bias), nor more conclusive (which is a subjective judgment based on many factors besides precision).
(For the record, I deeply deplore all scientific writing which does NOT provide confidence intervals, both in the popular press and in a discouragingly large fraction of academic publications.)
"I do find it interesting that you conclude that, all things being equal, one could fairly say the sample size was too small to be meaningful only if the 95% band straddled zero. As you say, if the confidence interval straddled zero, that would suggest some probability that the invasion had the net effect of saving lives rather than costing lives. So, it seems to me that you have decided, without even looking at the study, which results would be meaningful and which wouldn't. That is, if the result shows some probability that lives were saved, then it isn't meaningful. But if it shows no probability of lives being saved, then it is meaningful. That doesn't strike me as a particularly objective, scientific way of approaching the data."
????????????? Isn't that the normal everyday interpretation of the CI95 vs. statistical significance? If the CI95 doesn't include zero, the result is statistically significant (at the 95% level of certainty) and you fail reject the null hypothesis, but if the CI95 includes zero, the result is not statistically significant (at the 95% level of certainty) and you merely fail to reject the null hypothesis. You are not mathematically or logically required to state that this is evidence against the alternate hypothesis.
I don't believe I've seen any scientific publications or arguments wherein the fact that the CI includes zero is used as evidence that the mathematical inverse of the alternate hypothesis is correct.
"If the CI95 doesn't include zero, the result is statistically significant (at the 95% level of certainty) and you fail reject the null hypothesis"
The "fail" is silent here.
"Also, right now there is NOT an occupation of Iraq. There is a legitimate government - legitimised by the UN. Overseas troops there are acting in support of the civil power."
Wow.
"Iraq's ruling parties have demanded US forces cede control of security.
Jawad al-Maliki, a senior spokesman of the Shia Islamist Alliance and ally of Ibrahim al-Jaafari, the prime minister, said: "The alliance calls for a rapid restoration of security matters to the Iraqi government."
http://www.egyptguide.net/News/showArticle.aspx?ArticleID=1095
"Jaafari bucks U.S. pressure to quit
IRAQI LEADER WARNS AGAINST INTERFERENCE, DEFENDS TIES WITH RADICAL SHIITE CLERIC
HERALD-LEADER WIRE SERVICES
BAGHDAD, Iraq - Facing growing pressure from the Bush administration to step down, Iraqi prime minister Ibrahim al-Jaafari yesterday vigorously asserted his right to stay in office and warned the Americans against undue interference in Iraq's political process."
http://www.kentucky.com/mld/heraldleader/news/world/14219540.htm
"Overseas troops there are acting in support of the civil power."
I do not recall the left protesting the murderous dictatorship of Saddam as vigorously as they criticise UN-sanctioned presence of Coalition troops in Iraq.
And if you want a critique of the appesement mindset on the left that is trying to re-write the history of Iraq, try this historian's view :
http://victorhanson.com/articles/hanson033106.html
Hanson is correct on every single point - but the unholy alliance of the left and the Islamists/Baathists of Juan Cole who has been shown to lie to advance his case.
I notice how the Victor Hanson piece fails to provide references to evidence for any of it's assertions.
dumbisco: "And for a so-called "expert" he has made some very serious howlers of fact."
Perhaps you could, then, give a list of Juan Cole's "howlers"?
Hanson is correct on every single point - but the unholy alliance of the left and the Islamists/Baathists of Juan Cole who has been shown to lie to advance his case.
Poor VDH - still clinging to his wargasm fantasies like other wargasmers clung to their mommies when they and their Cheeto-stained boxers were kicked out of the basement and on to the street.
Really, how can anyone take a commenter seriously who writes comments like this - full of ululating false equivalencies, strawmen and FUD phrases?
Best,
D
Beck:
Merely an estimate? Well may be you would be kind enough to ask the coalition forces for an exact number? Good luck!
All academic studies have weaknesses, ego most mention the numerous weaknesses & areas for future study at the end. To list these all of these every time a study is referenced would take all day. More to the point to note a qualification that alone would most likely lead to misinterpretation is ludicrous.
Meyrick Kirby
The evidence to support Hanson's arguments are all over the net. You just read the wrong, blinkered sites. Try reading, for example, some of the stuff that has been released this month from the captured Iraqi documents. Try reading the Congress and UN debates about what people worldwide believed was the state of WMDs in Iraq.
By the way, apropos the estimates of deaths in Iraq, I expect you will concede that most have been committed by Iraqis on Iraqis - not by Coalition forces.
And what is your estimate for the annual number of Iraqis killed by Saddam - and the annual average of killings of citizens of neighbouring countries by Saddam ? Millions over the past 2 decades ? I expect you will say that is immaterial.
Allow me to take umbrage at the assumption that those who opposed the current Iraq war supported Sadaam Hussain at any point. Those who hate America often make this claim.
"I do not recall the left protesting the murderous dictatorship of Saddam as vigorously as they criticise UN-sanctioned presence of Coalition troops in Iraq.'
which may be a reflection on your memory rather than on the left.
Tell me do you remember Dick Armitage. former Assistant Secretary of Defence in the current US government, lobbying his former colleagues in the bush sr. administration against sanctions on Iraq in the months AFTER the Hallabj massacre?
Or perhaps, if there is more criticism of the current occupation it has soemthign to do with:
a. the fact that our tax dollars weren't funding Saddam;
b. the fact that a significant minority (on current polling) of our countrymen weren't actively applauding and promoting Saddam's actions at every opportunity (not after the end of the Iran-Iraq war anyway);
c. the fact that as patriotic citizens of liberal democracies we expect more of our governments than of third world dictators (foolish, I know);
d. the fact we didn't have friends or family members who were placed in direct physical danger as a result of Saddam's actions; or
e. the fact that our criticism of the current occupation is more likely to be effective?
Dumbcisco:
In that case why have you failed to include a single reference/link?
The particular causes are not the point; I fail to see how an Iraqi killed by an coalition solder matters, but an Iraqi killed by another Iraqi doesn't matter. The point, as you know but are ignoring, is that the invasion has made things worse for the average Iraqi not better. Triggering a civil war is nothing to boast about!
I would say they are extremely relevant, but are you refering to those killed when we were supplying him with weapons & military training, or after?
Perhaps you could, then, give a list of Juan Cole's "howlers"?
How about this then. Juan Cole: "According to the September 11 Commission report, al-Qaeda conceived 9/11 in some large part as a punishment on the US for supporting Ariel Sharon's iron fist policies toward the Palestinians. Bin Laden had wanted to move the operation up in response to Sharon's threatening visit to the Temple Mount, and again in response to the Israeli attack on the Jenin refugee camp, which left 4,000 persons homeless. Khalid Shaikh Muhammad argued in each case that the operation just was not ready."
Also here.
No list required. JC is a bad joke and so is the BBC who idolize him.
Gee max, is that the best you can do? A minor mistake that Cole corrected? And I note that your source, Martin Kramer doesn't seem to understand statistics either, endorsing an innumerate criticism of the Lancet study.
In the linked thread Robert dealt with the claim that without the exclusion of the Fallujah cluster the CI would extend below zero.
True, but, unless I missed something, mainly by using asequence of long and technical words to express the message 'it's strange and probably wrong, so lets quietly ignore it'.
Maybe a valid argument could be made along those lines, but the fact that the author of the paper didn't choose to mention the need for it doesn't exactly increase my trust in their findings.
Can someone explain why someone looking for an honest picture of Iraq would use the Lancet study, and not the UN one? It uses standard methodology, surveys all provinces with a massive sample size, and above all, actually answers the question asked, i.e. how many people died as a result of the war.
Mixing up casualty figures with the seperate issue of public health, right down at a low level so they can't be disentangled, just seems wrong.
If there had been a cholera epidemic in Basra in 2002, as there was in 2003, and in many other years under saddam, the Lancet figures would be rather different, maybe even to the point they could be used by right-wingers to claim it justified the war.
Knowing 'x' people died violently, and the impact on public health has been 'y' deaths per year, seems immensely more useful for meaningful discussion than only knowing 'x-y'.
Quite possibly there have been enough improvements to clean water provision over the last two years that now a repeat of the Lancet study would find a lower number, not a higher one. To what extent would that be useful evidence the war is going well?
Meyrick Kirby
If you want education on Victor Hanson's arguments, go read his artcles. He provides all the references you need.
It is debatable whether things are worse now in Iraq than before. I have seen estimates that there are a NET 100,000 less deaths per annum now than before - because Saddam was killing people at a ferocious rate, check out all those mass graves, and because Saddam's corruption of the Oil-for-Food programme meant that he starved his populace of food and medical aid.
And please don't trot out the old chestnut about the US supplying Saddam with weapons. A huge amount of his ordnance was from Russia, with the French following close behind, The US supplied less than 2% over the past couple of decades.
Meyrick Kirby
You might find it very worthwhile to study the essays and also the books of Victor Davis Hanson. He is an historian with a respect for facts rather than hyperbole. Not a chump like Juan Cole whose reputation for veracity has been exploded.
Dumbcisco is acting as an amplifier of the campaign to smear Juan Cole, which I am sure is unconvincing on this blog because he offers no evidence whatsoever. But it is worth reminding people that Cole is a major commentator on the Middle East, who does speak the language and is a significant adversary for the Right. For some of Cole's remarks about the campaign, go to this and here. There's a lot more, particularly about the attempts to claim he doesn't speak Arabic.
Soru wrote about CIs:
True, but, unless I missed something
You missed something.
Can someone explain why someone looking for an honest picture of Iraq would use the Lancet study, and not the UN one?
The purpose of the ILCS was to focus on living conditions, not mortality. There were only a few questions about mortality (see the household questionnaire); most importantly, they did not collect the date of death. That means you can't calculate separate pre- and post-invasion death rates. Nonetheless, it's not like the ILCS and the Roberts study contradicted each other--the ILCS findings are consistent with and reinforce the Roberts findings.
Dumbcisco, I suppose I could spend time going over your comments in detail, but I'd be wasting my time. I note you still haven't linked to sources to justify any of your numerous essertions.
Meyrick Kirby
You will find all the support you require for Victor Davis Hanson within his own writings. It will take you 5 seconds to google them. Or just go to his website - which by the way affords you the chance to ask questions :
http://www.victorhanson.com/
But let's be fair - you won't read him, will you ? Because he clearly has a different set of beliefs to yours. However, he is a published historian, you can check out on amazon.com that his many books average 4 stars on the many hundreds of reviews that Amazon cite. Just type his name into the search box and read them for yourself.
I have followede politics including US politics for probably a couple of decades longer than you, and I have worked at the heart of the UK governmental machine. I find Hanson factual as well as persuasive. If you broadened your view, you might too.
Incidentally, I note that your favoured Juan Cole has a solitary political book mentioned at Amazon. It garners 2 stars and just 4 reviews. Doesn't sound much of a scholar to me - try reading those reviews for a slapdown. That is why it is offensive for the BBC to keep ramming Cole's opinions as an "expert" at us. He is about as reliable as Robert Fisk - that is, he has an agenda.
Meyrick Kirby
You asked for a few references on Juan Cole's howlers. The web is chock-a-bloc with them, so I have selected just a few.
Here is Juan Cole informing us that the London bombers were not home-grown :
http://www.juancole.com/2005/07/update-on-london-bombing-investigation…
His piece is headed "Informed Comment" See why the man is a joke ?
And try these comments on his inaccuracies and bias, and the way he tries to cover up his errors :
http://obsidianwings.blogs.com/obsidian_wings/2005/07/juan_cole_and_h.h…
http://decision08.blogspot.com/2005/07/weekly-jackass-number-thirty-fiv…
http://www.windsofchange.net/archives/007187.php
http://elephantsinacademia.blogspot.com/2005/07/honoring-one-of-our-own…
http://foreigndispatches.typepad.com/dispatches/2005/04/juan_cole_errs_…
Oh, and an amusing glitch by this so-called "expert" was on 1 April last year, when he headlined a piece "The Government in Iraq is not even close to being formed". Two days later Reuters reported "Iraqi politicians elected a Sunni to be speaker of the Parliament on Sunday, ending a political impasse and taking a decisive step towards forming a government nine weeks after the historic elections."
In other words, the headline by the biased "expert" on whom BBC folk choose to rely was proved wrong before the ink was dry on his statement.
I expect you will want a reference - try this :
http://www.windsofchange.net/archives/006606.php
Here are some more references to Cole's clumsy errors - which many presume to be based on his bias and spleen against Bush :
http://www.jihadwatch.org/dhimmiwatch/archives/007322.php
And finally - can you remember when Cole made a truly spiteful and scurrilous attack on the reputations of a front-line journalist Stephen Vincent and his translator when they were killed in Basra last year. Stephen Vincent's widow summed up Cole's character to a T :
http://www.murdoconline.net/archives/002697.html
http://jarrarsupariver.blogspot.com/2005/08/juan-cole-takes-cowardly-sp…
http://www.samefacts.com/archives/the_war_in_iraq_/2005/08/note_to_juan…
http://www.murdoconline.net/archives/002713.html
But I suppose you will continue to treat people like Cole as the font of all wisdom on ME affairs ?
Actually, the most interesting part of the Hanson piece is the way he insists that there was a genuine threat from Iraqi WMDs. Someone who can't even acknowledge that Iraq did not have any WMDs should not be taken seriously, ever again.
Tim Lambert
All the intelligence agencies worldwide felt there was a WMD threat back in 2003. Even Blix came up with 170 pages of unanswered questions. There is still no clear answer as to whether WMD materials were moved out to Syria during the build-up to war - and a good deal of evidence that this was possible.
What intelligence agencies believed is irrelevant. Hanson contends that there was a genuine threat, not that some people believed that there was one. And there is a clear answer as to whether Iraq moved all their WMDs to Syria -- it's no. Hanson is delusional.
Mr Lambert,
The only one who's delusional is you. Hanson says it is not true that there was no real WMD threat. He then elaborates. He does not assert that Iraq had WMDs immediately prior to, or at the time of, invasion. It's not that hard to understand, really.
>There is still no clear answer as to whether WMD materials were moved out to Syria during the build-up to war - and a good deal of evidence that this was possible.\
There are a great many things that are possible.
For example it's possible that George Bush colluded in flying the WMDs to North Korea in exchange for several billion dollars and a liftime supply of hard drugs and pre-teen sex slaves.
It just isn't very probable.
You are surely misquoting Hanson in saying that he contends that there was a genuine threat, not that some people believed that there was one. And your use of the term "some people" is quaint, to say the least. The worries were shared by intelligence agencies worldwide. And to suggest that their views were irrelevant is ridiculous. And what about Blix's 170 pages of unanswered questions ?
Try reading what Hanson actually said on the WMD point :
"Second, we know now that worries over Iraqi weapons of mass destruction were both justified and understandable. Postwar interviews with top Iraqi generals reveal that Saddam’s own military assumed that his stockpiles of WMDs were still current; confirming the intelligence estimates from Europe and most of the Arab world.
In addition, Iraqi arsenals of WMDs, in the judgment of both the Clinton administration and the United Nations, were still unaccounted for in March 2003. And even if the stocks were moved or destroyed, the prerequisites for the rapid mass-production of biological and chemical agents ; petrodollar wealth, scientific expertise, alternate-use facilities, and a will to produce and use them were met in Saddam's Iraq."
Also, Hanson does not assert that all the weapons to Syria. Whatever happened to them, Saddam retained the ability to re-start his WMD programmes and had not complied with 1441.
So - it is delusional to say that Hanson is delusional.
Ian Gould
You suggest that it isn't very probable that Saddam would collude with another Ba'athist dictatorship in concealing highly-mobile WMD's. That sounds a touch naive.
The jury is still out on that. Some of his own senior military have suggested that stocks were moved to Syria. As further captured Iraqi documentation is released we may get a clearer view. In the meantime Syria is hardly a helpful witness.
(If I said my boss had hidden some stolen goods in a field, would the police be sensible to ignore this as it was just my story ? Because that is what you seem to be doing.)
And anyway, whether WMDs were moved or not, the main issue is whether AT THE TIME there were genuine perceptions worldwide, fostered in part by Saddam's own actions, that he still had WMDs that were unaccounted for. In leaving those perceptions in the air, Saddam had failed to comply with UN Resolution 1441.
Hanson claims that this is wrong: "There was no real threat of Iraqi weapons of mass destruction." The worries about Iraqi WMDs were not justified, because they didn't have any. And are you now trying to argue that they just haven't been found yet??
The worries about WMDs were real at the time, not fabricated. You do not dispute that intelligence agencies worldwide believed he had them. Nor that Blix produced a 170-page list of unanswered questions.
Whether or not he had actually destroyed them all, he retained the ability to restart his WMD programmes quickly. And he had used WMDs on his own people as well as neighbouring nations. If you don't regard that as a threat, neither is swimming with sharks.
The worries about WMDs were justified AT THE TIME. You are trying to use 20/20 hindsight, which is a false approach to what was the general perception back in 2003. I don't recall any senior politicians in the US or UK saying flat-out that Iraq had nothing. Even the opponents of Bush and Blair concurred with the worries about WMDs.
http://www.nysun.com/article/26514
Interesting if very hostile review of the Lancet death estimates :
http://www.logictimes.com/civilian.htm
"You suggest that it isn't very probable that Saddam would collude with another Ba'athist dictatorship in concealing highly-mobile WMD's. That sounds a touch naive."
No more naive than the assumption that because Iraq and Syria were both nominally Ba'athist they'd put aside decades of animosity with included attempts by the Syrians to both assassinate Saddam and overthrow his government and similar attacks on Syria by Iraq.
Given their history of antagonism, Syria's overt support for Iran during the Iran-Iraq War and of the UN forces during the Gulf War, I find it highly unlikely that a notable paranoid like Saddam would entrust the son of one of his most bitter enemies with absolute proof of his guilt.
dumbcisco, the NY Sun link is to someone telling a fanciful story because he has a book to sell. If WMDs had been transported to Syria there would have been some concrete evidence.
The logictimes stuff is just stupid. You seriously can't see what is wrong with their analysis?
Ian Gould
Your mind is obviously closed if you reject contemptuously the first-hand evidence of a senior military figure in Saddam's "court" just because he has published a book. He hardly earn much money from his book - and he is taking a lot of risk in publishing it, in talking openly about what he knows. It is blind of you to dismiss his evidence out-of-hand - without even knowing what he said.
Likewise you duck any discussion of the logictimes stuff, including the extensive comments on it. That too looks like a closed-mind approach to debate. And illustrates the general point that opponents of the Coalition's actions have a tendency to push high figures. Just as you are commend the analysis by Davies, an obvious polemicist - which includes the ludicrous high estimate of 500,000 deaths.
Which is where I came in. You attacked Scott Callaghan as a "war blogger", a silly and inaccurate epithet. He blogs against the anti-US and anti-Bush bias in a lot of the UK media - especially the BBC. On the Lancet study, all he was asking was that the BBC should state openly that there is a lot of dispute about them. An indisputable point.
You appear partisan in your view of the Coalition action, so I simply don't trust anything you say about death statistics. You are not a statistics expert any more than I am a deep-sea fisherman. Your blog, like most blogs, grinds axes. Just like I do. Can't we sometimes be honest and admit we all have bias ?
dumbcisco, My name is Tim Lambert, not Ian Gould.
That's the first time anyone has accused me of ducking discussion of the Lancet study. I have 80 posts on it and there have been upwards of 2000 comments on those posts. I've taken apart the vast majority of criticisms of it that have appeared on the web. I saw the logictimes one and decided that it wasn't worth bothering it. If you ask nicely I'll explain what is wrong with it, but you should really try to figure it out yourself.
It is misleading for the BBC to undercut the Lancet study by reporting bogus criticisms of it. Do you think that every time they report about evolution they should mention that Creationists dispute evolution?
Tim Lambert
Sorry about misaddressing you.
I did not accuse you of not discussing the Lancet study. I accused you of not discussing the article I posted and the comments that had been made on it. You confirm that you swept it aside. That looks like a closed mind to me.
It really does not matter how many times you have posted on the Lancet study. You are not a statistician so why is your commentary worth much - no matter how many times you repeat it ? Competent statisticians have criticised it - and their comments matter far more than yours.
It is ridiculous for you to say that all criticisms of the Lancet study are "bogus". People with real competence have disputed the Lancet methodolgy and the conclusions. Others have pointed out the fact that it is normally impossible to get anything published in the Lancet without peer review - which did not take place on the Lancet findings. I know how hard one of my children found it to get Lancet publication for highly original MRI research with detailed physics findings supported by an eminent UK statistician. It took over a year of peer review befor publication.
It was also asked how such laxity about peer review for the lancet study just happened to occur under a left-leaning editorial staff at the Lancet.
You will continue to cleave to the Lancet stuff. Others won't. The furore over the Lancet study should have been at least mentioned by the BBC - its findings should not be presented as Gospel. There should have at least been a qualifying phrase - eg "which some have challenged" That was Scott Callaghan's point, and it remains valid.
Tim Lambert
To repeat - you contemptuously rejected the first-hand evidence about WMDs from the second-in-command of Saddam's air force - solely because he had written a book. That marks to me a closed mind.
You are not a statistician so why is your commentary worth much - no matter how many times you repeat it?
Well dumbcisco, if that's your idea of an argument, it seems fair to ask you what, if anything, you think your comments are worth? How do you distinguish between "people with real competence" and hacks? You assert that the paper by Roberts et al was published without peer review, so it is clear that you are not even informed about the facts of the case, never mind the statistical arguments.
Kevin
If a competent statistician says that an article is not worth attention I would listen to him. If Tim Lambert says it is not worth attention I see a closed mind, not a competent comment.
Peer review was rushed and somewhat limited in Lancet case, I submit.
The central point remains - the BBC should not blithely quote the 100,000 figure without qualification.
Kevin
On statistical matters I would agree that my comments are not worth any more than Tim Lambert's or most of the commenters here. That is - not much. And I get very suspicious when there are attempts to elevate the widely-questioned Lancet stuff against any other estimates.
Peer review was rushed and somewhat limited in Lancet case, I submit.
I'm willing to consider your submission. Do you have any evidence for it?
And I question anyone who still regards Juan Cole as an "expert" as against a polemicist. I was asked to produce some of Cole's serious errors of fact - his "howlers" and did so, without challenge.
I am not a Lancet subscriber so I cannot access the publication. But I asked my daughter about it at the time, as she had been put through the peer review mangle for a very long time to get publication of her MRI research which was co-signed by a bevy of Professors and an eminent UK statistician. Her MRI method of diagnosis has subsequently led to a 75% reduction in UK death rates in the affected population of patients. She was amazed how quickly the Iraq death study - hardly a usual topic for a medical journal like the Lancet - had been published. And she is anti-war. She felt this was a clear case of editorial bias.
dumbcisco, The NY Sun article does not say that Sada had first-hand knowledge -- he is relaying a story he claims was told to him by others. Sada was 2inc of Iraq's Air Force but was fired in 1986. He wasn't in Iraq before the invasion but seems to have been one of the Iraqi exiles. It looks like the NY Sun is the only paper to have taken his claims seriously. Here's what the Washington Times said:
dumbcisco, I didn't just say that those criticisms were bogus, I went into the details in all those posts. You don't seem to have been able to challenge any of them. The fact that you can't see what is wrong with the logictimes post suggests that you are not competent with statistics.
Tim Lambert
So, we have a very senior military man who publicly claims at personal risk to have been told by Iraqi pilots that they flew WMDs to Syria. And you now add that we have the former head of the Iraq Survey Group saying there were promising leads that weapons had been transported to Syria but it was impossible to pursue them owing to security risks.
On the other hand we have an unnamed Washington Times source "close to the Iraq Survey Group" - whatever that means - who claims "we were unable to find anything we could verify". He does not say that anything was disproved.
On that basis I would say the jury is still out on whether WMDs were transported.
Other US media have paid attention to Sada - I have seen him being interviewed on TV for example. He was not in exile from 1986 onwards. And he is in Iraq now. To dismiss out of hand the views of a senior miolitary guy is still, I think, evidence of a closed mind.
On criticisms of the Lancet stuff, you specifically said "It is misleading of the BBC to undercut the Lancet study by reporting bogus criticisms of it."
You surely cannot say that all the criticisms of the Lancet study were bogus ? Now that truly would indicate a closed mind.
We who pay the enforced-on-pain-of-jail BBC licence fee want balance and we want facts. It is highly pertinent that there were serious criticisms of the Lancet study. It is not to much to ask for the BBC to mention that the Lancet study had been strongly challenged, which the reporter knew full well. And it certainly looks to be misleading to simply quote the 100,000 headline-grabbing figure after all the challenges to that trite figure.
I remain unimpressed by your apparent claims to particular statistics expertise. Just because you have defended the lancet study up hill and down dale does not give you any expertise. Indeed, I am pretty confident that I have as much competence in statistics as you have - that I have had as much stats training as you, and certainly that I have used and published more statistics while in Government service than you have had hot dinners. Your expertise in Open Source or other IT issues does not translate to stats expertise.
I submit that you are cleaving utterly to the Lancet figures, through thick and thin. I stop short from using the word obsession, but hobby-horse springs to mind. It seems odd that you should be attacking an obscure UK blogger who was not himself criticising the Lancet study but merely commenting on the way the BBC had referred to it.
You are not a statistics expert, you are simply a commentator like most of the rest of us, with no special expertise. So when you sweep the logictime article and the comments on it aside, it still indicates a closed mind to me.
I am not a Lancet subscriber so I cannot access the publication.
You don't need to be. It has been freely available online since it was published. Even among the bloggers who have written scathing critiques, there are many who actually have read it (albeit not very carefully in most cases).
She felt this was a clear case of editorial bias.
Even if she is right, that would only mean that the editor wanted to publish quickly. Your contention is that the referees (who are not usually the sort of people an editor wants to try pushing around) rushed their work.
It is entirely likely that the editor of the Lancet wanted to publish quickly. Axes to grind.
Also, there is peer review and there is peer review. I know of medical articles where the referees were all hostile, favouring the established party line. That sort of technique can block publication of valid research. And use of compliant referees can have the opposite effect, allowing dubious stuff through.
Meanwhile, is there still no chance of replication of the Lancet study ? Non-replicability is normally a bar on publication as it strikes against the essence of scientific method.
Dumbcisco wrote:
If a competent statistician says that an article is not worth attention I would listen to him.
I claim expertise in this area, and on my good days my colleagues consider me moderately competent. I can't swear that there aren't competent statisticians somewhere in the world who think the Roberts article isn't worth attention, but on the whole I'd say that there is roughly about as much agreement on the methodological issues involved in the Roberts article as there is, say, among climate scientists on global warming.
Accelerated review is unusual but it's not necessarily a sign of shoddy quality. I rarely spend more than a few hours reviewing a paper of the length and complexity of the Roberts' article. Accelerated review simply means that I spend those hours sooner rather than later because the paper goes to the top of my to-do list rather than the bottom. I don't cut the depth of my review, and I'm pretty sure the recipients resent my comments just as much whether the review was accelerated or not.
Yes, dumbcisco, there is peer review and there is peer review. Likewise, there are allegations and there are allegations. The ones with evidence to support them are generally more interesting. Alas, yours are not of that sort.
But now that you know that the paper is available to all, why not apply your stats training to it? You know, the training that you reckon is at least as good as Tim got in his B Math course etc.
Dumbcisco claimed:
Non-replicability is normally a bar on publication as it strikes against the essence of scientific method.
I've never in my life received editorial instructions that non-replicability should be a bar to publication. That would exclude from publication any non-laboratory research. I do not think you know what you're talking about, and I claim special expertise in this area.
Kevin
My MSc and direct Government experience using lots of stats trumps his BSc. Yah Boo.
The fact is that the publication of the Lancet study was extremely rushed. Didn't the authors insist on early publication - before the US election ? Or was that a rumour ?
The central point remains - the BBC should not still be waving around the 100,000 figure without qualification.
Well, dumbcisco, if you are such an expert in statistics and if expedited peer review caused the reviewers to miss something, how about you tell us what it was? Put up, or shut up.
Dumbcisco yah-boo'd:
My MSc and direct Government experience using lots of stats trumps his BSc. Yah Boo.
I believe Tim has a PhD. Plus, unlike you, he's actually read the article.
Robert
So he has a Ph D. Good for him. But his qualifications are not in statistics.
I did read the article at the time, and have read a lot of criticisms. I am amazed that people should still be promoting it as if it was gospel truth, and the vigour with which they attack any dissenting views. Especially people with little or no qualification to comment.
Tim Lambert
The central point which you keep avoiding is that the BBC should have alluded to the fact that the study was very controversial, had been seriously challenged. They should not wave around the bald figure of 100,000 deaths - a point which again you have avoided.
You presume you are qualified to comment ad nauseam on statistics, I don't. But you sidestep any commenting (other than abuse) on challenges to the study which you have spent so much time championing.
And I continue to note that you jumped to the defence of Juan Cole but have avoided commenting on his egregious errors and his heavy political bias.
I will walk away convinced that your mind is closed. Your original attack on Scott Callaghan's piece looked political, and still does.
Well dumbcisco, you've certainly failed to impress anyone here. You've blustered endlessly about how the study has been seriously challenged, but you've been unable to state what these serious challenges are.
I am not interested in impressing you. I was simply interested in seeing responses to criticism of the way the BBC used the Lancet study and the 100,000 figure without qualification, and their repeated use of Juan Cole as an "expert". You have failed any reasonable test on both. Even the BBC reporter involved in the Scott Callaghan piece had the grace to say that he realised there were serious issues about the Lancet study. You still appear to say there are no serious issues. Like I said, an indication of a closed mind. From which I make the perhaps unreasonable jump to political bias, axes to grind.
So let me get this straight, dumbcisco. The BBC should have mentioned that there were serious issues, but you can't tell us what these issues that they were supposed to mention are. Do you think that whenever they mention evolution they should mention that Creationists don't agree with evolution?
"I do not recall the left protesting the murderous dictatorship of Saddam as vigorously as they criticise UN-sanctioned presence of Coalition troops in Iraq. "
Depends what you call "the left". Do you recall Congress opposing the sale of helicopters to Saddam in 1983 but being overruled? Do you recall the Senate passing Senator Pell's (Rhode Island) Prevention of Genocide Act which would have ended all U.S. aid to Saddam in September 1988, but the act being killed in Congress under Reagan White House pressure?
"Non-replicability is normally a bar on publication as it strikes against the essence of scientific method."
Uh, as much as this may be true, the question is not re the author can repeat the study, but can anyone else; and the lack of motivation of other authors to do so is generally not considered a "strike" either.
"Mixing up casualty figures with the seperate issue of public health, right down at a low level so they can't be disentangled, just seems wrong. "
Why? In one post, opponents argue that the Lancet shouldn't have published the study as it's not really medical, now here the argument is that public health shouldn't enter into it.
When your children die because a war has thrown your country into chaos, it's not that much of a comfort that they died of dysentry which would have been easily treatable under alternate circumstances.
Having got round to reading some of your links Dumbcisco, I can help but notice that Prof. Cole does admits to his mistakes, unlike to certain scholar at the AEI:
source
I also found your following comment missed the point by a wide margin:
One step towards forming a government does not amount to forming one, as demonstrated by the fact the Iraqi parliament is yet to form a government. source
I also found you're own research to be somewhat lacking. For instance:
May be you should look beyond the comments of US residents: link
Also, you wrote:
No, I don't. Come to think of it I'd never heard of Juan Cole before this thread.
Lastly, I found this comment really funny:
Kimble made the same accusation above, yet it was he who choose to believe despite the odds. Plus, not that you're going to believe me here, but I'm considered by many in my area of research, including by a BAA distinguished academic, to be somewhat right wing; somedays I just can't win!
I also had a look one of your references that tries to rubbish the Lancet study link. It seems to make 2 points. Firstly that the Iraq Body Count has a far lower figure than the Lancet study, but as been pointed out numerous times on this blog, they measure different things. Secondly that most of the post-invasion death on IBC are by insurgents/terrorists, not coalition forces. So what? As I noted above, starting a civil war (or getting close to it) is nothing to boast about.
As far as I can tell all challenges, such as those found in the LogicTimes article, have been answered. Instead of strutting out your qualifications, maybe you could say what is wrong with the Lancet study?
I would lastly note that your ad hominem attacks on Lambert are getting pathetic. Lambert does have a degree in maths, which I suspect includes a fair amount in stats. And since I've taken courses in quantitative methods and econometrics, I'm not entirely ignorant on the matt myself, although I admit I'm no maths/stats grad. Far more to the point though is that Lambert has gone in to detail as to why the likes of the LogicTimes piece are wrong. In other words I suggest to you that it is better to judge people by the merits of their arguments, not their qualifications.
"All the intelligence agencies worldwide felt there was a WMD threat back in 2003. Even Blix came up with 170 pages of unanswered questions. "
Hardly.
"The intelligence is practically non-existent," one exasperated American intelligence source said. Most of the intelligence being used to support the idea of a link between al-Qaeda and Mr Saddam comes from Kurdish groups who are the bitter enemies of Ansar al-Islam, he said.
"It is impossible to support the bald conclusions being made by the White House and the Pentagon given the poor quantity and quality of the intelligence available.
"There is uproar within the intelligence community on all of these points, but the Bush White House has quashed dissent and written out those analysts who don't agree with their views," the source said.
"This could all be dismissed as a turf war between intelligence agencies were it not for the near unanimity across the British and American intelligence communities, including the Defence Intelligence Agency analysts whose bosses produced the line the White House wanted to hear.
"The Secretary of State has to win over the three permanent members of the Security Council who might veto a second resolution on using force against Iraq - France, Russia and China. (US goes soft on Iraqi link to al-Qaeda, The Age, February 5 2003)
"Of all the claims U.S. intelligence made about Iraq's arsenal in the fall and winter of 2002, it was a handful of new charges that seemed the most significant: secret purchases of uranium from Africa, biological weapons being made in mobile laboratories, and pilotless planes that could disperse anthrax or sarin gas into the air above U.S. cities. By the time President Bush ordered U.S. troops to disarm Saddam Hussein of the deadly weapons he was allegedly trying to build, every piece of fresh evidence had been tested--and disproved--by U.N. inspectors ... The work of the inspectors--who had extraordinary access during their three months in Iraq between November 2002 and March 2003--was routinely dismissed by the Bush administration and the intelligence community in the run-up to war, according to the commission ..." (Washington Post, on the findings of the 9/11 Commission)
"After three months of intrusive inspections, we have to date found no evidence or plausible indication of the revival of a nuclear weapons program in Iraq." (Dr. Mohamed ElBaradei, the chairman of the International Atomic Energy Agency)
"The activities we have detected do not ... add up to a compelling case that Iraq is currently pursuing what INR [State Department's Bureau of Intelligence and Research] would consider to be an integrated and comprehensive approach to acquire nuclear weapons." (State Department comment on National Intelligence Estimate WMD report)
"In the Middle Ages when people were convinced there were witches they certainly found them." (Hans Blix, on the evidence for WMD)
Dumbcisco:
You haven't named your list of "competent statisticians" who say that the study "is not worth attention". You haven't listed any of the statistical issues that you claim make the article disputable. When you thought Tim had a BSc, you tried to trump him with a MSc. When Tim was revealed to have a PhD, you say degrees aren't relevant. You claimed that non-replicability is a bar to publication, when it is not. You say that acclerated review is proof of editorial bias, when it is not. You claim that someone is unreliable because he was shown to be wrong in the past, but you excuse others for their belief in WMDs because "everyone" thought it was so in the past.
Is this about it? 'Cuz I'm startin' to think that maybe you'd want to work on your argument a little. Just sayin', is all.
"Accelerated review is unusual but it's not necessarily a sign of shoddy quality. I rarely spend more than a few hours reviewing a paper of the length and complexity of the Roberts' article. Accelerated review simply means that I spend those hours sooner rather than later because the paper goes to the top of my to-do list rather than the bottom. I don't cut the depth of my review, and I'm pretty sure the recipients resent my comments just as much whether the review was accelerated or not."
Amen.
Daniel Davies put it well:
"the time taken for peer review is determined by academic procrastination above all other factors. Every academic paper could complete its peer review very quickly if the reviewers got their finger out because they thought it was important. The suggestion that people are trying to make here is that reviewers for the Lancet usually spend six months humming and hawing over the data, to the exclusion of all other activity, and that this process was short-circuited by politically motivated editors wanting to rush something into print without anyone having a proper look at it. No such six month scrutiny ever takes place, and this objection is also Simply Not True"
'Why? '
For the obvious reason that the Lancet method of analysis would quite likely have concluded that signing the 1918 Armistice was a bad idea.
Tim, I tried to post a comment yesterday, but it still hasn't come through?
I think it has. Unfortunately, the order is based on when you posted it, not when I approve it.
Oops, sorry, my mistake.
"The central point which you keep avoiding is that the BBC should have alluded to the fact that the study was very controversial, had been seriously challenged."
I agree. Interesting thread. At the very least, there are questions about the Lancet Study. It should not be presented by a reputable media source as fact.
"Your original attack on Scott Callaghan's piece looked political, and still does."
Sorry, but he's right. Callahan is not a pro0-war blogger. I've read his blog and it's about the biased media coverage directed at the US, on a number of subjects. It is not a pro-war blog. There's no point in mischaracterizing it.