A Reader's Report: Deeper into the Lancet Numbers

Thousand of articles and blog-postings worldwide have already been generated in response to the so-called “Lancet Report” which suggests 100,000 Iraqi war deaths.

Let’s add one more. One of our most intrepid and industrious blog commenters – none other than the now infamous GMroper – who stirred it up here last week with his profiling of both John Kerry and GW Bush -- is back with another provocative contribution.

I want you to take a few minutes to read his entire argument which I have saved to a file and that you can read by clicking here. GM goes farther than the Fred Kaplan piece I posted yesterday in deconstructing and really demolishing the figure of 100,000 deaths contained in the Lancet Report. As Kaplan also noted, the real figure based on the report’s methodology might be as high as 198,000 or as low as 8,000 or maybe much lower or higher. Indeed, GM figures that there's 1/10th of 1% per cent chance that it is 100,000.

Says GM of his analysis-- which he has written up just for this blog:

"Two caveats should be noted here: (1), Figures do not lie, but liars figure. In addition, (2) there are three types of lies: Lies, Damn Lies and Statistics."

Reading GM you’ll see he’s a very handy statistician. And, while like most of us, GM has made his own political views quite open on this blog (GM’s one of our resident Bushies), his deconstruction is not ideological or political. Nor does it write off the crucial importance of what the actual number of war dead might really be:

"The loss of life for these people is horrendous enough without a biased and frankly partisan report sent out with obvious efforts to effect an election. Should you doubt that, please consider the following. (1) they claim peer review, but that was done in only a few weeks, I have talked to many colleagues who tell me peer review is seldom done in less than 6 months and often it takes a year. (2) Les Roberts, the principal author of this article insisted that it be published before the election, ostensibly so that both candidates would make comment about the need to protect civilians and plan to do so. Hogwash, only one of the two candidates has a chance to do that and that is the winner of the election. Publishing the results after the election would have given the winner then necessary message. And lastly, (3) Les Roberts was a vocal critic of the war in the first place, do you really think he didn’t want to “effect” the election process and used horrendous numbers to do so? Lies, damn lies and statistics indeed!"

I also want to go out of my way to clarify why I have now done two postings on this issue. I feel compelled to offer the explanation after I have been chastised by the politically correct (and the morally impaired) for, say, giving Fidel too hard a time without giving “equal time” to offenses committed in Haiti.

I have done two postings on this precisely because I oppose the war in Iraq and because I am rather disgusted with the way the administration not only bans photos of our own war dead but also doesn’t bother to publish some estimate of the Iraqi civilian dead.

This is an unacceptable standard – one that serves as an example of how not to do things. Therefore, IMHO. tossing around unreliable, imprecise and flawed tallies of the victims of unjust policy only plays into the hands of those who shape that policy by discreditng its critics.

I’m not speaking for GM here, only for myself. But what a pleasure to have such an engaged commenter as GM who has taken the time and effort to make one more outstanding contribution to our ongoing and open dialogue.

Again our ground rules: Flail away or applaud GM's work. Leave his person and your unfounded guesses about his motive and intent out.

Comments

[bains]

Your objections to the current Administration, in my estimation, are vastly bolstered by your objections to the unprincipled critics of said Admin. Unforntunately, aside from my Korean and Vietnam vet father, you are the only sane opposition voice I hear.

George Bush should be losing, nay should be lost, yet when the best the Democrats can set forth is JF Kerry speaks volumes to the dearth of character within US politics.

[Michael Turner]

Marc, I'd be careful about being prematurely impressed with GMroper's statistical acumen. Your own leaves something to be desired. There is actually far less than a 1/10th of 1% chance of the actual figure being 100,000 - exactly 100,000, that is. (GMRoper can probably explain why.) You might retort, "Well, you know what I meant." Actually, Marc, I don't. Because YOU don't know you mean when you say it. Statistics is a trickier subject than that. I know: I got a solid year of it under my belt in college, as part of two years of engineering-degree math prerequistites. (And I'm glad I did, because later, in physics, I was able to understand perfectly what Einstein REALLY got the Nobel Prize for.)

GMRoper doesn't cite sources in any detail. Doesn't that make you at least a little suspicious? His figures for UNICEF's estimate of infant mortality rate now in Iraq, for example. I go to UNICEF and I get a slightly different number:

http://www.unicef.org/infobycountry/iraq.html

You also get a different picture, one that suggest that UNICEF only gets numbers from the parts of Iraq it feels safe enough in.

The important thing to understand about methodology in cases like these is this: such radical uncertainties are themselves a huge, blinking red light. If a country can't count its own war dead to within a reasonable margin of error when civilian casualties are touted as being unprecedentedly light, you're looking at a country that's still in very big trouble. Could it be 8,000 or less? Could it be 195,000 or more? Geez, it's pretty messed up that we can't even tell, isn't it?

GMRoper is worried that a household can't produce a death certificate for the dead child it might still be grieving. Gee, it couldn't have been because the hospital was closed for weeks or months? Without electricity for too long? Maybe the doctors closed that hospital because they realized it was becoming more dangerous than leaving their patients were sitting out in the street, watching their babies die in their arms?

GMRoper supposes that people would exaggerate the number of deaths in their own household, and hypothesizes this on a basis that draws not from any previous studies of similarly distressed populations, and any follow-on corrections to those demographics , but rather from ... I still can't believe this ... a much-noted INDUSTRIAL-psych phenomenon he remembers as "the Westinghouse effect." Bzzt. Roper, that's called "the Hawthorne Effect." And please note: the original interpretation of the Hawthorne Effect has long since been debunked anyway:

http://www.nwlink.com/~donclark/hrd/history/hawthorne.html

But the real howler is Roper's "methodology" in even dragging the Hawthorne Effect into a picture like family grief. How do you work that, Roper? Families become ever more "productive" of exaggerations of their grief, the more "official" the census taker is? Oh please.

Is this recent study nevertheless flawed? And perhaps seriously so? I wouldn't be terribly surprised. That happens. Many supporting figures for the supposed Cambodian "autogenocide" trace back to a politically-motivated census conducted by the invading Vietnamese, during which they asked, merely, "did the Khmer Rouge kill any of your family?" In close-knit clan-based Cambodian rural society, same dead second cousin ended up being counted six times or more. This study sounds like it might have serious geographic skew, apart from having better-than-Vietnamese methodology overall. Interestingly, the authors defend it as the same methodology used to estimate the number of Kosovars killed by Serbs. Gee, could they be referring to a certain study that estimated 200,000 Kosovars killed (a number that figured heavily in the apologetics for NATO's collateral damage to Serbia), when the real figure turned out to closer to 20,000? Hm, I'm not going there.

Suffice it to say: I agree that this report, the haste with which it was released, the possible lack of adequate peer review, and the pre-election timing all indicate that there is partisan bias in the TIMING at least. What alarms me is that, as far as I know, nobody else is in an unbiased position to tell me, to within a reasonable confidence interval, just how bad the collateral damage is. Why is the picture still so fuzzy in the run-up to an election where Bush only puts rose-tinted Iraq-ulors before carefully filtered partisan audiences while out on this increasingly frenetic presidential stump? The U.S. invaded in spring 2003, and we don't even know how many people died as a result, to within TENS of thousands? So how can we possibly know how many January Iraqi votes will be legit?

[Michael Turner]

When I posted the above, I hadn't read Fred Kaplan's supposedly devastating "deconstruction" of the Lancet report. Now, having read Kaplan, I'm even more appalled by the statistical illiteracy of certain high profile detractors of this report.

It would be very easy to go blow-by-blow, tearing apart what Kaplan has so blithely thrown together ... so ... let's get started!

The New York Time article immediately cited by Kaplan leads off with this:

"An estimated 100,000 civilians have died in Iraq as a direct or indirect consequence of the March 2003 United States-led invasion ..."

This is ridiculous - the actual authors of the article don't claim anything like the precision implied. No wonder Kaplan was disappointed to read the actual article, which begins:

"The risk of death was estimated to be 2.5-fold (95%CI 1.6 –4.2) higher after the invasion when compared with the preinvasion period. Two-thirds of all violent deaths were reported in one cluster in the city of Falluja. If we exclude the Falluja data, the risk of death is 1.5-fold (1.1 –2.3) higher after the invasion. We estimate that 98,000 more deaths than expected happened after the invasion outside of Falluja and far more if the outlier Falluja cluster is included."

Now, note the word "expected." This isn't your father's "expected" as in "we expected the baby sometime in the last two weeks, but finally went for a C-section." This is the "expected" of the highly technical field of statistics, which allows for vary wide ranges, and yet is meaningful in a statistical sense. The *expected* IQ of the next person you run into is exactly 100. The probability that it will be 63 is quite low. The likelihood that it's significantly higher or lower than 100 is nevertheless very high, but the probability of a given IQ trails off as you diverge from 100.

A bit further on in the Lancet report: "Interpretation: an estimated 100,000 civilians have died in Iraq as a direct or indirect consequence of the March 2003 United States-led invasion ..."

Well, considering that they decided to exclude Fallujah, and avoided areas that even their fellow Iraqi researchers considered too dangerous, why is this unreasonable, in light of all their previous, careful qualifications to the number? In view of the endless qualifications (the main text of the article says right out that the number is "uncertain", why is Kaplan pretending that this report hides its uncertainties in statistical jargon? He quotes this bit:

"We estimate there were 98,000 extra deaths (95% CI 8000-194 000) during the post-war period."

Then says this - quite sneeringly, too - about it:

"Readers who are accustomed to perusing statistical documents know what the set of numbers in the parentheses means. For the other 99.9 percent of you, I'll spell it out in plain English—which, disturbingly, the study never does. It means that the authors are 95 percent confident that the war-caused deaths totaled some number between 8,000 and 194,000."

"Disturbingly"? It's an article for technical professionals! Are they supposed to have a "Statistics for Dummies" sidebar?

Kaplan continues:

"Imagine reading a poll reporting that George W. Bush will win somewhere between 4 percent and 96 percent of the votes in this Tuesday's election. You would say that this is a useless poll and that something must have gone terribly wrong with the sampling. The same is true of the Lancet article: It's a useless study; something went terribly wrong with the sampling."

That may be plain English, but it doesn't spell it out. Because if it did, you'd see that it's Kaplan who has gotten something terribly wrong.

Draw a bell curve. You've seen those, right? Now draw two vertical lines, one at each tail, just about where you think 5% of the area under the curve would be included to the left or right. That's what "(95% CI 8000-194 000)" means. It doesn't mean that the probability of 8,000 deaths or 194,000 deaths is equal to the probability of 100,000 deaths. Exactly 100,000 deaths is very unlikely, but it's still considerably more likely than those other two numbers. Kaplan's looseness with statistics extends to his "99.9%" - actually, more than one person in a thousand is likely to see the reasoning flaw implied above, if only because more than 0.1% of the population has taken some kind of statistics course.

Kaplan continues:

"Imagine reading a poll reporting that George W. Bush will win somewhere between 4 percent and 96 percent of the votes in this Tuesday's election. You would say that this is a useless poll and that something must have gone terribly wrong with the sampling. The same is true of the Lancet article: It's a useless study; something went terribly wrong with the sampling."

Apples and oranges. The authors of this study don't hide the reasons behind the high likelihood of error, much as Kaplan would like to pretend they did. They are quite forthright about the possible sources of error. The clearly stated uncertainties stem from conservatism on the part of the researchers, and from chaotic state of data collection in Iraq today. Amazingly, the Iraqi health ministry only counts infant deaths in hospitals, for example, a figure that certainly understates that figure significantly in a country where it's often hard to make a phone call to determine whether there's a hospital close enough (and functional enough) to be worth getting a death certificate from.

Kaplan goes on, quoting Burman as saying:

"We're quite sure that the estimate of 100,000 is a conservative estimate."

Then adding this:

"Yet the text of the study reveals this is simply untrue. Burnham should have said, 'We're not quite sure what our estimate means. Assuming our model is accurate, the actual death toll might be 100,000, or it might be somewhere between 92,000 lower and 94,000 higher than that number.'"

Kaplan probably misconstrues what Burnham was saying. As an EXPECTED value (in the statistical sense of "expected") Burnham may be saying "We used conservative assumptions in arriving at this expected-value number." And you can't interpret it as saying it's only within the range Kaplan says, because in this style of estimating, it could be higher or lower (but with a confidence of only 5%.)

Kaplan again:

"They ... took the results of their random sample and extrapolated them to the entire country, assuming that their 33 clusters were perfectly representative of all Iraq."

Actually, the authors take pains to point out several ways in which their methodology might not be "perfectly" representative, and make no claims of perfection. Here's a choice example:

"To account for the potential that the Falluja data are profoundly skewing the mortality estimate or the potential that there is a recall bias in the infant mortality data, a lowest plausible death toll has been calculated excluding the Falluja data and assuming that
half the measured increase in infant mortality has been an artifact of selective recall. Removing half the increase in infant deaths and the Falluja data still produces a 37% increase in estimated mortality. The inclusion of this estimate does not mean that investigators believe that either bias has occurred. Instead,this estimation reflects the concern that investigators cannot fully discard the potential for bias from these two factors."

You want that in plain English? OK, here it is: "When we get ridiculous enough to pretend that Fallujah didn't happen, and that any woman could actually forget that one of her babies died if it happened to be a few years back, we're still looking something IN THE NEIGHBORHOOD of a 37% increase over our baseline assumption."

Regarding this effort to make sure their EXPECTED value (note statistical sense of the word) was "conservative", Kaplan darkly hints at some hidden agenda:

"A question does arise, though: Is this difficulty a result of some peculiarity about the fighting in Fallujah? Or is it a result of some peculiarity in the survey's methodology?"

Reset your head Kaplan: all they were trying to do was make the point that they are willing to exclude a conspicuous outlier in the interests of conservatism, in view of all the uncertainties.

And more:

"There were other problems. The survey team simply could not visit some of the randomly chosen clusters; the roads were blocked off, in some cases by coalition checkpoints. So the team picked other, more accessible areas that had received similar amounts of damage. But it's unclear how they made this calculation. In any case, the detour destroyed the survey's randomness; the results are inherently tainted. In other cases, the team didn't find enough people in a cluster to interview, so they expanded the survey to an adjoining cluster. Again, at that point, the survey was no longer random, and so the results are suspect."

Gee, sounds like the results are "inherently tainted" to produce an UNDERestimate, if anything. They couldn't get past a checkpoint? Danger ahead - hence, a likelihood of greater death toll. Not enough people in one area? Well, what should that tell you, if it was hitherto a relatively populous area? (And if it wasn't, how does that skew the result very much? And if it was an *abandoned* area, well, the soon-cited Beth Osborne DaPonte of Yale would say that not counting anything there would tend to underestimate deaths, because refugees in wartime die in greater numbers than anyone except those in the neighborhood of a direct bomb hit.)

More: only *apparently* citing DaPonte, Kaplan says,

"According to quite comprehensive data collected by the United Nations, Iraq's mortality rate from 1980-85 was 8.1 per 1,000. From 1985-90, the years leading up to the 1991 Gulf War, the rate declined to 6.8 per 1,000. After '91, the numbers are murkier, but clearly they went up."

Slow down. 1980-85 corresponds to the bulk of the Iran-Iraq War - one of the deadliest post-WW II wars. That war ended in Aug 1988. Thus, the 6.8 mortality rate figure for '85-'90 probably overstates the mortality rate before Gulf War I, since it doesn't reflect about a year and half of peace with Iran.

OK, and now: how can numbers be "murkier" but "clearly" go up from this probably overstated 6.8 figure? What was Beth Osborne DaPonte's direct quote about these figures? Kaplan doesn't supply one.

So let's check her out, shall we?

And here's what we find, in a web-reprinted BusinessWeek article:

http://www.worldrevolution.org/projects/webguide/article.asp?ID=565

In other words, one reason why DaPonte might go for a generally higher figure during much of the 90s is the same reason she almost got ousted by none other than Dick Cheney over a decade ago. She went public with a figure of 70,000 Iraqi civilians killed directly and indirectly Gulf War I, way higher than Dick wanted to go for an invasion that many, if not most, Americans will characterizing as "hardly killing anyone." That count can only have gone higher after the report that doomed her government career, since she was counting after-effects that dragged on for awhile after it was issued. (One of the most persistent of which was denying Iraq chlorine for purifying water under the anti-WMD pretext, which was a significant contributor to infant and child mortality.

Kaplan makes no mention of this inconvenient fact about DaPonte. Nor does he mention anywhere that quite a lot of the killing of Kurds during the sanctions period was at the hands of other Kurds, working hand in glove with Saddam's forces, in putting down U.S.-sponsored destabilization attempts (of which Ahmed Chalabi's was only the most shameful and pathetic.) Since the Lancet report is aiming only to put some estimate on civilian casualties, using DaPonte's *aggregate* (but "murky", and uncited) figures is a bit disingenuous. One wonders what DaPonte herself thinks of Kaplan's citation of her. Probably not much, since it sounds like she'd say, "if the figures in Lancet aren't supportable, the fact is, we're not doing better while fighting this resistance than Saddam was going while fighting his own resistance, and it's unacceptably bad either way."

Kaplan continues:

"There is one group out there counting civilian casualties in a way that's tangible, specific, and very useful—a team of mainly British researchers, led by Hamit Dardagan and John Sloboda, called Iraq Body Count. They have kept a running total of civilian deaths, derived entirely from press reports. Their count is triple fact-checked; their database is itemized and fastidiously sourced ..."

Ah, how many journalists are forced to take a statistics course? I really wonder about that. The bias for cold, hard numbers accurate down to the decimal point will inevitably lead to significant underestimates. And in a country in chaos, it will lead to laughable underestimates. Wouldn't it be great if somebody had the resources to triple-fact-check every death, everywhere? I don't think countries do that very often even in peacetime.

[Michael Turner]

I made some errors in the above, including quoting the same section twice, and some typos.

But you know what? My errors are nothing compared to how statistically illiterate Fred Kaplan is. So I'm outa here. If this is the level of debate right now, I have much better things to do with my time. Obviously, trying to reason with people about statistics in this political environment is pissing in the wind. Spin it all however you want.

[GMRoper]

A deconstruction of Lancet far better than mine can be found here http://obsidianorder.blogspot.com/2004/10/pick-number-any-number.html

[Michael Turner]

Yeah, I looked at that. Obsidian Order was good as far as it went, and the CDA report cited was fine work - within some limits that should be understood. Those limits are defined largely by what sorts of numbers you can get faxed into the Green Zone. I.e., they are based on what numbers you can get SAFELY.

The report under discussion is primarily flawed by the small numbers of deaths reported, using a methodology that, with more field workers and more effort, and more area coverage, might eventually produce something that we haven't seen yet: a truly credible estimate with tolerable sampling error. The dilemma is obvious: anyone willing to take those sorts of risks in Iraq these days is likely to have an axe to grind. Anyone not willing to take such risks is likely to produce an underestimate based only on official sources. Isn't that a bad enough sign in itself? The Lancet report closes on a note of encouragement: it says that better numbers are possible. I hope it encourages some other people to step up to the plate.

Now, I *really* am outa here.

[justpassingby]

Hi, Marc!

Was it Emerson who said that the radical of today is the stuffed shirt of tomorrow?

Happy Sunday!

[silent cal]

As a "statistically correct" projection, The Lancet report is, as the authors themselves admit in an interview with Spencer Ackerman on TNR, imperfect - and perhaps terribly flawed or damnably worthless as a statisticans construct to mathematical cogniscenti. As an empirical project conducted on the ground in Iraq, the results of the thousands of random interviews with familes are not refutable and should be disturbing to anyone who actually cares about the issue at hand. I remain more impressed by the comittment of the people who conducted the study to digging out facts of the matter than the dissectons proffered by various bloggers from - if you'll forgive my Rumsfeldism - their air-conditioned offices.

Hey, Mike, thanks for that. I found your taking apart the claims of Berman and others about Cambodia to also pretty darned sharp. That was on Totten's blog that you did that, nice job.
I find it odd that Marc thinks I'm pointing at him for not giving Haiti, a much worse human rights violator than Cuba, equal time. I guess that's what happens when aiming to stereotype people from the left that disagree with you, just don't read what critics from the left actually write.
I haven't actually responded to Marc directly except when I think there are alternative explanations for things he sees as major catastrophes and the like. My equal time comment has been primarily directed at that which has far greater capacity (and therefore responsibility) for the American obsession with Cuba, namely the "liberal" media, which systematically ignores more serious violations in places like Haiti.
Re: the Iraqi numbers. It is odd to hear those who were willing to believe any number thrown out by the US on "WMD"s or numbers killed by Sodom, Milo, Ortega,...Or who throw out the craziest stats about the Great Leap Forward famine, Cambodia 'genocide', etc [cue to the robots, this is the chance to say--"AHA, you support the Khymer Rouge..."]... Might one suggest a tad inconsitency here?

[Marc Cooper]

Michael Turner: Y'know... you spent an awaful lot of words to only further convince me that the Lancet Report lacks any relaibility. You put some dings in Kaplan and GM< but do nothing to bolster the Lancet study which remanins--in my eyes-- un-bolsterable.

[GMRoper]

Dings I have had a few.
Dings are, to me, nothing new!
I labor for the Truth
Capital "T" you know
And lots of dings I show.
But truth is out there you see
It's still there for you and me.
I'll keep trying to find it there.
So a few dings will come, I do not care!
And when Truth finaly itself reveals,
away the darkness it steals;
and glorious light doth appear
that to brotherly love we finally near.

[rosedog]

Terrific food for contemplation GM and Michael T.

Michael, by the way, although I generally tend to list more toward your analysis than GM's (although you both make excellent points), it should be noted that the “Westinghouse effect” that GM cites is not a misnomer, but another version of the “Hawthorne effect” that you mention. Both suggest that studies regarding worker productivity are likely to be skewed by---not only the variables in any given study---but because those being studied tend to perform better in the light of attention.

GM also mentions the Heisenberg effect….You remember Werner Heisenberg...the father of quantum mechanics, and author of the uncertainty principal. Heisenberg posited that the act of observing something essentially changes the thing itself.

How exactly this relates to Lancet, I’m not sure---unless GM you want to completely throw out any and all studies compiled using self-reporting.

Nonetheless, frankly I'm delighted to be reminded about all of the above principals and plan to bring them up the next time I guest lecture at a journalism class and the little darlings start nattering to me about journalistic "objectivity."

As for the bottom line---Silent (and occasionally Surly) Call says it well: As an empirical project conducted on the ground in Iraq, the results of the thousands of random interviews with families are not refutable and should be disturbing to anyone who actually cares about the issue at hand." In other words, as I said before, we should facter Lancet in---not take it as gospel---or throw it out altogether.

OT: A normally very sensible friend---also a respected therapist---has just called me to ask if the fate of the Presidential election hangs on the outcome of today's football game between the Greenbay Packers and the Washington Redskins.

Thoughts?

[Obsidian]

silent cal: "As a "statistically correct" projection, The Lancet report is ... perhaps terribly flawed or damnably worthless... As an empirical project conducted on the ground in Iraq, the results of the thousands of random interviews with familes are not refutable and should be disturbing to anyone who actually cares about the issue at hand."

I agree completely, on both counts. However, if you look at what they're actually counting - which is not only civilian deaths, and not only deaths due to violence - the numbers begin to make sense, within the huge margin of error. I can't say anything more without seeing the primary data which I am attempting to obtain from the authors.

silent cal: "I remain more impressed by the comittment of the people who conducted the study to digging out facts of the matter than the dissectons proffered by various bloggers"

So am I. However, I cannot help asking "cui bono". Clearly the authors wanted to exaggerate their numbers, and to time the release just right. The abstract of the paper bears almost no relation to their findings... this is not neutral scientific research, there is clearly an agenda. The big question for me: was the enemy involved? They have conducted psy-ops directed at journalists before, and they have released plenty of extremely-hard-to-believe statistics, especially in Falluja... food for thought.

[rosedog]

I meant "Cal" not "Call." (The perils of spell check.)

VERY moving poetry, GM. ; -)

[Obsidian]

rosedog and GM: I don't think that a Westinghouse effect will apply here very much. However, there are several effects that specifically affect studies of this type: telescoping and imperfect recall. Telescoping is the tendency to mis-remember when effects happened, making older events appear more recent. If you ask people whether they were robbed in the past year, or in the year before that, you will uniformly find that more people say they had were robbed in the year before that. Why? The longer ago something happened, the harder it is to remember exactly when. They probably *were* robbed 3-4 years ago but do not remember the exact time. Imperfect recall is that less-important events tend to be more easily forgotten the longer ago they happened. If you ask people whether they stubbed their toe, more will say they did in the last year than two years ago. In this study infant deaths *may* fall in this category which is why the pre-war number is lower than other estimates.

[silent cal]

"was the enemy involved" ?

Probably depends on how you define "enemy"...

[silent cal]

I'm curious just how - aside from the obvious issue of polling in a war zone - the Lancet study compares in relative accuracy to the myriad polls we're being treated to on the presidential race which vary radically poll to poll and week to week. Does GM think that Gallup is doing a better job than Lancet ? I mean 8 or 9 hundred interviewees based on a subjective guesstimate of likely voters broken down by party affiliation strikes me as something just bit better than a shot in the dark ? Should we be as dismissive ? If not, why not ? If so, why are these projections the focus of reams of commentary by the punditry and pack journalists ?

Also, I've read that Zoby just did a phone poll directed toward cell phones and younger voters. Kerry crushed Bush. Very interesting. Of course, I'm sure the sampling errors and confidence rate and whatever and et cetera mean it's total garbage from a methodological perspective, but it's garbage that no one else has picked through yet. As potential wild cards go that suggest a Kerry win, I find it interesting.

"The big question for me: was the enemy involved? They have conducted psy-ops directed at journalists before, and they have released plenty of extremely-hard-to-believe statistics, especially in Falluja... food for thought."

That's almost a farcical question. The resistance couldn't even begin to compete with the US in psyops operations, either in absolute senses or in quality. The US has journalists joining them as 'embedded' members and are very vulnerable to all kinds of psyops campaigns in the most direct sense. One need only think of the last few months' 'reporting' that occurs everytime the US drops a 'precision' bomb on civilians. The reporters then dutifully repeat the military press handout that states the 'precision' bomb targetted and hit 'insurgents' and/or 'terrorists' [the latter word being equated with anyone the US opposes]...end of story. The guerillas fighting the US could never match such marketting sophisitication, even if they tried.

Also, I've read that Zoby just did a phone poll directed toward cell phones and younger voters. Kerry crushed Bush.

--then add on the new voters that the Gallup folks understimate bigtime...

From Juan Cole, who is probably a lot more able to give an expert opinion on the topic of the study at hand:

I think the results are probably an exaggeration. But they can't be so radically far off that the 16,000 deaths previously estimated can still be viewed as valid. I'd say we have to now revise the number up to at least many tens of thousand--which anyway makes sense. The 16,000 estimate comes from counting all deaths reported in the Western press, which everyone always knew was only a fraction of the true total. (I see deaths reported in al-Zaman every day that don't show up in the Western wire services).
The most important finding from my point of view is not the magnitude of civilian deaths, but the method of them. Roberts and Burnham find that US aerial bombardments are killing far more Iraqi civilians than had previously been suspected. This finding is also not a surprise to me. I can remember how, on a single day (August 12), US warplanes bombed the southern Shiite city of Kut, killing 84 persons, mainly civilians, in an attempt to get at Mahdi Army militiamen. These deaths were not widely reported in the US press, especially television. Kut is a small place and has been relatively quiet except when the US has been attacking Muqtada al-Sadr, who is popular among some segments of the population there. The toll in Sadr City or the Shiite slums of East Baghdad, or Najaf, or in al-Anbar province, must be enormous.

[John x Moore]

As to who is going to win, at this point, it's just a matter of waiting.

The Hopkins study has been, I think, thoroughly discredited, while it's goal of causing people to look at civilian casualties right before the election has been achieved.

The real numbers of civilian casualties would be useful to know, but I don't see how it can be done. Marc, I disagree with you about it being the military's responsibility to keep track of them, simply because it is impractical.

Also, I would like to put this in context. Has any military ever been able to keep track of civilian casualties? If so, who? If not, what makes this war special other than your objection to it?

My other challenge to opponents of this war is to tell me your approach and what your approach would have produced.

I think this war could have been run better, but that applies to every war, including the one in Afghanistan, where 750 Americans, allied with various groups, replaced a government in two months, and produced a relatively stable outcome.

"As to who is going to win, at this point, it's just a matter of waiting."

Au contraire, it's already over. It's just a matter of when the US leaves, not whether or not it will be defeated anymore.
On one matter I certainly agree, this war is not that much different from any other war, thus the point. The largest numbers of casualties are civilians, and among them women, children, and the elderly, for the obvious reasons. On what could have been differently? How about not invading a country to help a president out with his reelection chances? In a democratic country, when the majority of the population are against a war in the first place, surely that was one option.

[John x Moore]

Actually I was referring to the election.

If you really think this war was about improving Bush's election chances, you haven't been paying attention. The Afghan war by itself would have done that, because it was a historic success. The president is catching nothing but criticism over Iraq. It is hardly a war for an election boost.

The argument that the largest number of casualties are civilians is completely unsupported. We have killed a large number of bad guys - sometimes at the rate of hundreds per day.

As to whether the US is defeated, it depends on who gets elected. Under Kerry, we can be assured that the Vietnam bug-out will be duplicated, leaving chaos behind us. With Bush, that is not going to happen. We have hardly been defeated. The insurrection is localized and we have just started a campaign of taking out its sanctuaries, such as Fallujah.

But you do sound as if you want us to be defeated. Well, nothing changed on the left. I remember from rallies I attended the chant "Ho Ho Ho Chi Minh, NLF Is Going To Win," always coming from the looney left like the SDS and VVAW. They got their wish, thanks to thee Democrats. It was a shameful event in our history.

"If you really think this war was about improving Bush's election chances, you haven't been paying attention. The Afghan war by itself would have done that, because it was a historic success. The president is catching nothing but criticism over Iraq."

You talk as though Bush had some crystal ball or people around him that felt that was a strong possibility. Neither is the case of course. It's not great secret that he felt he was going to gain political capital from the relatively easy task of overthrowing Sodom. There was never any sense, aside from obligatory statements made in line 456 of any given speech that there was an element of political risk. The Iraq 'victory' was clearly a major part of Rove's strategy. In fact, it has thrown the Rove team off balance since the US has lost Iraq.

I wish the claim that civilians suffer most greatly and constitute the greatest # of casualties, but if they weren't in this war it would be a major first. Your claim that the US has killed hundreds of insurgents can only be believed if we are to believe every word out of command central. few do anymore, for obvious reasons.

My guess with Kerry is he might even involve the US longer in Iraq, for fear of being attacked for doing otherwise. LBJ would be the model for that.

While I agree with you that Nixon was a great Democrat, I disagree that the SDS, VVAW, antiwar activists, represented a small proportion of the US opinion of the Vietnam War. Whether I wish or don't wish the US lose is immaterial, since it already has lost in Iraq. Maybe there's some way to invade a country under false pretenses and then try to privatize everything under the sun there, keep in place anti-union laws that Sodom made, and let foreign companies send home 100% of profits tax free...but Iraq isn't going to be the model for that I'm afraid.
And, since it is a matter of such little importance that even armchair warriors (not you, you're older than 30) and the Bush clan younguns' don't see the war as that terribly worth picking up a gun and going to Iraq to die for...I wouldn't get to worked up about the reality that we have lost the war. It's obviously not as important as the marketers proclaim.

[Obsidian]

Anonymous writes: "The resistance couldn't even begin to compete with the US in psyops operations, either in absolute senses or in quality."

On the contrary, I think that is the one area where we are much weaker, especially because of the ease with which our own media (even the Lancet) can be subverted. The Belmont Club has more about that:

http://belmontclub.blogspot.com/2004/05/news-coverage-as-weapon-historian-john.html

and examples of specific psy-ops:

http://belmontclub.blogspot.com/2004/04/coincidences-circumstances-fueling.html