20 October 2006

Another discussion of the Lancet Survey, entirely gakked

There are various claims in the survey; such as, 7% of the men in Iraq have died since the invasion, and since the population of Iraq is approximately 27,000,000, that means that 7% of half of that number have died; that is, 945,000 MEN.

Yet the figures are based on 'cluster samples', and two thirds of the deaths are said to have resulted from taking one 'cluster sample' in Fallujah. That is, that the surveyors asked people in Fallujah how many people had died in their family, and they then incorporated that with all the other data and extrapolated to calculate a figure for the whole of Iraq.

Any realistic researcher, on finding that their figures relied almost entirely on one cluster sample out of many, (but not nearly enough for accurate figures, which is another issue), would have asked themselves, can this be a real figure? But they obviously didn't do so, because here's how you can tell immediately that this is NOT a real figure.

Suppose for a moment that two thirds of the deaths in Iraq happened in Fallujah, which is what this seems to be saying. That would mean that 945,000 x 2/3 = 630,000 people died there. Isn't that a catastrophe? No, it's a falsification, because the pre war population of Fallujah was 375,000.

--- but if we try to work out how it is that so many men have died (and in the central region it's said to be 10%, not 7%) then we note that there are only 375,000 x 1/2 = 187,500 MEN in Fallujah, and for enough to have died to support the figure given for MEN in Iraq, then 187,500 people have died 630,000 times; that's 3.36 deaths for each man in Fallujah.

Now at this point, isn't it appropriate to ask, how it was that these 'statistics' were gathered?

By asking Arabs in Fallujah (of whom by their own report there are none left) how many of their relatives died. Well, they are bound to be truthful, aren't they? They've got a massive grievance. No one disputes that Fallujah was the centre of a lot of the resistance to the occupation. Is it really so hard to believe that they would inflate the death figures out of a desire to harm the reputation of the invader? Or did each of their men really die 3.36 times?

This survey bears no relation to reality, and the merest attempt to investigate it shows that this is so. What it also shows, is that the Lancet team didn't think to check their figures in this trivial manner; they just accepted them without further thought. I submit that this is a level of incompetence that is willful. Because it took me less than five minutes to work it out. They knew that the deaths quoted in their survey nearly all came from Fallujah. They should have investigated further. They obviously didn't, because the above is such a simple way to check. They asked people who bore a grudge, they didn't think to check the honesty of their account, and they didn't think to check their own arithmetic. Then they put this survey out to the world as if it meant anything.

Since 1.3% of the population of Iraq lived in Fallujah at the start of the war, then if we discount the figures from Fallujah and suppose that all the others are accurate (which we have no reason to suppose they might be, given what happened in Fallujah), then the rest of the data from this survey shows that the death figures for 98.7% of Iraq are three times lower, 315,000 men. Rounding up for Fallujah, that makes 319,148. That would be a tragedy, of course. But seriously, what reason have we to suppose that lies only emanate from Fallujah? The whole survey is shown to be based on accepting the word of those who have a grievance, never checking it, and imagining the impossible as a result.

It is said that 10% of men in the central region have died. Well let's suppose that these are mainly from the fighting age group, about one third of the male population. That means that about 30% of the fighting men are dead. Now let's ask, if that is the case, how many are injured? Army veterans can advise what figures would be realistic for injuries given the level of fatalities. Shall we say three times as many? In that case, we would find that 30% of men of fighting age were killed, and 90% were injured, a casualty rate of 120%. Well that ought to stop the fighting! Does anyone find that remotely credible? Instead, we have to ask why it is that hundreds of thousands of deaths have gone unrecorded in Iraq, unreported by the media, unknown to the authorities, why hundreds of thousands of people who have been injured have never turned up at hospital, and why the media have not previously picked up on this problem.

There's a simple reason for it. Because the problem does not, on this scale, exist. It's a fantasy woven by the Lancet out of the lies of the aggrieved - and it could only be accepted by them on the basis of wishful thinking and a blatant disregard for the need to check their arithmetic, let alone the need to check the truthfulness of their sources.

So on the basis of the above arithmetic, this survey proves certain things:
- That people in Fallujah lied to the researchers
- That the researchers made no attempt to check the truthfulness of their sources
- That the researchers didn't make arithmetic checks on the viability of what was said to them
- That people in Fallujah have a bigger grievance than those elsewhere in Iraq
- That there is no reason to suppose that the figures reported anywhere in Iraq are accurate.

So the world is being fed statistics that do not face the most trivial of investigation, by people with an agenda to do 'research' that has no validity, and can trivially be shown to be untrue; and the media are either incapable of unpicking such trivial inaccuracies, or have no interest in trying to do so.

'Sir_Dave'
19th October 2006.

7 Comments:

Anonymous Anonymous said...

Linked to QuickNews

http://community.livejournal.com/quicknews/65098.html

- SpecialRpt

12:49 PM  
Anonymous cMAD said...

So the world is being fed statistics that do not face the most trivial of investigation, by people with an agenda to do 'research' that has no validity, and can trivially be shown to be untrue; and the media are either incapable of unpicking such trivial inaccuracies, or have no interest in trying to do so.


Do you have Excel on a computer you can play with?
Or, better yet, can you (or the prettier half of you) get Mathematica at a student discount?

I think it would make sense for you to figure out how to verify the plausibility of statistics in a meaningful way.

7:28 PM  
Blogger A Soldier's Girl said...

CMAD- I don't care how you play with the margins for error, it's not possible for every man in Fallujah to have died over three times. Unless Islam has some version of voodoun and zombification, it's just not physically possible.

Also, if you noticed, John did mention that the comments on the survery were not his, they were gakked from someone else. The person who he gakked it from is a scientist, actually, although he's a physicist rather than a statistician.

2:28 AM  
Anonymous cMAD said...

I just read the Lancet article again and didn't find the part where two thirds of the deaths counted occurred in one cluster from Fallujah.

Do I have to take Sir_Dave's word on this, or did I miss some reference to the raw data set?

If the claim that most of the deaths were counted in just one cluster is true, failing to mention this would be questionable science indeed.

What I did find in the text is that out of 629 deaths reported, in 545 cases the researchers asked for a death certificate and found one in 501 cases.

This makes Sir_Dave's version somewhat less credible.

10:46 AM  
Blogger Just A Decurion said...

I do find amusing that no, they don't give you the raw data. The 'one cluster in Falluja' is from the full version of the study, not just the Lancet paper.

As near as I can tell, their margin of error is estimated by them to be about +/- 200,000 deaths, or about twice what the Iraqi Ministry of the Interior total casualty figures are.

Either way, it is simply lidicrous to suggest that Iraq is being depopulated, with hundreds of thousands of military-aged males being killed and well over a million wounded without any notice in the press or without any of these people arriving in hospitals.

The claim in the Lancet survey is that Iraq has sustained proportionately greater losses in the Military Age Male demographic (15-59) than Germany did during WWI. If the Lancet were correct, and assuming that the ratio of 3 wounded to 1 KIA is correct, there are practically no military aged males left alive and unwounded in Iraq. I suggest that this is not true.

12:40 PM  
Anonymous cMAD said...

The 'one cluster in Falluja' is from the full version of the study, not just the Lancet paper.

Do you know where to find the full version of the study? I find only the paper and the appendices you already quoted.

3:24 PM  
Anonymous nerdasaaurus said...

So the world is being fed statistics that do not face the most trivial of investigation, by people with an agenda to do 'research' that has no validity, and can trivially be shown to be untrue; and the media are either incapable of unpicking such trivial inaccuracies, or have no interest in trying to do so.

A perfect example of this is the first Kinsey report which is still widely cited, but is pure unadulterated CRAP. Kinsey reported that 10% of the population is homosexual. Its just a lie.

It was only in the 1980's was his work researched thoroughly. Then it was discovered that he did his statistical research in the prisons, insane asylums, and universities. Further, Kinsey himself was revealed as a homosexual. But his work is still widely cited. There is a Kinsey Institute of Human Sexuality. But it is all based on lies, not science. Its all agenda-driven and cloaked in a mantle of academia.

9:07 PM  

Post a Comment

Links to this post:

Create a Link

<< Home