« Election Prediction. | Main | Binge Drinking and Minimum Prices. »
November 01, 2004
Eating Crow and Umble Pie.
Following my piece at Techcentralstation on Friday a number of people pointed out that I was in error. To be less euphemistic, that I had completely bollixed the statistical part of my argument, to the point where it was complete drivel.
My apologies. I shall stay rather closer to subjects I understand in future. The following has just appeared at TCS appended to a Michael Fumento piece on the subject:
"Further to my article of Friday on this subject. I'm afraid I mangled the statistical argument. My inadequate knowledge of the subject led me to make an argument that is incorrect. I stand by my contention that there is something fishy about this study (leaving aside the politically motivated timing of its publication, something the author has been clear about himself) yet have to admit that I have not found it, leaving me with nothing but personal prejudice upon which to stand my argument. I would also like to make clear that this subject was not "assigned" to me, the idea, research, argument and errors were all my own, as was my request for this clarification. Just in case you are wondering, being fact checked by the Pajamahaddin and being found in error does hurt and I hope that future writings will be, where necessary, so corrected."
My thanks go to Tim Lambert and Daniel Davis (amongst others) for helping to correct my error.
It is with a certain sense of irony that I present this email that I received, in fact the one immediately after the one with the link to my TCS apology. It is from a Research Fellow at a University in the UK (not giving the name yet as I do not as yet have permission):
it's the unverified self-reporting that troubles me most about the study, and
makes it next to worthless to my mind.
Also, they report suspiciously low infant mortality before the war:
http://www.guardian.co.uk/Iraq/Story/0,2763,1338749,00.html
"They found an increase in infant mortality from 29 to 57 deaths per
1,000 live births"
Yet, Unicef reports:
A rate of 102 for the year 2002.Assuming their estimate for infant mortality after the war is correct (57),
we'd actually have an improvement of 45, or in other words roughly a
twentieth of nearly a million births per year, multiply by one and a half
years, and you'd get 80,000 children saved or so ...
No, that does not excuse my error in any manner. I would be interested to hear what people have to say on this matter (although I think I know what the answer will be. Both the pre and post war estimates in the paper come from the study itself).
November 1, 2004 in Techcentralstation | Permalink
TrackBack
TrackBack URL for this entry:
https://www.typepad.com/services/trackback/6a00d8341c2d3e53ef00d834575c0069e2
Listed below are links to weblogs that reference Eating Crow and Umble Pie.:
» Tech Central Station 0, Lancet 2 from Deltoid
Tech Central Station has published Tim Worstall’s admission that his critique of the Lancet Iraq study was completely wrong:
Further to my article of Friday ... [Read More]Tracked on Nov 2, 2004 3:46:25 PM
Comments
I am surprised that you did not point out that it has long been the contention of the Left that "US sanctions" have been responsible for the deaths of 500,000 Iraqi children and an unusually high infant mortality rate. . .but now they are reporting that pre-war infant mortality totals were negligable...Has no one caught this blatant ideological switcheroo?
Posted by: Irene Adler | Nov 1, 2004 6:18:18 PM
Irene, I think that point would be addressed (not by me, but by those defending the paper) as above: the pre and post war mortality rates come from the study itself, not outside sources.
Posted by: Tim Worstall | Nov 1, 2004 6:39:01 PM
But would such a study's conclusions on infant mortality have been accepted "pre-war"? Of course not.
Posted by: Irene Adler | Nov 1, 2004 8:01:32 PM
I have now checked that it is the same definition for infant mortality.
For further discussion by myself I refer to Daniel Davis's comments page.
Posted by: Heiko | Nov 1, 2004 9:29:50 PM
I suppose this bit makes it even more embarassing:
"the left have never really been any good at numbers".
Ouch.
I'm not an expert statistician either, but I don't think you have to be to see that the Lancet study is flawed. Michael Fumento's post sums it up really, even if he mixes in some pig-ignorant generalisations of his own: 'the Iraqis' like nothing better than to 'pad body counts', apparently.
Posted by: Jim | Nov 1, 2004 11:24:56 PM
http://mikejericho.blogspot.com/2004/11/100000-deaths-in-iraq.html
Posted by: John Ray | Nov 2, 2004 6:22:04 AM
http://obsidianorder.blogspot.com/
There are some good points in this article.
Notably, it points out that, ex Falluja, they only have reports of 4 children and 2 women killed out of 21 violent deaths.
The conclusions he draws from the data seem reasonable: a few thousand civilians killed by the coalition and of the order of 25,000 combatants.
Posted by: Heiko | Nov 2, 2004 5:34:07 PM
Ok, I blogged again on Fumento (click on my name). What kind of person criticizes a study without reading it?
With regard to your question, the reports were (mostly) verified with death certificates. The increase in infant mortality they found is not, by itself, statistically significant.
Posted by: Tim Lambert | Nov 3, 2004 4:03:04 PM
