----- Original Message ----- From: "Martin Lewis" <[EMAIL PROTECTED]> To: "Killer Bs Discussion" <[EMAIL PROTECTED]> Sent: Wednesday, November 10, 2004 4:39 AM Subject: Re: 100,000 Deadâor 8,000
On Tue, 09 Nov 2004 21:36:55 -0800, Doug Pensinger <[EMAIL PROTECTED]> wrote: > http://slate.msn.com/id/2108887/#ContinueArticle > > So, let's call it 15,000 orâallowing for deaths that the press didn't > reportâ20,000 or 25,000, maybe 30,000 Iraqi civilians killed in a > pre-emptive war waged (according to the latest rationale) on their behalf. > That's a number more solidly rooted in reality than the Hopkins figureâ > and, given that fact, no less shocking." > I'm baffled by this bit. How is that a number "more solidly rooted in >reality than the Hopkins figure"? He has simply said "oh, it's >probably a bit more than the IBC number so I'll just double that", >there is no basis for it whatsoever. It doesn't even fall outside of >the Hopkins CI. Well, lets look at another number: the number of people killed in car accidents in the US. If one simply looks at the count in official reports, one can get very close to the true number. If one were to use methodogy similar to that used in the Lancet report, one could get a number that was far off. Experimental science is rooted in methodology and an understanding of the inherent errors in methodology. Simple techniques, properly used, are far more trustworthy than sophisticated techniques, poorly used. > As far as I know The Lancet study remains the only scientific study >into deaths so until more studies have been done we really have no way >of knowing what an accurate figure is. But, we do have ways...For example, in Fallujah, we can look at the implications of the Lancet numbers. We can ask about the sources for violent deaths during that time. We can consider the likelihood of a few precision strikes killing 30,000....when many more similar strikes in Baghdad killed far fewer people. We can look for mass graves. There were people going in and out of Fallujah during those months; we can consider the likelihood that none of those leaving would mention the tens of thousands that were killed. We had reporters going elsewhere in that area...if tens of thousands were killed within the month, someone should have known. Experimental science is not a cookbook. It is a combination of a toolbox full of techniques, the explicit statement of assumptions, and estimation of potential error sources. "Scientific" techniques such as sampling are overwhelmingly dependant on the random nature of the selection. Anything that has the potential for introducing bias makes the results suspect. So, in short, using estimations such as numbers of deaths reported at hospitals, and then obtaining a range by determining likely sampling frequency in that number is far better science than using a non-random sampling of a small number of households. Dan M. _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l