[FoRK] The Lancet publishes Johns Hopkins' research on Iraq - 100, 000 excess deaths, or more

Luis Villa luis.villa at gmail.com
Mon Nov 1 17:01:03 PST 2004


On Mon, 1 Nov 2004 09:27:27 -0800, FoRK Ian Andrew Bell
<fork at ianbell.com> wrote:
> Right, but if they're accurate at all it's due to dumb luck.  That
> said, it's about as statistically valid as most of the political polls
> done in the US.  The problem is must of those polls tend to be
> predictive in influencing thought as much as they measure it, while I"m
> hoping that a death toll estimate is less so.

I haven't looked yet at their math, but it would appear that they do
explicitly claim to be much less accurate than most polls- like
someone else said, it's as if CNN's latest poll said 'we think Bush
will get somewhere between 5% and 98% of the vote.'

In a similar vein today, I was disappointed to see people who should
know better:
http://www.lessig.org/blog/archives/002275.shtml
http://www.electoral-vote5.com/

give credence to a 'poll' of kids taken by SMS. It's nice publicity
for Zogby, but it's not actually meaningful, given that the people who
were polled /had all signed up via rock the vote./ It's a neat trick,
but your average Young Republican is not watching MTV...

Luis

> On 1-Nov-04, at 9:05 AM, Damien Morton wrote:
> 
> > Actually, the estimate isnt that bad, considering that the estimates
> > based on journalistic sources alone are in the 10-50K range.
> >
> >> I've got a better word for it:  propaganda.
> >>
> >> -Ian.
> >>
> >>
> >> On 1-Nov-04, at 6:09 AM, Luis Villa wrote:
> >>
> >>> Given that confidence level, they should be embarassed as scientists
> >>> to be going around saying that 'conservative assumptions' place them
> >>> at 100K. And the Lancet should be embarassed to publish it.
> 
> _______________________________________________
> 
> 
> FoRK mailing list
> http://xent.com/mailman/listinfo/fork
>


More information about the FoRK mailing list