Re: Dumb questions and comments on ecological/case studies

[BERNARD L COHEN <blc+@PITT.EDU>]

On Thu, 6 Jun 2002, Ted de Castro wrote:



> Ok - so its not correlated - but its still a massive background noise

> far exceeding the signal we are looking for.  Although not correlated -

> still it cannot be ignored - it doesn't just cancel out.  At the very

> least it removes subjects from our analyses before their test exposure

> can be completed and therefore represent exposures that cannot be used

> as part of the analysis.



	--There are two cases where smoking cannot change the results, if

smoking was exactly the same in every county, or if there was zero

correlation between smoking and radon levels. These two conditions are

relatively close to being true in my study. But of course I do not assume

them but use the data on them in my analyses.

	--The issue of background noise due to uncertainties in smoking

data is a quantitative one and is fully addressed in my analyses. The fact

that my results are discrepant with LNT predictions by 30 standard

deviations makes it clear that this noise does not have a large effect.



>

>

> BERNARD L COHEN wrote:

> >

> > On Wed, 5 Jun 2002, Ted de Castro wrote:

> > >

> > > It is also truly impressive that a small effect can be discerned let

> > > alone quantitized or signed in the presence of an overwhelmingly large

> > > confounder.

> >

> >         --According to LNT, something like 10% of lung cancers are due to

> > radon; moreover, variations in radon levels among U.S. counties are very

> > much larger than variations in smoking prevalence; moreover, radon levels

> > are determined by geology and house construction, and neither of these is

> > obviously related to smoking, so correlations between radon and

> > smoking cannot be very large. In fact, correlations between radon and any

> > socioeconomic variables are not large.

>



************************************************************************

You are currently subscribed to the Radsafe mailing list. To unsubscribe,

send an e-mail to Majordomo@list.vanderbilt.edu  Put the text "unsubscribe

radsafe" (no quote marks) in the body of the e-mail, with no subject line.

You can view the Radsafe archives at http://www.vanderbilt.edu/radsafe/