Re: Radon and Lung Cancer

[Muckerheide <muckerheide@mediaone.net>]

Hi Otto,

"Otto G. Raabe" wrote:
> 
> At 08:05 AM 03/13/2000 -0600, Jim M. wrote:
> >Otto,
> >
> >So...
> 
> Dear Jim:
> 
> Lubin's arguments are primarily mathematical. For him it is easy to show
> that, given an appropriate cross-correlation among the factors affecting
> lung cancer induction, ANY SLOPE in the radon vs. lung cancer is possible.

Not true. You could say, "For him, he could make up any mathematical
relationship that would give any slope he wanted."  Bernie's context
is whether any such relationship is "not implausible," but  without
meaningless mathematical manipulations there is no possibility Lubin
is 'right.' 

> Given a specific cross-correlation, the observed slope will be consistently
> found in various ecological experiments. It's a kind of built-in
> phenomenological bias.

Right. But such a "bias" can not be enough to change the results! 

Bernie says, "make one up, test it" to challenge a definitive result.
But no such relationship can exist in Bernie's data. Lubin is just
dishonest (I doubt he's that incompetent, though most are) in
rejecting Bernie's definitive results.

> "Where is the example of a credible ecological study that demonstrates any
> such effect?" Why, it is Cohen's data itself that provides that example, if
> Lubin is right.

You're being facetious, right? Nobody's that incompetent. 

A real example: anywhere a substantial ecological analysis is inverse
to the demonstrable result of the actual biological dose-response,
e.g., selenium, etc. They can make Bernie an 'example' only when the
result is biologically refuted, or ONE of Bernie's 100s of analyses of
a significant subset produces an inverse result from the whole
analysis. The consistency of Bernie's results of such subsets of the
data is an indication of the statistical unlikelihood that such a
result CAN exist!