Re: Cohen's Fallacy

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

On Tue, 29 Jan 2002, Rad health wrote:



> I really sincerely question that you want to identify the cause of your

> inverse association.



	--I demand respect for The Scientific Method, requiring that a

plausible explanation be available for data that disagrees with

predictions of a theory. If you can suggest in concrete terms how that can

explain the discrepancy, please do. You insist on general statements that

would be impossible for me to evaluate. Why don't you give a specific

example that I can evaluate?





  I have pointed out with abstracts the limitations of

> using census data.  Your studies start with questionable averaged data and

> then you use other questionable data to try to stratify to correct for

> confounding.



	--Why don't you propose specifically how this can affect my

results?



  Why not spend your time trying to improve and understand the

> source of co linearity within a county by collecting information on that

> rather than asking the scientific world to explain your findings for you?

>

	--Why don't you spend some of your time reading and digesting my

papers; your comments show that you don't understand them.

	--It is not I but the scientific world that should be demanding an

explanation for the discrepancy between my results and LNT. Anyone who

understands The Scientific Method would be demanding such a

not-implausible explanation.

	--I would have read the Guthrie paper by now if I hadn't spent so

much time responding to your attacks.



> If you believe and others really believe that your one ecologic study has

> shown that the LNT is invalid I can respect your view without agreeing with

> it.  I would just urge you to try to improve your data quality by using SEER

> based incidence data for those states and getting some real data on

> cross-level bias and within county variability.

>

	--Thank you for your advice. I may even do these things.





> The one paper I posted shows how blacks are under identified in census data

> along with errors in SES.  There is no way to correctly stratify on a county

> level with this type of data when the stratifiications are based on

> misspecified data.  Black women and men have much higher lung cancer rates

> now because of various behaviors (smoking 20 years ago) and collinear

> correlated factors whose joint affect can not be theorized, but will result

> in inadequate adjustment thus creating a negative association between radon

> and lung cancer. It is up to Cohen to show this is not occurring by

> providing real data on intercounty variability.

>

	--There are plenty of counties that have virtually no blacks so

using those counties only would solve that problem. Also, I have used lung

cancer statistics for whites only, as well as for the total population.





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

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/