Re: Re: Cohen's Fallacy

["Rad health" <healthrad@HOTMAIL.COM>]

Kai,



He could perform a case control study that eliminates cross-level bias or he 

could use the methods of Guthrie to help control the aggregate bias.  Have 

you read that paper yet?  If Cohen was really interested in improving his 

study he would incorporate such methods.  But, he always go back to 

theoretical and then uses his methods to disprove the theoretical that he 

constructs.  As he said before, I want people to give me an explanation I 

can disprove.  The point that I am trying to make that falls of deaf ears is 

why not determine the inter county correlations as suggested by Guthrie, 

then you would ned no theoreticals.





>From: Kai Kaletsch <kai@eic.nu>

>Reply-To: Kai Kaletsch <kai@eic.nu>

>To: Rad health <healthrad@hotmail.com>

>Subject: Re: Re: Cohen's Fallacy

>Date: Mon, 28 Jan 2002 23:10:14 -0600

>

> > The group confounding can cause

> > it to go either positive or negative without bounds.

>

>It will go positive or negative depending on how the group confounding is

>related to the average county radon level. If it aint related it will move

>everything to the null. I liked Lubin's example, where the same average

>levels of smoking and radon give different numbers of lung cancer. What is

>missing is an argument that relates this effect to the average amount of

>radon in the county. Without such an argument, the example would result in 

>a

>shotgun type scatter plot of county radon vs lc.

>

> > From the two

> > papers I posted, tell me how Dr. Cohen can correct for this error?

>

>If he didn't correct for it, his error would have moved his results toward

>the null ....

>

> > I

> > understand your an engineer type

>

>I prefer "scientist" or "physicist" type (but then again, physics is the

>only true science).

>

> > and engineers like to see specific causes,

> > but there is always not one cause

>

>Give me 100 causes! I am not looking for one single one. (I think the huge

>variation in miner data results is due to a whack of different causes.)

>

> > that is explainable without using methods

> > like the ones found in Guthrie's paper.

>

>I don't doubt that the analysis can be improved and I think it should. If

>Cohen doesn't want to do it someone else should.

>

>Kai

>





_________________________________________________________________

Join the world’s largest e-mail service with MSN Hotmail. 

http://www.hotmail.com



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

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/