[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Cohen's Ecologic Studies



In the case you used, it confounds by adding "noise" to the data.  An analogy:
 
Take a large and aggressive skunk into a small room.  Kick it hard.  Now try to determine which perfume/aftershave your assistant is wearing.
 
That's a confounding factor.
 

Dave Neil               neildm@id.doe.gov

-----Original Message-----
From: RuthWeiner@AOL.COM [mailto:RuthWeiner@AOL.COM]
Sent: Tuesday, June 04, 2002 12:08 PM
To: blc+@PITT.EDU; tdc@XRAYTED.COM
Cc: EPIRAD@mchsi.com; mcaceci@radal.com; radsafe@list.vanderbilt.edu
Subject: Re: Cohen's Ecologic Studies

Dear RADSAFERs:

The following may very well just be stupidity on my part, but I have been puzzling over it for some time:

A "confounding factor" seems (to me) to be some factor or parameter that confuses (confounds) an otherwise clear cause-and-effect relationship.  Thus, I do not understand how smoking, especially more than a pack a day for 20 years, can confound the relationship between lung cancer and radon exposure.  The causal relationship between lung cancer and heavy smoking has been clearly established, but a causal relationship between indoor radon exposure and lung cancer (pace Bill Field and the EPA) has not been,  I could understand the reverse statement -- that radon exposure might be a confounding factor for  establishing smoking, or eve second hand smoke exposure, as a cause of lung cancer.  

I can see, for example how smoking can be a confounding factor in the relationship between black lung in miners and mine dust exposure, or in the relationship between asbestos exposure and asbestosis IN ASBESTOS WORKERS, but I don't understand how a factor that has been established as a primary cause of a disease can confound the relatioship between that disease and a secondary, not-well-established, somewhat speculative factor.

Can someone explain this, please, in some way other than with statistics?  I have nothing against statistics, but they show correlations, not causes.  Looking at correlations alone, it is perfectly possible to draw the wrong conclusions about causes, as in  the following canard:

Q:what is it in a highball, a gin-and-tonic, and a scotch and soda that gets you drunk?  
A: it must be the water, because they all contain more water than anything else.


Ruth Weiner, Ph. D.
ruthweiner@aol.com