Re: Cohen's Ecologic Studies (R-squared)

[epirad@mchsi.com]

Kai,



If you hope to determine if the relationship is linear 

or not between radon exposure and lung cancer, you need 

to be able to account for smoking.  I would have more 

confidence that he could account for smoking if he was 

able to show that his surrogate measures for county 

averaged smoking rates could at least help to predict 

the lung cancers within the county.  The low r-squared 

value indicates to me that he is not able to adequately 

account for smoking. 



Bill Field

> OK - we might not be able to agree on everything, but we should be able to

> agree on the meaning of a simple statistical indicator, such as the

> R-squared value. Please consider the following:

> 

> 3 counties, each with 3 million people. The number of smokers and lc deaths

> per unit time (what ever that is) is as follows:

> 

> County 1 has 1 000 000 smokers and 10000 lc deaths.

> County 2 has 1 000 001 smokers and 9900 lc deaths.

> County 3 has 1 000 002 smokers has 10120 lc deaths.

> 

> Perhaps Drs. Cohen and Field could both calculate the R-squared value (I

> think its pretty small) and explain how this value is related to the ability

> to "account" for lc due to smoking.

> 

> (To me, this data doesn't show anything about a smoking-lc relationship, but

> it doesn't disprove it either. The fact that R-squared is small gives no

> clue about the correctness or incorrectness of anything. It just says that

> it is a poor data set to use if you want to draw a conclusion about a

> smoking-lc relationship. It might be a good dataset, if you are looking at

> an "anything else-lc" relation.)

> 

> Thank You,

> Kai, the poor miner trying to understand this stuff

> 

> ----- Original Message -----

> From: "BERNARD L COHEN" <blc+@PITT.EDU>

> To: <EPIRAD@mchsi.com>

> ....

> >>  As for Dr. Cohen's

> > > ecologic analyses, he can account for only about 30% of

> > > the lung cancer mortality with his smoking data.

> >

> > --Wrong, wrong, wrong. The fact that R-squared is only 30% derives

> > from the small up and down statistical variations. The true indication of

> > predictability is the standard deviation of the slope of the regression of

> > lung cancer on smoking prevalence which is very small percentage-wise

> ....

> 

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

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/