Re: Ecological Fallacy?

[John Goldsmith <gjohn@bgumail.bgu.ac.il>]

Scott:
     Among epidemiologists, it has been the practice to designate as an
"ecological" study a study of a population for which the exposure
estimation is based on such variables as area of residence, and usually
time is also part of the exposure matrix.  What is considered to be a
fallacy is to asssume that all persons in such an area or population are 
having the same exposure, and therefore than any disease difference by
area is due to such exposures. Ecological studies in this sense are
relatively frequent because of their low cost...they avoid study of how
the exposure varies across various population attributes, for example, and
usually involve no new measurements.  From such ecologic studies, the 
strongest conclusions (based on excluding other obvious factors, usually 
such as age or occupational differences) can be ones of association.
If one can show that other relevant population attributes really differ
among the population groups, this may be considered as evidence for an 
"ecological fallacy" as the reason for the putative differences in
population health.
                John R. Goldsmith, M.D., M.P.H.
      Professor of Epidemiology, Ben Gurion U. of the Negev, Israel
     gjohn@bgumail.bgu.ac.il
On Wed, 25 Feb 1998, Sorensen, Scott wrote:

> RadSafers,  
> 
> I've been reading with interest the thread regarding radon epidemiology
> and its implication to the LNT theory.  Not being one intimately
> familiar with the science, could anyone post a description and describe
> the implications of "ecological fallacies?"
> 
> Thx,
> Scott Sorensen
> ssorensen@doeal.gov 
>