Others have commented on the "correlation is not causation" aspect, the
ecological fallacy aspect and the autocorrelation aspects. But I felt that
another logical/statistical flaw was being featured.
I believe this summarizes the argument at issue:
Suppose that factors A, B, etc. explain X% of the variance in Z.
Then
1. all other factors explain 100% - X% of the variance.
2. any other single factor explains at most 100% - X% of the variance.
The author points out correctly that this syllogism is not valid.
This thesis does not involve causation or the ecological fallacy per se.
It does involve autocorrelation. But most importantly it involves a mistake
about the nature of R-squared as measured by a percentage.
In an intro stats class, I would say:
Explanatory power (as measured by R-squared) is not "exclusive."
["Exclusive" is not technically right, but it has the right connotation.]
My "name" for this fallacy is the "fallacy of exclusivity."
The proper name for this is the fallacy of orthogonality. In probability it
might be termed the fallacy of independence, but since we are dealing with
fractions of 100%, the fallacy of exclusivity hopefully communicates the
most understanding -- even if it is not technically correct.
================================================================
Gene Gallagher wrote in message <8fd1h4$g00$[EMAIL PROTECTED]>...
>Here is an error that is subtle, but very common. The statistical
>test (multiple regression) was applied perfectly, but the
>statistical inference was wrong.
<snip>
> The logical/statistical flaw in the Australian thrip story was
>published in Smith, F.E. (1961) Density dependence in the Australian
>thrips. Ecology 42: 403-407. Since weather accounted for such a
>high proportion of the variance in the data (78%), A&B assumed other
>factors could not be important. This is a fallacy.
<snip>
> This fallacy should have a name, but I don't know it. I point
>my students to Wright's path analysis and structural
>modeling approaches (LISREL, and AMOS) to show alternatives
>to the misleading inference based on an R^2 in a multiple
>regression equation.
<snip>
>--
>Eugene D. Gallagher
>ECOS, UMASS/Boston
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
