Re: correlation, critical temperature

[pwillis@ozemail.com.au (Paul Willis)]

>Dave wrote:
>
>>...variables that correlate yet have nothing to do with each other
>simply do not exist in reality.  If anyone can cite me just one example I
>will be only too happy to eat my words.<
>
>If I applied correlation analysis to two sets of truly random numbers I
>would get a correlation coefficient between -1 and 1. However, the
>usefulness of random numbers is that they are completely independent of
>one another. Munch, munch?

The human population of the world plotted against the distance between any
two galaxies would be pretty close to 1. Where's the connection? There are
some great Gould (the Great Gould) essays on this subject. I suggest that
list members who see causation in correlations get some of these essays and
read them.


Cheers,

Paul


Dr Paul M.A. Willis
Consulting Vertebrate Palaeontologist
Quinkana Pty Ltd
pwillis@ozemail.com.au

If smoking kills, how come it cures salmon?