Jim wrote:
> I think that Stephen's examples work fine when talking about the
> correlation across an entire sample, not particular pairs.
> Rick's comments here and for the others (deleted) demonstrate the
> difficulty with trying to make generalizations from a few
> selected cases and also the imperfect nature of correlations.
> Some heavy smokers do live to a ripe old age.
They probably do work across the entire population.
The problem is, in a classroom demonstration (which, of course, is what
was being disussed), Murphy's Law almost _guarantees_ the exceptions will
be present and volunteer! :-(
Rick