Lies, Damn lies and Statistics

[P3D Gregory J. Wageman <gjw@xxxxxxxxxx>]

Just to make it clear-- I'm not disputing George's personal observations.
I assume they are valid-- for him.  I simply maintain that no useful
conclusions about the world at large can or should be drawn therefrom.

Absurd example: I observe that I have four sisters and no brothers, and
one parent each of the opposite gender.  This is an inarguable fact.
However, were I to then generalize that the ratio of women to men in
the general population must therefore be 5:2, I would be wrong.  My
"data" is not false, although my conclusion is because my sample size
is simply too small.

This is, in my not-so-humble opinion, what George was doing.  He was
trying to start (no doubt with the best of intentions) a fruitless
discussion of why women favor collecting ViewMaster over stereoviews,
when it is far from demonstrated (to this reader's satisfaction, at
least) that this is even so.

Nine out of ten people* say you can prove anything with statistics.
They're wrong.  You can't prove ANYTHING with statistics (except,
perhaps, that you should have payed more attention in college). :-)

	-Greg W.

---
*surveyed and expressing a preference


------------------------------