Re: demo fervor

["Devine, James" <jdevine@xxxxxxx>]

I wrote:
> strictly speaking, no propositions at all can be "proved" with statistics.
All one can say is that your hypothesis survived a statistical test.<

Jurriaan writes: >True, but it occurs to me that this might be too strongly worded, insofar as
SOME propositions could be proved with statistics. For example, take
proposition p which states that y will always greater than x. I go and
measure y and I measure x and by golly, yep, I cannot find any single case
where p is false. Now you could of course say, tomorrow I could encounter an
example which contradicts p, and that is possible. This would demonstrate
what we knew logically anyway, namely that there is no absolute proof of p,
and, any counterexample of p is proof of that theorem. <

if the data is poorly measured, then even a case where there are no empirical counterexamples may be wrong.
Jim D.