On 14 Jul 2003 10:12:42 -0700, (Richard Alexander)
wrote:


Many things don't fit anything except
statistics/probability. There you have to go by the weight of the
evidence. The notion of 'testable' is not a binary yes or no answer. It
may be testable with a certain percent confidence level or have a
certain correlation coefficient. On the other hand, you could plot
bubble gum sales as a function of meteors seen in the southern
hemisphere and might find a pretty good correlation......

The term "testable" means that any random person who correctly
performs the experiment would get similar (generally within 10%)

there is no such universal guideline about reproducibility. In
physics, they argue about things that can only be distinguihed out at
several decimal places.



results, that is, the results are universally repeatable. If plots of
bubble gum sales as a function of meteors correlates testably, that
would be an amazing coincidence!



But the point is to use an odd example to illustrate the idea of how
science works. If someone does it and publishes it, they would state
"We have observed a correlation ..." And then someone else would try
to do it. If they get comparable results, that reinforce that the
correlation may be valid. If their results are not comparable, we now
have two expts with contradictory results, and over time people will
try to figure out why they got different results, or at least collect
more data to establish whether or not the effect is true. If
substantial data accumulates supporting that it is true, then it can
become a scientific theory... that there is a correlation. That
statement is useful for making predictions, and is a valid scientific
statement. Note that nothing is said here about the reason for the
correlation, which might then become the subject of further
investigating.


As others have noted -- and shown! -- the terms hypothesis and theory
are not used consistently. But "officially", in the context of formal
discussion of that somewhat abstract notion of "scientific method", a
theory is something that is generally well accepted (meaning that it
has been tested). Also note that it does not convey "understanding"
why it is true.

bob