> But my approach is NOT statistical approach. The statistical approach is more
> or less like you cannot explain the things, so you conduct tests and see the
> correlation among the variables. My approach is to explain things physically
> (thus the "physical model."), so the tests will not be to find statistical
> data, but to verify or disprove hypothesis. When one obtains a good model,
> then s/he will not need all the prohibitive amount tests because things can be
> explained easier.
I think that the term "statistical" is a loaded word here. Yes, one could
try to build a "statistical model" for which the physics is not known or
one can build a model based on physical principles. The later approach is
certainly the prefered route especially since it provides the basis for
choosing tests from the large pool of "statistical" possibilities. It
also affords a real understanding of what is going on.
- Wayde
(wallen@boulder.nist.gov)