RM>Or ANY single factor.... Statistics, in general, with regard
RM>to education ARE, in fact, damn near useless for exactly the
RM>reason you cite above!
CB>I disagree.  Done over a large enough sample, as was the case in 
CB>Tennessee, other factors in the study of class size should become 
CB>trivialized.
I'd be interested in your explanation of how other factors
become trivialized with a large sample; I can't see how that
would happen?  If the factors are present, the size of the
sample, it seems to me, would be immaterial.... You're
suggesting that other factors cancel out?  That requires 
that there be an equal number participants on opposite sides
of each factor.  Demographics in any given state (citing
your example) are not equivilent to those of any other state
(at least to my knowledge), so the expectation that factors
cancel is overly optimistic, imo... I agree, however, that
a large sample, in general, improves the validity of a study.
My point was that many studies in social areas are VERY open
to interpretation, and often the conclusions are more a
function of the interpreters point of view than of the study
itself; this isn't ALWAYS the case, obviously, but it IS the
case more often than I'd like...
CB>Far too often, in education, we follow our hearts to the 
CB>new teaching techniques instead of the research.
I agree with you...
CB>It is time we stopped chasing educational fads that sound good on paper 
CB>(whole language, NCTM Standards), and followed the research.
I agree with you in principle... Obviously I'm going to be
a proponent of research and study; first of all, it's in my
nature, secondly, I'm trained as a scientist... Despite all
that (or, perhaps because of it) I'm skeptical of a lot of
the research as applied to education.  Since I am trained
in this type of thing, I recognize the limitations of 
testing when variables are not able to be isolated and
controlled.
___
 * MR/2 2.26 * OS/2 WARP: The choice of the next generation.
--- Silver Xpress Mail System 5.4P1a
---------------