Author: Herman RubinHerman Rubin Date: May 14, 2008 14:36
>On May 13, 11:48=A0am, kev...@ mit.edu (Beth Kevles) wrote: >> Citing "The Bell Curve" makes me suspect everything else in those
>> posts... "The Bell Curve" is really NOT good science. =A0It's based on
>> ideas that lost respect after World War II (and for good reason)!
>Might this "good reason" be that it is not politically correct???
>Please substantiate your claim that it is not good science.
This is very definitely the case.
>The Bell curve *is* good science. I base my claim on the Central
>Limit Theorem. The sum of any set of iid random variables MUST
>be approximated by the bell curve.
This is not as good as it seems. The Central Limit Theorem
holds for independent, but not identically, distributed
random variables, but the approximation is nowhere near as
good as you seem to think it is. Even worse, the educationists
transform their data so that what they get is normal.
Also, a mixture of two different normal populations is
never normal. But one can still get a bell curve.
| 
