Re: IQ and Testosterone?

Stephen Barnard (
Mon, 02 Sep 1996 16:18:13 -0800

Larry Caldwell wrote:
> Yep, the tortoise wins every time, right? In fact, you are falling into
> a common fallacy. Probabilistic information about a population is only
> valid if you have a statistically valid sample. An individual is not
> a statistically valid sample. Therefore, statistical information about
> a population conveys no information about an individual. I know, I
> responded the same way when I first studied statistics. There is a
> mind set that automatically leads to generalization and prejudice, but
> it is wrong. I'm not just talking morally wrong, I mean it gives the
> wrong results.

It goes without saying that you need a reasonable sample size to draw
statistically significant conclusions. Actually, I don't know whether the
MRI data amounts to a large enough sample size, so you may well be right
about that. You seem be be claiming, however, that statistically valid
inferences based on large sample sizes give no (probabilistic) information
about individuals the population from which the sample is drawn. That is

Steve Barnard