From: Peter C. McCluskey (pcm@rahul.net)
Date: Tue Oct 22 2002 - 12:47:33 MDT
chris@pancrit.org (Chris Hibbert) writes:
>> [Les Earnest] pointed out that the correlation is less than 1.
>> "Extremely low" suggests that it is below 0.1. Nothing that I
>> have read suggests that it is below 0.9.
>
>I'll admit that I haven't seen statistics to back up my sense of
>agreement with Les' points. His arguments and my experience were
>enough to convince me of the validity of his points. Can you point to
>credible statistics that show that the correlation between the
>categories produced when Americans self-categorize according to any
>common system and the underlying genetics (in any reasonable sense) is
>higher than .8?
I'm not aware of any carefull attempts to quantify this. My estimate
is based on casual observations of the degree to which people agree on
who should be classsified as white or black, and that they appear to
base these classifications on features such as skin color which seem
to be strongly correlated with genes.
If the correlation was not above 0.8, I would consider it somewhat unlikely
that a mere 12.9 percent of Americans would label themselves as black.
>> It seems to confirm my guess that you are claiming that the existence of
>> noise in data shows that the data have a signal to noise ratio of zero.
>
>That's a strong overstatement. I'm surprised that you would believe
>(or merely suggest) that I reason that erratically.
>
>I believe (subject to challenge, of course) that there's so much noise
>in the data we're discussing and so little analysis has been done so
>that conclusions based on the data should not be trusted.
"Should not be trusted" sounds very different from "nonsense".
I agree that The Bell Curve's conclusions should not be trusted because
there has been inadequate analysis of some alternative hypotheses.
I don't understand why noise in the data would cause significant doubts
about their conclusions. Could random noise in the racial classifications
cause a persistent pattern of nonzero correlations between race and IQ?
That seems wildly improbable to me. Could the noise involve a large number
of high IQ people classifying themselves as white when they would have
classified themselves as black if they had lower IQ? If the correlation
between genes and racial classification is 0.8, then maybe that is
enough to explain the correlation between race and IQ. But if that 0.8
estimate is way too low, then it seems rather farfetched.
-- ------------------------------------------------------------------------------ Peter McCluskey | Free Jon Johansen! http://www.rahul.net/pcm |
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:17:43 MST