Re: OT: Racialist evolutionary psychology


Manuel Doria wrote:

Due to my personal political convictions, I hope that the research
conducted by J. Philippe Rushton is wrong on the affairs of his partly
genetic approach to race differences in intelligence based on E. O.
Wilson's r/K selection model.

Whatever the nebulous IQ concept means, twin studies suggest that it is
highly heritable. There exists some other correlations, between cranium
size and IQ, for instance. So, I expect to find fluctuations of
"intelligence" which are genetic in origin in different populations.
..

For those not familiar with the thesis, here is a summary:
http://en.wikipedia.org/wiki/J._Philippe_Rushton

There is good evidence that intelligence (however one tries to measure
it) does have a heritable component.

But that isn't what is important in Rushton's work or your concern.
What both of you are interested in are: 1. Statistical significant
differences between groups and 2. implications for social policy.

In the table at Wikipedia, the most important information is missing:
THE STANDARD DEVIATIONS. In ANY test -- height, weight, upper arm
length, intelligence, brain volume, etc -- every population is going to
exhibit a bell-shaped curve of values. That is, when you plot the
number of individuals with a particular value on the y-axis and the
values on the x-axis (such as cranial volume), you get a bell-shaped
curve where you have a few individuals at the extreme left and right of
the curve and the most individuals (highest point of the curve) at the
mean.

When comparing groups, BOTH the mean and the width of the curve are
important. And it is the standard deviation that gives us the width.
2/3 of the total number of individuals lie within 1 standard deviation,
95% within 2, and 99% within 3 standard deviations.

Let me give a simple example. We have a curve whose mean is 100 and
whose standard deviation is 10. 2/3 of the people will have values
between 90 and 110. 95% will lie between 80 and 120, and 99% between
70 and 130.

Now we have a second curve with a mean of 106 and still have a standard
deviation of 10. 2/3 of the people are between 96 and 116, 95% between
86 and 126, and 99% within 76 and 136.

Do you see the overlap between the two curves? Less than 1% of the
people in the second group have values totally above those in the first
group.

My example is drawn from the IQ values and arbitrary SD's plugged in.

But what happens if the standard deviation of the second group is only
7? 99% of the people have IQs between 85 and 127. IOW, the first
group has more people with IQ's greater than 127 than the second, even
tho the second has a higher mean!

If the standard deviations are larger, it becomes even more difficult
to have meaningful differences between the groups.

Statistics try to determine the odds that two curves represent 2
different populations or are really from the same overall population
and the curves result from sampling error. We don't have the
statistical analysis. But, even if we did, it's only odds. A p value of
0.05 (usually considered statistically significant) means that 1 time
in 20 we would draw samples from a single population that would give us
the 2 curves. It could be our unlucky day.

In terms of social policy, the overlap of the curves makes
discrimination untenable. Yes, on average, Orientals had a higher IQ,
BUT that doesn't mean the individual applying for a job has the IQ of
106 or is the one with 86. Because of the overlap, you can't make
judgements from the group to the individual.

Also, you have to consider whether the differences are BIOLOGICALLY
IMPORTANT. For instance, does the difference in cortical neurons
between 13,665 and 13,767 really important? That's 112 neurons out of
over 13,000 or less than1%. Considering that we only use about 10% of
our brain, is that important?

I notice that much of the Table consists of value judgements. For
instance, how does one judge "cultural achievements"? What objective
criteria were used?

I can see several flaws with Rushton's work but, even IF (a big "if")
it were completely valid, any attempt to judge individuals within the
groups would be fatally flawed.

.

Relevant Pages

  • Re: Standard Deviation Chart
    ... thought of it as a "standard deviation" chart. ... "bell curve" is probably a more readily understandable description. ... Advanced Excel Conference - Training in Charting and Programming ... remember that I could not find an easy way to produce such a chart. ...
    (microsoft.public.excel.charting)
  • Re: OT: Racialist evolutionary psychology
    ... Statistical significant ... exhibit a bell-shaped curve of values. ... And it is the standard deviation that gives us the width. ... the overlap of the curves makes ...
    (talk.origins)
  • Differences in the Sexes
    ... I don't know which curve ... but the standard deviation for IQ among men is much ... One highly successful male ... to me that a horny female will never have problems finding sex. ...
    (talk.origins)
  • Re: How to minimize the absolute sum in curve fitting?
    ... I cannot use any software package or human judgement for this curve ... fitting, since I need to put this into a C routine to run on a device. ... Squares" and "Least Median Squares". ... I have improved the standard deviation of the estimated mean by 20% by ...
    (sci.math)
  • Re: Area of an envelope of a curve
    ... If I move a line segment of length 2d along this curve in a such way ... My guess is that again a similar inequality should be valid. ... As long as the curves sweeping out area don't overlap you are okay. ...
    (sci.math)