Re: Race and IQ

On 20 Nov, 03:09, "Daniel T." <danie...@xxxxxxxxxxxxx> wrote:
Vend <ven...@xxxxxxxxxxx> wrote:
Daniel T. wrote:
Here is a great article explaining the
subject.http://www.shortenurl.com/8aezz(tinyurl seems to be
down.)

I scanned the article and it seems to me that its point is that
various tests which should supposely measure different mental
abilities give correlated results and with some statistical
analysis this correlation can be explained in terms of an
univariate hidden variable which has been called General
Intelligence Factor ("g"). The article then claims that "most
intelligence experts now use g as the working definition of
intelligence.". But since g is a statistical property of the test
scores, the article is effectively claiming that intelligence is
what intelligence tests measure, which is an obviously circular
definition.

I like the analogy used in wikipedia...

Let's say I wanted to compare the size of several people, but all I have
is their height, inseam length, chest circumference, waist
circumference, hip circumference, and arm length. I can use factor
analysis on all of these measures to get an idea of each persons size. I
would end up with a single measure for each person and can compare them
by size, despite not actually having a "size" measurement.

Further, suppose that studies showed that this "size" measurement had a
strong predictive capacity (better than any one measurement on its own)
as to the person's ability to do things like fit in a car, or look good
in a thong. :-)

Would you say that "size" measured this way is a statistical property
that the various length measures measure, and therefore it's a circular
definition? Would you say "size" isn't real?

In this example you measure quite well-defined physical properties and
thus the general factor obtained through factor analysis can be
considered a physical property itself (in fact, i'm pretty sure that
it would correlate well with the cube root of the volume if you were
able to measure it).

In the case of IQ tests and specific mental abilities tests instead,
the properties being measured have no clear definitions besides the
circular ones of being the properties measured by the tests.
Therefore the g factor is also circularly defined.

Moreover the article shows a table of correlations between IQ and
various "social outcomes". Some of those "social outcomes" refers
to both males and females, other to males only and other to females
only. This sounds me very much suspect of cherry-picking.

I don't know about that. I expect that comparisons between 'g' and and
some "social outcomes" didn't show a correlation. And there are probably
many social outcomes that simply haven't been studied (it's politically
dangerous to make such measurements after all.) Although 'g' does show a
strong correlation among several social outcomes (some much stronger
than others,) I doubt it does so for all of them.

.

Relevant Pages

  • Re: Race and IQ
    ... abilities give correlated results and with some statistical ... analysis this correlation can be explained in terms of an ... circumference, hip circumference, and arm length. ... despite not actually having a "size" measurement. ...
    (talk.origins)
  • Re: entangled states
    ... > you perform a measurement of an observable A with the outcome x then ... The correlation, described, ... then photon 2 must be vertically polarized ... polarisation of photon 2 before or after photon 1 since these ...
    (sci.physics.research)
  • Re: entanglement question
    ... Measurement of 's property would confirm ... The correlation is enforced even across ... Roll a die. ... The top-number is just as random as the bottom ...
    (sci.physics)
  • Re: Race and IQ
    ... circumference, hip circumference, and arm length. ... despite not actually having a "size" measurement. ... some "social outcomes" didn't show a correlation. ...
    (talk.origins)
  • Re: entangled states
    ... the correlation between the outcome of measurements is ... a purely positivistic point of view, where only observables are relevant. ... > you perform a measurement of an observable A with the outcome x then ... a notion of a theory is not meaningless if it is possible to ...
    (sci.physics.research)
Loading