🚀 go-pugleaf

In article <73600017@uiucdcsp>, pax@uiucdcsp.cs.uiuc.edu writes:
> 
> Stephen Gould in 'The mismeasure of Man' explains why the concept 
> of IQ is flawed in the first place.  Basicly there is no such thing.  
> Given that, how can a test purport to measure it?

	What Stephen Jay Gould argues against is the scientific opinion
of the early part of this century, which I should add, is still the
prevailing popular opinion, that intelligence can be described as a 
unitary value. All modern (Post-1955 or so) psychometric testing depends
heavily on subtests.  In general, however, those who tend to do well on
one subtest tend to do well on all subtests, hence the belief that diverse
tests combined with sophisticated statistical analysis can tell us something
about the underlying nature of "intelligence."  However, the IQ score is
not this measure of intelligence - it is a weighted average of the subtest
scores.
	I have often compared IQ tests to benchmarks, and as most people 
here know.  It is very easy to cook a benchmark.  However, if you enough
of them, you can average them to get a good picture of how the systems
compare.  Intelligence testing is no different, and probably even more
complex, since measurement is more difficult.
	I should also note it is not necessarily true that performance
on subtests is absolutely correlated. Some people have selective defects,
some have selective talents.



-- 
	        Craig Werner   (future MD/PhD, 3.5 years down, 3.5 to go)
	     werner@aecom.YU.EDU -- Albert Einstein College of Medicine
              (1935-14E Eastchester Rd., Bronx NY 10461, 212-931-2517)
          "It's hard to argue with someone who knows what he's talking about."