On Fri, 13 Sep 1996 pcapper@actrix.gen.nz wrote:
> (1) We inappropriately shoehorn people into IQ bell curves when the
> testing is done to the very young in age cohorts where, in fact, the age
> of the cohort can have an 11 month spread. This early distortion feeds
> through to all subsequent statistical manipulation of test data:
We also make the unsupported assumption that the sources of variation in
*childrens'* intellectual performance are the same as the sources of
variation in *adults'* intellectual performance. But no other aspect of
childrens' development behaves this way. Look at height, for example. In
a group of adults, all of whom were adequately nourished while growing up
and none of whom have any endocrine disorders, practically all the
variation in height is due to heredity. But in a group of children, all
of whom are adequately nourished and none of whom have endocrine
disorders, much of the variation in height will be due to differences in
the timing of growth spurts. This is a source of variation that is simply
not present in adults. If one were to look at a group of kindergartners
and attempt to predict their adult heights based on their relative
ranking, one would be wrong at least half the time.
We assume, however, that children don't experience intellectual growth
spurts, but instead gain intellectual maturity in a steady, linear
(sciences definition of "linear" [1]) fashion. But as far as I know, the
only test of this has been studies showing that children tend to maintain
their rank ordering on IQ tests as they grow older, and many of the
studies have used unrepresentative samples consisting primarily of
children who scored either very low or very high. The problem with these
studies is that they treat variation in scores between students who have
developmental disabilities and those who don't as the same phenomenon as
variation in scores within the population of students who don't have
developmental disabilities. Most students are neither gifted or retarded;
a score of 100 on an IQ test does *not* represent a balance between
retardation and giftedness; rather it represents the absence of both (just
as an adult height of 5'9" in a male is not the result of an equal
combination of conditions leading to dwarfism and gigantism, nor is a body
temperature of 98.6 degrees Farenheit the result of equal parts of
hypothermia and fever).
The fact that we can successfully predict that young children with certain
kinds of developmental disabilities will have a low level of adult
intellectual ability does not in any way imply that one can make similar
predictions of adult intellectual performance from childhood intellectual
performance for individuals without such disabilities. To do so is to
confuse special causes of variation with common causes.
> One of the consequences of these different polices for holding children
> back is that it comes to be believed that Scottish people (for example)
> are much more intelligent than English people (for example). In fact the
> observed difference is a function of different policies on class
> advancement of children feeding through into IQ test scores. Neverthelss
> the illusion leads us to the myth that Scottish education must be much
> better than that of anywhere else.
Similarly, the average SAT score for high school students from Iowa is
considerably higher than it is for most US states. But most Iowa students
who go to college attend schools in Iowa, which mostly use the ACT
(developed at the University of Iowa) rather than the SAT as their
admission test. Therefore, the group of students taking the SAT consists
mostly of students planning to attend college out-of-state, which means
that a) they generally come from wealthier backgrounds, since out-of-state
schools are more expensive and b) the group includes a
higher-than-expected proportion of students who are applying for admission
to highly selective schools like the Ivy League schools (both these
factors also mean that the students in the group are probably more likely
to have participated in test-prep programs that teach test-taking
strategies (*not* basic academic content), and these programs do work).
--Eric Bohlman <ebohlman@netcom.com>
Learning-org -- An Internet Dialog on Learning Organizations For info: <rkarash@karash.com> -or- <http://world.std.com/~lo/>