__2/5/2001:__

__SAT-Practice
Word:
__spurious__Intermediate
Word:__tarantella__Difficult
Word:__pretermit

__Last night's disquisition deals
with the Flynn Effect.__

I've read recently
that the Flynn Effect applies only to IQ's in the vicinity of 100. By the
time the IQ has reached 140, the effect is minimal.

It seems to me
that this doesn't compute. It appears to me to lead to absurd internal
contradictions.

Over the 85 years
since the release of the 1916 Stanford Binet (S-B) IQ test, IQ scores have
risen by about three points per decade on both the S-B and the Wechsler
IQ tests. By now, S-B scores have escalated a total of about 25 points.
That means that an average child from 1916 who took the up-to-date (not-quite-yet-released)
Fifth Revision of the Stanford-Binet would be expected to average an IQ
score of 75 on that test. Working the other way around, today's child who
took the same up-to-date S-B Revision V test and received an IQ of 100
on it, would be expected to earn an IQ score of 133 on the 1916 S-B Test.
(As I've mentioned elsewhere, a shortfall of 25 points looking backward
to 1916 would imply a gain of 33.3 points of IQ looking forward from 1916...
4/3rds versus 3/4ths.) Now let's suppose that the claim that the Flynn
Effect doesn't apply to IQ's of 150 or above is correct. (To simplify our
calculations, I'll assume that it's zero for a ratio IQ of 150 or above.)
(One sticking point that arises here is the abandonment of ratio IQ's for
children, with no formula for relating ratio IQ's and deviation IQ's. We
don't actually need such a reconciliation formula, but we will need to
employ the concept of children's mental ages. The forthcoming Fifth Revision
of the S-B IQ test will reinstate the mental age scale that was abandoned
with the Third Revision of the S-B.)

. Let's suppose that we find
one or more 6-year-olds with an average IQ of approximately 100 (as measured
by the Fifth Revision of the S-B) and a mental age of about 6. We also
find one or more 6-year-olds with an average ratio IQ of about 150, and
an average mental age of about 9. If these children take the 1916 S-B,
the average children would be expected to secure an average IQ of 133,
and a corresponding mental age of 8. The children with IQ's averaging 150
will score about 150 on the 1916 S-B, corresponding to a mental age of
9 (In other words, their average mental age will be about the same
on both the 1916 S-B and the impending 2002 S-B. This means that on the
upcoming Fifth Revision of the S-B, the children would difffer by three
years of mental age, but on the 1916 S-B, they would only differ by one
year of mental age. Thus, 6 years of mental age today would correspond
to 8 years of mental age on the 1916 test, and 9 years of mental age today
would correspond to 9 years of mental age on the 1916 test. Consequently,
when the children who scored 100 at 6 years of mental age on the 2002 S-B
took the test three years later at the age of 9, they would still score
100 on it, corresponding to 9 years of mental age. But as we've seen, 9
years of mental age on today's test equates to 9 years of mental age on
the 1916 S-B. Consequently, their IQ's of 133 as measured by the 1916 S-B
at age 6, should now have dropped to 100 on the 1916 S-B at age 9. Furthermore,
their IQ's would have declined about 8 points per year. If we extrapolated
that another seven years, we might expect them to reach the age of 16 with
an average IQ on the 1916 test of 100 - (8 X 7) = 44, down from 133 at
the age of 6! Of course, that's nonsense, and a *reductio ad adsurdum*.

I have a fantasy
about how this conclusion that people at the upper end of the log-normal
or Pearson Type IV distribution aren't subject to the Flynn Effect might
possibly have been reached. For the past 15 or 20 years since the Flynn
Effect has come to everyone's attention, children's IQ's have largely been
measured with the Stanford Binet Revision Four test, and the WISC-III.
I believe that these tests measure

(1) deviation IQ's rather than
ratio IQ's; and

(2) are notorious for compressing
IQ's range near the upper end.s of their ranges

My wild guess is
that IQ's approaching 150 are really much higher, and that this hasn't
been taken into account in assessing the Flynn Effect at higher IQ levels.

Of course, this
is an uninformed speculation on my part, and may be entirely wrong. If
so, the Fifth Revision of the Stanford Binet, when it appears, should lay
this particular ghost to rest. But the idea that the Flynn Effect is as
great as it is near the mean and fails to operate on the wings of the distribution
seems to me to lead to contradictions.

** 2/1/2001:**
The only modern screening of extremely gifted children of which I'm aware
is that performed by Miraca Gross, who conducted her study in Australia
in the mid-80's using (among others) the 1973 Third Revision of the Stanford
Binet. That test would have been standardized 57 years after 1916, so the
Flynn-Effect rise in the average IQ of 100 should have been of the order
of 17 points looking back over that 57-year interval, or 20 points looking
forward from 1916. Out of her sample of, perhaps, 3,000,000 children, Dr.
Gross found four with IQ's of 200 or above, including one (Adrian Seng)
with an IQ above 220. Statistically, for deviation IQ's, that's impossiblefor
all practical purposes, but not for ratio IQ's if they relate to deviation
IQ's in about the same way as they did in the Terman Study (where one child
out of 250,000 had a ratio IQ of 201). Dr. Gross' findings tally nicely
with the Terman results obtained about 65 years earlier, and with the relative
frequencies of children with IQ's of 200 or above among the Quiz Kids about
40 years earlier. But this can only happen on ratio-IQ tests like the 1973
L-M Terman Test is if the Flynn Effect for children with IQ's of 200 is
twice as great as it is for children with IQ's of 100. That follows from
the definition of mental age, and from the fact that,

A principal problem with accepting the arguments I've espoused above is that of explaining how the great geniuses could have accomplished what they did with IQ's that couldn't have exceeded a present-day ratio IQ of 150, corresponding to a present-day deviation IQ of ~143. On the other hand, if we buy into the idea that there is little relationship between intellectual accomplishments and IQ once the IQ exceeds about 120, then this objection doesn't seem so strenuous. And on the third hand, what still seems to me to be hard to explain are the childhood precocities of great geniuses like Gauss. Leta Hollingworth's book "Children Above 180 IQ" gives us a glimpse of her subjects' levels of precocity during the 1920-1940 era. Their ratio IQ's might map into present-day ratio IQ's above about 145. Her children with IQ's at or above 200 would have present-day ratio IQ's at or above 160, or present-day deviation IQ's above 150. Is that reasonable?

** News Flash:** The UltraHIQ
site has gone off the air. The Mega Foundation has set up its s own in-house
server. I've altered my links, redirecting them to the Geocities website.
In a few days, I'll try to transfer the files to the new server.

I have temporarily removed the "Prior Material" link because of storage limitations. I'll reinstate it on the new Mega Foundation website, or elsewhere,.if that site update is delayed. You can find the same information in the Daily Indices above.

*"NOT EVERY MOUNTAIN"*

Not every mountain is an Everest.This bit of hillside sloping from my door,Whose yellow daisies bloom and sparrows nest,Is but a slant of earth and nothing more.This lazy creek that ambles on its wayThrough sun and shadow for a crooked mileSeeks but the tranquil margin of the bay,Not every strip of water is the Nile.

Yet Nature in her most expansive moodsHas dug no river bed with greater careThan this small gully twisting through the woodsAmong the wild blue flags and maidenhair,Has carved no mountain peak with fonder skillThan this green slope, this daisy-scattered hill.--"Window to the South."Vivian Smallwood