Opposition to Using the SAT as an
Admissions Test for the Society
In the last issue Ron Hoeflin proposes that we admit Brian
Schwartz based upon his score of 1593 on the SAT. While Ron has chosen to propose only that we
admit Brian, his arguments are based upon the acceptability of the SAT as an
admissions test. I therefore will oppose
this proposition as if it were a proposal that we accept the SAT.
I oppose this proposal for the following reasons.
Mega Society Constitution states that we are “an organization of people
who have scored at the one-in-a-million level on a test of general
intelligence which is credibly claimed by its authors to be able to
discriminate at this level.” The
ETS makes no such claim about the SAT; it even eschews claiming it is an
intelligence test. If you’re a
strict constructionist, that will be enough to oppose the proposal.
think we can all agree that the SAT does not contain any questions that
require one-in-a-million intelligence to answer. How then can it tell that someone has
such intelligence? Only if asking
many mid-range questions somehow boosts the range of the test. Where is the proof that this is
possible? We know from the example
of coin flipping that merely showing that the scores are distributed in a
bell curve is not a proof. Indeed,
as discussed in the next reason, we have evidence that there can be no
such proof, because it is not so.
- In a
previous issue I published the results of a simulation showing that
because of the errors of test taking a mid-range test will have a handful
of lucky people getting a perfect score.
Ron has argued that there is something wrong with the simulation
because the raw scores do not show the same bell curve that the ETS
publishes for the SAT. This is because
the ETS smoothes the raw score into what they call the “scaled” score,
which they purposefully ensure has a bell curve distribution. This paper contains details of this
non-linear scaling: http://www.ets.org/research/dload/RR-02-04.pdf. Incidentally, this paper contains graphs
of scores that look a lot like my simulation. To repeat the result of the simulation: the
most intelligent person has less than a 1% chance of getting the highest
raw score. No smoothing scheme can
fix that. The SAT is not a high-range
- If the
Society did not have another method for admission, it might be argued that
the Society will die off if we don’t find some way of admitting
people. But we have recently voted
to accept the Titan Test for admissions.
If Brian or anyone else wants to qualify for membership, they
should take the Titan Test. That
(or something like it) is how the rest of us were admitted.