Chris [Cole] wrote: ``The old Mega Society (whether intentional or not) had the same entrance requirements as the current one, namely, a score at the one-in-a-million level on suitably normed high-range tests.'' That was the intent, though it isn't possible to discriminate accurately at the one-in-a-million level using any existing test, in my opinion; the present state of the art in high-range psychometrics does permit discrimination a few points below this level. The existing admission criteria go beyond generalities, specifying the two tests accepted for admission and the score accepted on each.
. . .
In an open letter to Chris Cole and Rick Rosner, Mr. Maxim wrote:
In the third paragraph on page 29 [of Noesis #121] in Jeff Ward's letter, he wrote as follows: ``Regarding the enclosed two requests for admission, my feeling is `no' for ---- and `yes' for PAUL MAXIM.'' At this point, you blacked out not only my name, but the entire following line, which must have aroused some curiosity on the part of the readership. Enough material came through, however, to indicate that the ``no'' individual submitted a score on a Langdon test. In the final line of his letter, Jeff Ward wrote as follows: ``On the other hand, Maxim's score does appear to be in the Mega range on a valid, objective test.''
I am now questioning whether the Editor or Publisher, or both of them acting conjointly, have the right to censor and suppress the remarks of the Society's Executive Officer on an important issue, and whether this does not represent an unreasonable infringement by them on the Executive Officer's prerogatives?
. . .
The wording of Jeff Ward's letter may lead to some doubt about what standards have actually been applied in admitting members. I have been informed by Chris Cole that no one has been admitted to Mega since the merger on any other basis than LAIT 175, Mega 43, or Titan 43. The only pertinent question is whether the applicant has submitted a score which accords with the qualifying scores established by a vote of the members of Mega (although a qualifying score on the Titan Test was not set by the recent membership vote, the Titan Test is comparable to the Mega and appears to be a little harder, so using 43 shouldn't get us into any more trouble than we're in already).
In this same letter, Mr. Maxim expressed the opinion that the 4.75-sigma level is reached on the Mega Test at raw score 46. My own analysis of Ron's Mega Test data and one of Ron's norming studies (as Mr. Maxim noted) also places the one-per-million level at 46. (This is not to say that the Mega Test can discriminate reliably at this level; it's too close to the test ceiling.)
Mr. Maxim's argument that members have been admitted with scores below the 4.75-sigma level, though, is irrelevant to his own application for admission. It may not make sense to lock the barn door after the horse is stolen, but anyone who doesn't take precautions after some of the horses have been stolen is a damn fool. A society which has made admission mistakes in the past can tighten up its policies and wait for attrition to bring the actual composition of its membership into accord with its theoretical standard, but those whose good-faith applications have been accepted by the duly elected officers of Mega, acting in accordance with the society's estab-lished admission policies, cannot later have their memberships revoked--not ethically and not legally, either. (Note that Mr. Maxim was ``admitted'' to Mega by a non-officer acting unilaterally in violation of the admission policy established by vote of the membership--and thus Mr. Maxim is not a member of Mega at all.)
. . .
In a letter in Noesis #129, Ron Hoeflin observed what he characterized as a discrepancy in my position on straight-line vs. curvilinear fitting of I.Q.'s to raw scores. Ron pointed out that, while insisting on straight-line fitting, I was willing to allow one point on the LAIT and the Mega as a ``ceiling bumping effect.'' This is something a little different. A diagram will illustrate this better than a text description:
| o | |||||||
| o | o | ||||||
| o | o | ||||||
| o | o | o | |||||
| o | o | o | |||||
| o | o | o | o | ||||
| o | o | o | o | ||||
| o | o | o | o | o | |||
| o | o | o | o | o | |||
| o | o | o | o | o | o | ||
| o | o | o | o | o | o | o | |
| o | o | o | o | o | o | o | |
| o | o | o | o | o | o | o |
[<-- mean of distribution] [ceiling]
I regard it as insupportable to claim to be able to make a finer discrimination between raw score points very close to the ceiling of a test, where there is much less data, than at mid-range. Our activities are coming under public scrutiny and it is incumbent upon us to draw conclusions with the parsimony expected of those engaged in serious scientific work.
On the other hand, the LAIT and the Mega Test (and other tests from the same sources) are all we have; the ceilings of the standard tests are significantly lower. The number of persons included under a one-point ceiling-bumping allowance does no more than smooth the extreme right tail of the graph above to provide us with a small pool of probably-qualified candidates--but this is all that can be said about someone who scores 46 on the Mega Test, 99.9999th-percentile by the reckoning of all parties who have expressed opinions on this matter; there isn't enough top to discriminate accurately this near the test ceiling.
. . .
Chris Cole's ``High Range Tests,'' in Noesis #133 [reprinted in this issue], clearly explained why the new high-range intelligence tests are the only appropriate instruments for attempting to select members of high-I.Q. societies with cutoffs above four sigma. He also provided a powerful argument (many too many very high scores) against accepting childhood scores such as those which Mr. Maxim thinks entitle him to a place in the Mega Society. There are additional arguments against the childhood tests, including low correlations with adult scores (not too serious a problem when the child is as old as ten), low g loading, and poor statistical methodology.