Psychometric Issues

A number of questions about the Mega Society's admission standards have been raised recently. In this issue and the next, #141, correspondence on this subject published in Noesis over the past two years and more recent letters on these matters from a number of Mega members and from Grady Towers, one of the most psychometrically-knowledgeable members of the higher-I.Q.-societies community, have been collected. I have not included Paul Maxim's self-serving criticisms of the LAIT and the Mega Society's admission policies. I have been highly selective with regard to Chris Langan's remarks, because little of what he has written addresses the issues with which we are concerned here.

One of the differences between the Mega Society and other high-IQ societies is that questions regarding psychometrics are treated as of general interest rather than being considered by a Psychometrics Committee; most of the authors of the major high-range, scored-by-mail IQ tests are Mega members, as are several other people with a background in psychometric statistics.

Many questions have been raised about the basis of our organization, among them questions about what our qualifying percentile should be and what tests are suitable for discriminating at the target level. A number of pertinent questions have also been raised on the Prometheus Society's "fire" listserv (e-mail list). (A number of non-members of Prometheus are on this list. For information, e-mail the list co-ordinator, Fredrik Ullén [e-mail address; Fredrik is no longer the list co-ordinator; see Prometheus' Web site, http://www.prometheussociety.org])

At the present time, the tests we are using reach ceilings in the vicinity of the fifth standard deviation from the general population mean. Scores above about 4.5 sigma are subject to very large uncertainties, which poses a problem for us if we are to adhere to our nominal entrance standard (1 per million; 4.75 sigma).

The society is under outside scrutiny as has never before been the case. We will be expected to define what we claim to be. At this point we must ask ourselves what we can justifiably claim, given the present state of the art in high-range testing.

One contingent maintains that we must accept the attempts in two states to prohibit the tests we use for admission to Mega. Various suggestions have been made for other means of testing applicants (e.g., "chronometric" [reaction time] tests, creative use of the Ferguson formula), but the truth is that there aren't any other adequate methods available today. There is no evidence that the states of California and New York are interested in what the high-I.Q. societies do, as opposed to what is done by the people who administer the tests by mail without benefit of a license to practice "psychology" issued by the state. But if they should attempt to interfere with our First Amendment right to select members any way we choose, then, in the interest of continuing to exist, we must not allow these bureaucrats to dictate to us, though it might be prudent, at some point, to remove certain operations from jurisdictions where the authorities take too much of an interest in our business.

Others want to maintain our 99.9999th-percentile entrance criterion without addressing the question of what instrument is to be used to discriminate at this level.

There are partisans of the curvilinear fitting technique used by Ron Hoeflin in norming his tests, but this type of fitting is on questionable ground statistically. Ron Hoeflin now agrees with me that claims that the highest scores on these tests indicate ability much beyond five sigma do not have a solid foundation, given our present state of knowledge.

I have suggested that we lower our sights to the one-in-300,000 level, 4.5 sigma, which seems to represent a practical limit to where we can credibly claim to discriminate given the available psychometric instruments.

Another possibility is the development of harder tests. Alan Aax' Eight Item Test shows that this is possible, though it's too short and it hasn't been normed.

If there are other views, they should be made known also. What do you, the members of Mega, think about these issues? We need to make some decisions about our admission criteria. Clearly, everyone's opinions should be heard before we take a vote on these matters.

Your views are solicited.

Send material for Noesis #142 (hard copy only) to Dr. Ronald K. Hoeflin, P.O. Box 539, New York, NY 10101. Send corresponding electronic files (ASCII or HTML, please) to me if you're interested in having your material made available on the Web.

[About the Mega Society]