Reply to Paul Maxim's Criticism
of the Norming of the LAIT

Kevin Langdon

This letter was originally published in Vidya #147/148, October/
November 1995. It has been slightly revised for publication here.


I have received a copy of Paul Maxim's letter to the Editor of Vidya which contains numerous assertions regarding the norming of my Langdon Adult Intelligence Test. I will respond to the points raised in Paul's letter below.

It is, of course, the case that it is very difficult to establish reliable norms for the extreme right tail of the normal curve. I have learned a great deal since the LAIT was first released, but I still have reservations about the precision of percentiles calculated for high-range tests.

What is quite certain is that there is a huge difference between an I.Q. of 100 and one of 150. The I.Q. scale is useful as a sorting filter. And the LAIT is a good measure of I.Q., in the sense of fluid g.

Estimates of the ceiling of the LAIT have varied from about 174 to 180. According to the official norms, the ceiling is 176, or one in a million. But no test is accurate at the extreme limit of its range. Scot Morris was right when he wrote in Omni that the LAIT is ``most effective in measuring IQ's between about 130 and 170'' (Paul reproduced this quotation). An I.Q. of 170 indicates a rarity of about one in 165,000.

Also contained in the Scot Morris quotation reproduced by Paul Maxim was the statement that the average LAIT testee scored ``just short of 150.'' This was on the first norming of the LAIT, dated January 1978 (N=147). Subsequent norming studies showed that this was somewhat of an overestimate. The average for this sample was actually in the mid-140's. The mean I.Q. as of the second norming (July 1979, N=553) was 141. A sample of 20,000 testees (the vast bulk of them from Omni) had a mean I.Q. of 137. The difference in means reflects the fact that the initial sample was largely drawn from a population consisting of ISPE members and selected Mensa members.

Paul wrote:

By July 1979, Mr. Langdon reported (in his ``LAIT Norming Report #2'') that he had scored 553 LAITs to that point in time. But then, due to computer problems, he fell behind in scoring the LAITs which were being sent in by OMNI readers--a circumstance which utimately led OMNI to file a lawsuit against him in 1982.

This is not quite accurate. There were long delays in scoring the LAIT after its publication in Omni because neither Omni nor I had anticipated the over 20,000 answer sheets that were submitted during the first few months after publication of the test in Omni.

Shortly after I had finally caught up with the backlog, in mid-1980, scoring approximately 23,000 tests, the computer that I had purchased to score the LAIT suffered a catastrophic breakdown. I was unable to score any tests for several years, as I had no money to replace the computer (the scoring fees for the LAIT had been consumed by expenses, particularly the labor of the people that I had opening mail, entering answer sheets, and mailing score reports), and a backlog accumulated again, though a considerably smaller one.

It was during this period that Omni sued me for a million dollars. I was later able to purchase a new computer, score a backlog of about 2500 tests, and catch up again. I settled the lawsuit with Omni by scoring these tests and supplying Omni with a list of names and addresses for verification that the scoring had actually been completed; I did not have to pay Omni a cent.

Paul wrote:

Mr. Langdon recently stated . . . that his Four Sigma Society reached a membership peak of 250 in 1980. I do not know exactly how many LAITs he had scored by that point in time, but by way of comparison, it should be noted that ISPE, a 3-sigma group, had 150 members in 1980 and fewer than 100 in 1979. In other words, even though 4-sigma IQ's are thirty times rarer than 3-sigmas in the general population, Mr. Langdon claimed to have recruited more 4-sigma individuals in three years than the number of 3-sigmas ISPE had enrolled in six.
Although Mr. Langdon has not disclosed the number of LAIT tests he employed to arrive at his claimed ``250'' qualifiers, I estimate that (by 1980) it cuold not have exceeded about 2,500, and might have been considerably less. This means, in turn, that Mr. Langdon is claiming (or attributing) a 4-sigma IQ to more than 10% of his sample--an incredibly high figure, considering the ``one in 30,000'' average incidence of 4-sigma in the general population.

The figure of 250 was not the total number of Four Sigma members but the number of members who responded to the Four Sigma membership survey of 1980. (There were about 150 subscribers to Four Sigma's journal, Sigma Four). Approximately 600 people have made four-sigma scores, out of about 27,000 who have taken the LAIT.

There are two reasons for these large numbers. One is that I had a far greater sample, from Omni's circulation of nearly a million, than the ISPE was able to draw from. The second is that there is a very significant self-selection factor in submission of answer sheets for a difficult high-range test which takes many hours to complete, offering ample opportunity for those who are not doing well to abandom the project. The percentage of four-sigma scores is closer to 2 or 3 percent of testees than to 10 percent.

Paul wrote:

As regards the category of ``OMNI readers who take high-IQ tests,'' the anticipated incidence of 4-sigma scores is even lower, based on the estimated IQ for such individuals of 127 (please see OMNI, May 1993, p. 94, Col. 2). In other words, about one thousand people must be tested, to arrive at the expectation of one 4-sigma score.

The test on which the average score of Omni participants was 127 was the Quest Test, by Daryl Inman. The Quest Test has severe problems and, to my knowledge, has not been accepted for admission to any high-I.Q. society except Ron Hoeflin's Top One Percent and One-in-a-Thousand societies; Ron has explicitly stated that he is more interested in increasing the membership of these groups than in adhering to strict membership standards.

The items comprising the Quest Test have been the subject of devastating critiques by many different authors in In-Genius, the journal of the Top One Percent Society, and Noesis, the journal of the Mega Society. There is clearly a very high percentage of bad items. Daryl Inman has not responded to any of this criticism.

There are even more serious difficulties with the norming of the Quest Test. As Daryl Inman does not have a background in psychometric statistics, Chris Harding performed the norming study that resulted in Quest Test scores. As with the norming of his own tests, Chris has not provided a detailed description of the procedures used in this norming (as Ron Hoeflin and I routinely do for our tests). It is suspicious, to say the least, that the highest I.Q. among almost a thousand Omni readers was only 160, though this number matches the theoretical expectation if the mean of Omni readers were actually 127 and there were no self-selection factor.

The average I.Q. of Omni readers who completed the LAIT (on both reported previous scores and the LAIT itself) was 137. The average I.Q.for the Mega Test was approximately 140. The difference in mean score between the Quest Test and the LAIT and Mega Test is significant, as it indicates a difference of a factor of five in rarity.

In a table on page 5 of Ron Hoeflin's report on the sixth norming of the Mega Test, Ron reported that 120 people, out of 3920, had scores at or above the four sigma level; this is a slightly higher percentage of four sigma scores than I obtained with the LAIT.

It is apparent that Paul has not done his homework. The numbers he has used in his calculations are wrong and he has not considered the self-selection factor.

Paul wrote:

Since I was not affiliated with any of the ``super'' high-IQ groups during the time period aforenoted, I have no idea of whether Mr. Langdon's announced results were subjected to any scepticism during that period.

The LAIT norming reports (with some additional data, in some cases) have been reviewed by a number of independent investigators, including Ed Van Vleck (one of the founders of TNS), Fred Britton, Grady Towers, and Alan Aax. All have found the LAIT to be a valid test with a high loading on fluid g.


[Intelligence: Essays and Reviews]

[Consciousness and Mind]

[Polymath Systems Home Page]