Response to "Are not your norms too high?"

I.Q. Tests for the High Range

The long response

Every once in a while, candidates or other interested individuals come with questions like "Are not your norms much too high? Is it not much too easy for people of average or below-average intelligence to score in the I.Q. 130+ range? Should you not let more people of average or lower intelligence take your tests to get better norms?"

The answer to all of these is "no". A number of things have to be said about this:

Firstly, questions like this stem from an intuitive assessment of the difficulty of the test problems and of the ability level of people of average and below-average intelligence. Such intuitions tend to be detached from reality due to the phenomenon of psychological projection; one involuntarily and unknowingly assumes one's own ability level in others, and thus overestimates the ability level of those of lower intelligence than one's own, and underestimates the difficulty of the problems one has been able to solve. This tendency is extremely persistent, to the point that one keeps doing it even when it is pointed out to one, such as in the present paragraph. The test norms, on the other hand, are based on objective data, not on subjective intuitions, and thus may appear too high to those who suffer from this projection.

Then, one is reminded that high-range tests are meant for people who score in or near the ceiling on mainstream I.Q. tests, and attempt to distinguish among those people. In effect this concerns mostly the top one to two percent of the general population, with their mode falling just above the 98th centile. Still, a minority of high-range candidates is of lower intelligence than that, simply because no one is keeping them from taking the tests. And a very small proportion of high-range candidates is around or below the average of I.Q. 100. This last group is actually increasing as a result of the ever greater ease of use of devices with Internet access. From the viewpoint of high-range intelligence research, there is no need to include more people of average or lower intelligence among the candidates. Having some of such candidates may indeed help to obtain good norms in the bottom range of high-range tests, but measuring instruments will always be less accurate near the extremes of their range, and high-range tests were never meant for this group.

And do those with I.Q.'s around or below 100 indeed score too high, as some believe? Do they easily end up with I.Q.'s of 130 or more? What really happens is the following:

When someone from this I.Q. range (100 or lower) succeeds in ordering a test (which is a moderately high hurdle in itself) and subsequently succeeds in opening and viewing the attachment containing the test (which is not at all obvious at lower intelligence levels) such a person tends to be utterly stumped by its contents and understand almost nothing of it. The person may either discard the test altogether, or embarrassedly ask for a refund, or ask for clarification of the test. When the test scorer refuses to give clarification (which would give the candidate an advantage over all other candidates, after all) the person may again discard the test, ask for a refund, or eventually send something that resembles a test submission.

Invariably, these submissions result in low scores, mostly zero raw. It is also observed that these candidates are likely to not report the requested personal data (although that happens with high-I.Q. candidates too) let alone that they would be able to submit the test registration form. Then it is observed that these candidates are likely to not understand the reported score, as in not understanding that a raw score of zero means that they solved nothing and are below the measured range of the test. Some understand from the report that they actually have a very high I.Q., and explaining otherwise to them tends not to repair this misunderstanding. They are also likely to require a test or report to be resent frequently due to their inability to save and/or retrieve a file on their device. They are also relatively likely to disagree with the score, displaying abusive and offensive behaviour in the process. They are also likely to possess poor written communication skills, as in writing incoherent rambling in all lower case without punctuation and without including their name in the correspondence. They are also likely to not give meaningful responses to simple questions like "What is your name/age/nationality?" Not because of privacy, but because they do not understand what is being asked, are cognitively overstrained. All in all, dealing with these candidates involves more work, correspondence, hostility, abuse, and stress for the scorer than does dealing with high-I.Q. candidates. No test scorer is dying to have more of these candidates, although having some of them does help to prevent the norms of the lower raw scores from being too high. An attentive test scorer would most definitely notice it if candidates of average or lower I.Q. scored in the 130+ range.

The above observations are not given here to ridicule these people, but to illustrate the discrepancy between the intuitive notion of "it is much too easy for people of average or lower I.Q. to score 130+ on your tests" and the actual state of affairs.

Then it must be asked how one can know if a person is in this intelligence range. This can be known by interpreting information as to possible reported test scores, educational achievement, job history, and personal history of the person, in combination with observation of the person's communication and further behaviour. Some observable characteristics of low intelligence are listed in this article. One may also ask the person questions to gain more insight into the person's personality, or have samples of text produced by the person analysed by something like an online Writing to I.Q. estimator. The lower someone's intelligence, the less usable information will be actively provided by the person, and the more one has to rely on observation.

As an aside and to avoid misunderstanding: For software products delivered by electronic mail such as these tests, there exists no legal obligation to refund the fee in the event the client is dissatisfied or regrets the purchase, because the product can never be truly returned (a copy would be retained on the client's electronic computer). In practice, refunds are very occasionally given out of leniency, and under certain circumstances PayPal will bluntly take the money back from the vendor, such as when the buyer's account was supposedly hacked at the time of the purchase.


So, norms are not systematically too high by objective standards, but may be too high as judged intuitively and subjectively (test creators and scorers themselves may have this intuitive and subjective notion too). Candidates of average or lower intelligence are not scoring way too high, like in the 130+ range. Also there is no need to have more people from that group take the tests, because (1) high-range tests are not meant for this group, (2) enough people from this group are spontaneously taking the tests to have some confidence that the norms in the lower range are not much too high, and (3) candidates from this group cause more work, correspondence, stress, and abuse.

This does not mean that high-range tests are necessarily free of problems or biases, but too high norms in the lower part of the range are not the essential problem. Speculatively speaking, and to play the Devil's advocate, a more fundamental phenomenon may be that high-range tests, being administered by correspondence, inherently focus on theoretical, abstract-logical abilities and lack practical tasks involving motor coordination, dexterity, timing, oral communication (speaking), dealing with people, and other performance-type, executive abilities. This may not be a problem for candidates with even aptitude profiles, because the contents types that are included in high-range tests happen to be those with the highest g loadings so that those with balanced profiles will have their intelligence measured correctly still, but this will cause bias in rare cases of sharply uneven profiles:

Those who are good at the lacking types of tasks (practical, social, performance) but bad at theoretical abstract-logical matters will do poorly on high-range tests, but may succeed in real life, have jobs with good incomes, get married, and form families with many healthy children. Those who excel in the abstract-logical field while being exceedingly clumsy in practical and social matters, on the other hand, will score highly on high-range tests, but fail miserably in society, may not be able to get or hold jobs, may remain unmarried, and may never form families and procreate (this is purely theoretical of course, not to imply that we know any such people). These two examples of uneven profiles are caricatures, in reality there is no dichotomy and most people have more even profiles, but the second group will necessarily be overrepresented among high-range candidates. A truly sceptical critic might even suggest that the high-I.Q. community be largely composed of members of the second group, but this is probably an exaggeration, as many with balanced aptitude profiles appear present among those of high intelligence, up to the highest levels. These last two paragraphs are speculation, not supported by sufficient proof or evidence yet, and unravelling this matter is one of the goals of high-range psychometrics.