One of the latest members of the high-IQ club Mensa is a mere 4 years old, with an IQ of 159 — but psychologists warn against pulling out the Albert Einstein comparisons just yet. ... "All you're doing with IQ testing is testing within a certain age group," Lawlis told LiveScience, explaining, "You're saying the 4-year-old is smarter than 99.5 or 99.8 of [her] age group, but that doesn't mean you can compare to another age group."
I'm a little confused here. It seems to me that the biggest issue with IQ tests given to very young individuals is expected variance in outcomes across tests. Raise your hand if you know a moderately bright person who "tested off the charts" as a very young person. If the charts are such that people test "off them," I am skeptical of the precision of the tests being used.* My own hunch when someone tells me their IQ is that if they are moderately gifted they are reporting from the high range of distribution of scores they've received. Many gifted children have taken a fair number of standardized tests, so it isn't too difficult to sample from the higher of the range of outcomes. 90 percent of people think they are better looking than average, and 90 percent of people report they are more intelligent than average. Why should we credit the cognitively gifted set with much greater self-awareness of the usually unconscious bias of overly positive self-assessment? (yes, I know of the Dunning–Kruger effect) If someone took the SAT three times, do you think they'll be reporting the average of their outcomes, or the best of their outcomes (probably the last time they took it)? Which gets to my title: from what I recall Mensa allows you to keep taking IQ tests as long as you want (you may have switch tests though, but that's not too hard, there are plenty of them out there to select from). If you are within spitting range of the top 2 percent you will eventually cross the threshold simply by taking enough tests because the results are somewhat noisy. An individual whose average score is in the 5th percentile across tests may still find a test which allows them to break the 2nd percentile floor for Mensa. I suspect that measurement error of this sort is at the heart of Richard Feynman's legendary "low IQ" by the way. He was smart. He had to take lots of tests. * Personal story, when a friend told me at one point that they'd been tested at a 150 IQ when they were young, I unwisely replied that it was obviously measurement error, and their real IQ was probably closer to 140. I knew their SAT test scores, and I also knew the person well enough to gauge their intellect. My friend did not press the point after I told him baldly that he was surely reporting his highest score. But I don't recommend this course of action to people who are interested in maintaining friendships. A many an ego has rested uncomfortably within the house which measurement error has built.