I am hearing two questions here:
1) is the distribution of high-IQ individuals the same in adulthood as it is in childhood?
2) will children identified as high IQ continue to be identified as such in adulthood (stability of IQ over the lifespan)?
Answers:
1) Yes. Contemporary IQ scaling is deviation-based, which means that, by definition, there will be 2% of the population at any given age that can be measured as 130+ IQ (the 98th %ile, on most cognitive assessments). This is purely a reflection of the bell curve and the test-norm development process. (Granted, there is discussion as to the applicability of the bell curve to the distribution of intelligence, as to many other human attributes.)
2) Mostly yes. The research on IQ is that it is reasonably stable from middle childhood on, but with regression to the mean, scores that are further to the right tail of the bell curve are likely to fall more on retest than those that are hovering around the mean (which are likely to continue hovering around the mean). Longitudinal studies have also found that unusually low IQs (on the left hand tail of the curve) are more stable than unusually high IQs. So it appears that the regression to the mean of outlier IQs is greater for high than for low scorers.
So to the OP's question about kids qualifying for Mensa, but then scoring below the qualification cutoff as adults: yes, that could very well happen, especially if the original qualifying score was close to the boundary. (Notice the GAI exemption does not apply to the WISC-V.) It does not necessarily follow, though, that in 10 years Mensa will consist of the top 20% instead of the top 2%, partly because not everyone who is eligible applies. (Every tested member of my FOO has qualifying scores, all by wide margins, but none of us have felt any particular inclination to join. Nothing against the organization, or anyone else joining. Just no interest.)
Frankly, the question of questionably qualified members of Mensa is probably affected more by the range of qualifying tests that are allowed than by regression to the mean. I've found pretty different numbers for the same children across combinations of the SBLM, SBV, WISC-IV, WJIII, and RIAS. Let alone the CogAT and OLSAT. A motivated person with the resources and borderline eligibility could probably test-shop until a qualifying score was obtained.
The norms for the SBLM are over four decades old now, and based on a completely different structure (mental age, rather than deviation IQ), both factors which generally result in higher numbers on the LM than on contemporary IQ tests. Mensa still accepts SBLM scores; I didn't see a comment anywhere about when the scores were obtained (perhaps others know otherwise), so one could, hypothetically, bump borderline qualifying scores over the threshold pretty easily by having an SBLM administered. (If you can find someone to do it--though Linda Silverman still likes it, so obviously some legit psychs are still making a case for using it.)
Last edited by aeh; 01/07/16 02:48 PM. Reason: typos
