Originally Posted By: Eagle Mum
I don't expect any IQ test to produce a 'normal' or Gaussian curve because many factors influence or impact intellectual function. If a measurable outcome is affected by multiple factors, I'd expect a log Gaussian curve. If the magnitude of the effects of these factors on the measurable outcome are very different and there is genetic clustering, I'd expect a nonparametric curve. I suspect that the actual distribution of intellectual function in the general population would predominantly resemble a log Gaussian curve with non parametric waves which would be more evident in the long right tail of the log Gaussian curve. Both of these would explain why there are far more HG, EG & PG individuals than predicted by a test which assumes the population is normally distributed.


As far as I know, deviation-based tests are designed to make the distribution they produce be a Gaussian curve, no matter what the actual distribution of whatever they're trying to measure is. I'm not sure how successful that actually is at the tails, but that's the goal.