The normal distribution is baked into the scoring of intelligence tests. I do not know what the distribution of raw scores looks like, but the calculation of the IQ score is done by transforming the raw scores to make them normally distributed with a mean of 100. There is surely not enough data to do this transformation out to ±6 SD.
Why assume Gaussian?
The normal distribution is baked into the scoring of intelligence tests. I do not know what the distribution of raw scores looks like, but the calculation of the IQ score is done by transforming the raw scores to make them normally distributed with a mean of 100. There is surely not enough data to do this transformation out to ±6 SD.