This is pretty simple. A highly conscientious person outperforms their native intelligence in academics, and vice versa, and both are well-formed normal distributions. We can reasonably expect that a person with scholastic achievement two standard deviations above the mean will have scholastic aptitude somewhere around two standard deviations above the mean. Given a large number of individuals with +2SD IQ (that is, holding IQ constant), I’d expect their subject SAT scores to form a bell curve around 720 (mean of 500 and SD of 110).
This assumes that IQ and conscientiousness do not affect one another, which is obviously not the case, but the effect is smoothed out in practice. If +2SD IQ grants a lower time preference, it will do so for all +2SD IQ folks. It is then the difference in relative work ethic which creates the bell curve around 720. We can then describe a person’s conscientiousness relative to their IQ cohort, which effectively controls for IQ. If the +2SD crowd typically scores 720 with a sigma of 50 (I made this up, although I’m sure the numbers exist in a chart somewhere), then a 135 IQ chap who scores 770 could be described as having a +1SD conscientiousness quotient within his IQ cohort.
I’ll use myself as another example. I don’t remember my scores very clearly, but my ACTs, SATs, PSATs, and so on tended to cluster around the 90th percentile, or roughly 660 on an SAT subject test. As I’ve laboriously estimated beforehand (unwilling to pay for a proper test), my IQ is probably right around the 99th percentile cutoff of 135. If the made-up standard deviation of 50 points above were correct, this would put me at the -1.2SD achievement level for the 135 IQ stratum. That’s around the 23rd percentile.
It bugs me a little bit that we don’t have a primary test for conscientiousness like we have for IQ, because SAT scores depend more on self-discipline than they do on intelligence. Google it if you need to. But this will serve for the time being.
One of these days, I’ll have to throw together a chart of standard deviations on SAT scores for each IQ centile. I checked the NLSY, and it has both variables. Now I just need a few extra hours to remember how to get the data into R and play around with it.