In reality it applies nowhere.
The more people appear in your statistic the more it will even out.
In the end it's going to be even.
Unless you manipulated it somehow which I think this particular statistic is.
Also IQ doesn't tell that much.
Wrong on all accounts! A statistical relation in the form of a correlation, regression or some other association measure does not "even" out with increasing sample size. Two things happen in such a case:
1. The estimate of the strength and direction of the association becomes more precise (the statistical model becomes more accurate)
2. The generalizability of the association from sample to population becomes more robust (lower p values).
Thus, increasing sample size leads to the opposite of what you suggest.
And- IQ is one of the most important and robust predictors of academic and work performance as well as socio-economic status.
NN