So we can't look at the statistical distributions of students and the entire population to see whether individual groups have statistically significant advantag

So we can't look at the statistical distributions of students and the entire population to see whether individual groups have statistically significant advantages? Why not? Even if we can't get to 100% fairness, we can move away a bit from unfairness.