The idea you describe in your second paragraph is to some extent rational, too. For example, a degree from a top college means less for a black job applicant because of the heavy affirmative action advantage given to them.
Are you attempting to assert causation in the absence of correlation?
Suppose that for University X, the minimal SAT requirement for white applicants was 700/700, and for black applicants was 400/400.
Suppose that the SAT scores of all white students at X are uniformly distributed over (700-800/700-800), and those of the black students are uniformly distributed over (400-800/400-800).
Now randomly pick one white student, and one black student.
Which of them is likely to have the higher SAT score?
And which do you want performing heart surgery on your child?
It's a whole lot more "obvious" here in South Africa. There are actual quotas, rather than "score" enhancements and juggling like there is in America.
With the assumption that the "range" of ability is equal in both groups. The simple fact that you take a smaller quantity of one, rather than the other, implies that you're taking a different ability range of one group over the other.
So if you have 100 black, 100 white students, and both their group scores range using a normal distribution from 500-1000. But you enforce that only 10 whites can be taken, and 40 blacks should be taken. Then you're guaranteeing that only the "top" 10 of the white students get picked, while with the blacks the top 40 get picked. This means you're going down the distribution and selecting black individuals that have lower scores.
As for your question about which number drawn at random from two differently distributed populations would be higher on average - yes, in this hypothetical college that admits black students with SAT scores below the black population's average but only takes white students who are 1+ STD above the white population's average, I would expect the white students to have higher SAT scores.
What I would interrogate is why you seem to be defining the worthiness of students by their SAT scores, which most elite colleges now essentially disregard, since studies have shown that it correlates alarmingly well with parents' income.
I also don't understand why so many people, when discussing the merits of college or job applicants, are so focused on the exact point of application, to the point of refusing to consider earlier opportunities and environmental effects.
Wouldn't you consider a 500 SAT score from a low-income family's student more impressive than a 500 SAT score from a high-income family's student, based on the relative availability of SAT prep classes to the two students? Shouldn't such contextual information be considered when evaluating candidates? Wouldn't you say that the candidate from the less advantageous background who has achieved identical "on paper" results is more deserving of selection, considering what they must've overcome to achieve such results? If the candidate from the less advantageous background is more deserving, why do studies find that, when presented with identical applicants, hirers will choose the one with the whiter name? [0] http://www.politifact.com/punditfact/statements/2015/mar/15/...
Perhaps my use of SAT scores in particular was a mistake, but hopefully you came away with a better understanding of garanduss's original post.
Secondly, many elite schools pride themselves on "need-blind" admissions, meaning they do not know the applicant's financial situation when deciding whether or not to admit him/her. This is to protect poor kids from being discriminated against (since they would need extra financial aid from the school), but it also means they can't give a leg-up to poor students.
Given that family income is highly correlated with SAT performance, it's pretty perverse that the best way for a poor student to gain admission into an elite university is to ace their SAT's and masquerade as a rich kid on their application.