These types of studies always miss the point - the difference in entry input (SAT, GPA, class rank) do not necessary result in a proportional difference in output (reputation in a variety of fields).
As an example, if you assume everyone who applies at Harvard would be equally successful, you would want those who would have the most broad range of outcomes to boost your schools reputation. In other words, with a class size of 100 you would not want 100 famous scientists - you'd want famous scientists, politicians, writers, etc.
So as a result you'd have to control for perceived interests among the committee that reviews applications.
Nevertheless, discrimination is definitely in poor taste.