If there is a positive correlation between admission standards and graduate ability, what you have done is lowered the average but left it above the bar (assuming that the bar isn't curved).

There is still the question as to whether certain admission standards are actually correlated with graduate ability. I wonder if there is a way to find out?