Though this implies that 10% of the 4.0s don't deserve it, it's not proof; a single cheat may or may not have affected the grade.
It wouldn’t actually change the result of the metric (because survivorship bias: you don’t count students who never graduated in the graduated student population), but it changes the believability of and the ethic behind the metric. Now we can say that at least 10% of the folks who graduated didn’t earn their grades, and the school’s reputation is less for it.