ucbalumnus makes a good point about predicted yield. A Harvard faculty member once told me that with regard to graduate admissions (where the students are selected by the department) the % of students they really wanted who actually came there for grad school was lower than the % of those they considered good or admissible. One year, they were worried that they had admitted more students than they had funding for. The faculty member suggested that they should move some of the students into the “really want them” category. Tongue in cheek of course.