Speaking Out
Predicting Performance
In a letter last week, Professor Charles J. McMahon, Jr. said that "student athletes generally performed better than was predicted by the standardized scores." The evidence was data that he later acknowledged to me "pertained only to people who graduated."
I first encountered this maneuver in the mid '60s when we were constructing the admissions policy now known as the McGill Report. At that time it was widely believed that athletes at Penn had higher GPAs than non-athletes. With some effort we were able to extract from the Admission Department's files the lists of students who had been recommended by the coaches and then with even more effort we extracted each of those students' records from the files of the various schools. A very different picture emerged.
Yes, those students who had not flunked out and who had not been barred from athletics because of their grades had higher than average GPAs. The performance of the entire group of student-athletes, though, was dismal.
Since that time I've seen this maneuver resurface about once every ten years. Start with a group of students-athletes, remove all those with unreasonable GPAs and--behold--exhibit a group with reasonable GPAs. Professor McMahon brought a new variation: instead of trying to prove that the athletes' academic performance was above average he tried for a much more modest goal, to wit, that they "performed better than was predicted." The prediction in question was a freshman GPA of 1.61. Professor McMahon's sample, alas, was not the group of student-athletes for whom the prediction was made, it was the subgroup of those known to have, indeed, done better than was predicted.
Professor McMahon said "the majority of these students obtained a final GPA above 2.0" and went on to say "One can only speculate on the reason for these results." There's no need to speculate: the rules of each of the undergraduate schools require a final GPA of at least 2.0 to graduate.
--Peter Freyd, Professor of Mathematics
Speaking Out welcomes reader contributions. Short, timely letters on University issues can be accepted, normally by Thursday at noon for the following Tuesday's issue, subject to right-of-reply guidelines. Please note, due to Spring Break there is no issue March 13; the deadline is extended until March 15 for the March 20 issue. Advance notice of intention to submit is appreciated. --Ed.
Almanac, Vol. 47, No. 25, March 6, 2001