The Unpleasant Ethics of Capping Grades

I am adamantly opposed to the recently proposed faculty rule that mandates a maximum number of A’s in any given course. My opposition is primarily based on three ethical issues.

First, as teachers, are we not striving to be so effective and motivating that all of our students learn and internalize the entirety of what we teach? Is that not our ultimate goal? And if we should succeed, then how do we distinguish the students who will get A’s from those who won’t? If we succeed in our Quixotic goal of masterful teaching, a cap on A’s would require an artificial, capricious choice of which students get the A’s and which don’t.

Second, the grade cap proposal asks us to treat our students as statistics, not as unique individuals. My very central ethic (perhaps yours too?) says to treat each person as a unique individual. As a unique individual, a student should be graded solely on the merits of their own achievements and what they have learned — not on the statistics of one or another cohort.

Third, a grade cap erroneously assumes that grading is always exact. But grading is never exact. There is always noise in numerical grades: from misunderstandings, subjectivity of the graders, and the inherent inexactness that comes from turning a qualitative notion of mastery into a quantitative score. We obfuscate this imprecision by adding up numbers from assignments, exams, and quizzes. Noise is already immensely magnified by the binning inherent in a letter grade system, which assigns very small numerical differences to distinct letter grades — and a strict cap will make those cutoffs even more arbitrary.

Ponder this: What do we do in a large class where differences of tenths of percentage points will be magnified by a factor of 75 by letter grade assignments? This is especially troubling given that those tenths of percentage points are often due to different graders — or simply a grader in a particularly snarly mood one day and not the next. Because grading is never exact, it should err on the generous side. A grade cap renders this impossible.

That said, Harvard has a legitimate grade inflation problem. Instead of arbitrary, top-down grade caps, the University needs a sensible, coordinated effort to fight it. What follows is a suggestion with two parts: Part one is for the administration. Part two is for the teachers.

For administrators: The University should stop using raw student feedback statistics to judge faculty retention, promotion, salary or hiring. Under the current system, students are asked to rate their instructors on a scale from one to five. And as October’s grading report points out, the use of such metrics is a major driver of grade inflation: It strongly motivates instructors to design workload and requirements to earn fives rather than to cultivate mastery.

A better feedback system would ask students only for written feedback on their classes and professors. Such written answers are actually useful for teachers — and should also be useful for administrators if they can take the time to read the responses. Simple numerical feedback is useless for teachers and incomparable across classes — and its use by deans and administrators is destructive (and a misuse of the statistics).

For instructors: Any instructor who believes their course has “too many A’s” should make their course more demanding. Teach more and deeper! In any event, course syllabi should make clear what level of work and achievement is required to earn an A. In contrast with the vague language — “extraordinary distinction” — distributed to students and faculty, instructors should set out clear expectations. Crucially, grades must depend only on each student’s achievement and that achievement alone, not on some comparison measure against the rest of the class — which is not only unethical, but a bar whose height is unknowable until the course is over.

In this regard, the distinction between A and not A should be robust in two ways. First, the distinction should not be sensitive to a sneeze by the student, the graders, or both. Second, a reasonable student, when assessing their own work, should agree that their given grade is a fair assessment of their achievement in the class. These robustness criteria for grading are the hard part of assessing student performance. To my mind, a grade cap is a shirking of this duty. With a robust distinction in place, if, at the end of the term, all students achieve the highest level, then all should receive A’s.

What if the proposed grade cap comes to pass? I will ignore it and give A’s as I see fit. And, if it turns out that the Registrar rejects my grade sheet because of “too many” A’s, then I will let the kind and gentle folk at the Registrar’s Office choose which students get the A and which do not.

Clifford H. Taubes is the William Petschek Professor of Mathematics and the Math Department’s Director of Undergraduate Studies.