Quomodocumque

The Harvard faculty is proposing to hard-cap A’s at 20% per course, leading to an undergraduate crashout the good old Yard has not seen the likes of since,the dining hall discontinued the Chickwich.

I am in cold hard print telling people not to worry about grade inflation. I think I was right about grade inflation as it was in 2002. The situation of grading at Harvard now, on the other hand, gives even me pause. Half the class having GPAs of 3.8 or above is probably too much.

Still, the current argument leaves me with nobody to really root for. When a Harvard student says “We pay to go here to get the product, which is to have a better signal of performance… If you’re just lowering that for everyone, then you’re just lowering the value you provide as a business for the same cost, even while raising tuition year over year,” I just want to cry in my soup a little bit. Is Harvard a product and a business? Sure. But is it just that? And is the product being purchased really “a better signal of performance?” Another undergrad has a slightly different spin on the new policy, and on the point of college: “It misses the point of college, which is to network, go out there, have fun…It would create so much pressure where life wouldn’t be worth that much to live.”

Again: is that the point of college? Is it, specifically, the point of Harvard? It is a point of Harvard, that I’ll grant you. I networked there. I had fun. I believe I probably did whatever is meant these days by “going out there.” But I also went to class, and made a point of really trying a lot, both in the areas that would become my job and those (painting, poetry, Biblical interpretation, the politics of the Chinese Cultural Revolution) which weren’t and were never going to be. That, too, is a point of Harvard.

Anyway: this hubbub is all a bit rich considering the Monterey Jack-level mildness of the proposed policy. The faculty wants to limit the proportion of A grades to 20% — or more precisely, 20% of the class + 4, so small seminars could still have a solid complement of A’s. (Maybe this entire thing is secretly a plan to incentivize grade-grubbing undergrads to take a seminar once in a while?) But no other grade would be limited; a professor would be free to give 80% A- and 20% A if they saw fit. And 20% A’s isn’t a return to the pre-Vietnam gentleman’s C era; it would be roughly in line with grading practices in 2006, at which time, as far as I know, life at Harvard was considered worth living.

But I don’t think the faculty’s document supporting the plan is very convincing either. The ostensible problem to be solved is that the present system offers no way for really excellent students to distinguish themselves, that grades have ceased to offer any meaningful signal at all. Is that true? That median 3.8-ish GPA translates into about half A’s and half A-‘s in your 32 college credits. The difference between straight A’s and 16 A’s and 16 A-‘s is actually pretty noticeable on a transcript!

You can actually see this in the faculty’s own dataset. They propose using APR (average percentile rank) rather than GPA for internal awards. That just means you get a score that averages your rank in each class; if you get the median grade, you get 50, whether that grade was a B or an A- or an A. Look, they say, this is highly correlated with GPA, but no grade inflation!

Well, sure — but you can read that same correlation to say maybe GPA isn’t doing such a bad job capturing academic variation among the students!

The report makes much of that vertical line at the far edge — students with 4.0 averages, who are indeed not distinguished from one another by GPA. How are you supposed to know who’s the best of the best of the best? I’m not moved by this argument. The true academic standouts are the ones who are easy, not hard to tell from the grade-grubbing masses. They do things GPA doesn’t measure, and faculty know they did these things, and they write about it in their letters.

(In passing: the faculty also complain that “the shift of the summa cum laude cutoff so close to 4.0 that summa eligibility now depends on GPAs carried out to five decimal places.” How can this be? Getting 32 A’s is a 4.0. Getting 31 As and an A- is just about 3.99. I only need two decimal places to tell those apart! Are students taking tens of thousands of courses and getting an A- in just one? Update: My source inside Harvard makes the good point that 31 As and an A- and 32 As and an A- give much closer GPAs; but that still only gets you to the third decimal place.)

The report also makes the claim, several times, that a broader grade range would improve “fairness” in the overall academic assessment. Would it? I can see the argument that the relation between GPA and academic success would be less noisy if the grade range were broader; but I’m not sure how much. More compelling to me is their argument that narrow GPA range gives an advantage to students with strong networks who have access to back-channel endorsements orthogonal to GPA (“A fine touch with the epee, that one, just the sort of man for the job.”) I can look at a 3.8 and 3.9 and see a measurable difference in how well the student did in classes, but to the outside world I will concede those numbers look pretty damned close.

Still, I can’t help feeling that the motivation for the policy is simpler. Professors don’t really believe signs are arbitrarily assigned to the signified. We have a sense of what an A means, and don’t want to give an A for work that’s not A work. But somehow we’re embarrassed to just say that; we have to dress it up in concerns about fairness and credibility and get all scatterplot about it all.

One thing that makes me sad: despite the policy difference between faculty and undergrads, they seem to agree on one big thing; that grades are the reason for students to work. The students worry that with scarcer A’s, they’ll have to work too much; the faculty worry that with too many A’s available, the students aren’t working enough. (This article in Harvard Magazine is a really nice, humane rendition of this view.)

Let me be honest about something at this point. When I was a Harvard undergrad I was a huge grade-grubber. I always knew exactly what my GPA was. I once went to office hours to argue with a very patient sociology TA that my final exam essay was better than he thought it was and how could I possibly be getting a B+? (I got the B+.) So I understand the drive for the grade and I don’t look down on it. (OK, maybe I look down on it a little, but in an “I’ve been there too” kind of way.) But I don’t think the grades were why I was working hard in class. I worked hard because I was there to learn. Not everybody who goes to Harvard is aiming to be a professional scholar, I get that. But my possibly sentimental view of Harvard is that it’s where you go because you want to learn things, in every area, with the same kind of depth and intensity as the scholars-to-be of that thing.

Since I spent this post disagreeing with everyone, and since I do more or less agree that the Harvard grading system isn’t ideal, I ought to tell you what proposal I would have written if I were king of the Harvard faculty! Here is, my three-point plan:

1. Bring back the 15-point scale. When I was in college, the GPA was out of 15. An A was 15, an A- was 14, a B+ was 12 (see why I was so mad at my sociology TA?) and so on. Harvard undergrads worry that a Harvard with fewer A’s will leave them with lower GPA’s than their competitors from other colleges, and employers and med schools won’t recognize “Harvard grades different.” Under this plan, they would have way higher GPA than kids from other schools! That’s partly a joke but partly not — a completely incompatible scale would make it clear that there was no direct translation. Many schools couldn’t get away with this, but Harvard can. And one of Harvard’s goofy charms is that they call everything by their own Harvard-y names. Political science is “Government.” Dorms are “Houses.” Majors are “concentrations.” Why shouldn’t a 4.0 be a 15?
2. Establish the A+. This option (which is what Princeton does) is brought up in the faculty report only to be rejected. That’s coming from the “A should mean A!” perspective. But an A+ would solve the “how to distinguish really superb performance” problem. You don’t even have to call it an A+ or average it into GPA; you could call it “with distinction,” and there could be a clear understanding that you were only allowed to give one or two per class, that zero distinctions is fine and even expected, and that any student who asked about distinction was disqualified for distinction. A student who racked up three or four in their career would be a visible standout.
3. More pass-fail. I teach a first-year seminar at Wisconsin, a writer’s workshop where the students, mostly data science majors, learn how to write good sentences and put those sentences together into good, clear paragraphs. I am going to be honest; I give mostly A’s in this class. And I stand by that practice, and would be pissed if the dean told me I had to stop, because I’m teaching a hard skill that most first-years don’t feel confident in and I think fewer students would take it if they thought they might get a bad grade for not being a fully developed writer of English prose. I want them to take the risk of coming to my class! The big downside of low grades, for me, is that they might discourage students from signing up for a hard course in an area where they don’t feel fully confident. Those are the exact courses we want students to take! (Sidebar: of course, this is also a potential downside of high grades; if you feel everybody around you has a 4.0, maybe even taking a class where you might get an A- is too great a danger to endure.) My son, as I think I’ve mentioned, goes to Brown, and I think the relaxed culture of taking courses pass-fail there really does help students stretch themselves more. I just looked this up in the student handbook, and it looks like Harvard students are also essentially unlimited in how many of their electives they can take pass-fail. So maybe this is a cultural change that Harvard needs — sure, limit A’s, but also create an Brown-like environment where students are encouraged to take that weird but interesting course with a great professor where you really might not master everything, and take it pass/fail. Heck, Harvard could require every student to take a certain number of pass-fail courses that met no requirement. Imagine the undergraduate crashout that would cause….