My recent post on grade inflation drew several interesting emails. I’ll catalogue and respond to them here, in lieu of the comments section we’re still hoping to re-open once we have a working spam filter.
One reader points out that grade inflation involves an element of fraud, because not everyone — especially, not everyone outside the academy — knows the code.
So, for example, an unwitting employer might learn that an applicant had graduated from Harvard “with honors,” without knowing that Harvard awards a quasi-distinction called “Cum Laude in General Studies” which requires a 2.7 GPA: a B-, in the age of the Gentleman’s B+. About 80% of each Harvard class graduates CLGS or better. If that employer were to imagine that hiring an honors graduate of Harvard meant hiring a brighter-than-average Harvard graduate, he would be sadly deceived. (As I recall, IBM used to have a similar “distinction” for its sales force.)
That’s a fair point; even if employers know in general that there’s been grade (and honors) inflation, it’s hard for them to keep track of which institutions have debased their currencies, and by how much.
A related point made by another reader is that grade inflation makes it very hard to compare people who graduated from the same school in different decades. I’m somewhat less concerned about this problem than about the problem of deception, because I’m not sure anyone should be paying much attention to the academic record of someone who’s been out of school for a decade.
One of my correspondents disagrees with my claim that high-stakes grading discourages cooperation among students; that problem is a characteristic, he says, of grading on a curve. Fair enough. However, there is almost always an implicit curve, and if the bottom of the pecking order is getting F’s instead of B-‘s the pecking is going to get more vigorous.
Yet another reader reminds me that the Law School Data Assembly Service used to supply its client law schools with a fudge factor for each undergraduate institution, along the lines of the one I ask for above. By comparing the distribution of LSAT schools from each school with the distribution of GPAs for LSAT-takers from that school, LSDAS could essentially convert a student’s GPA to an LSAT scale, comparable across institutions. Whether such a service still exists for law schools I don’t know; if one exists for grad schools, using the Graduate Record Exam, my department hasn’t heard of it. That would be especially useful for judging candidates from overseas; I have no clue what a 3.4 GPA from, for example, Sun Yat-sen University is worth.
That would fix part of the problem, though not all; there remains the problem of the distribution of grades across departments and individual courses within a given college. Most of us assume that grades in math aren’t as inflated as grades in sociology, and it’s not hard to spot a student who has crammed his senior year with introductory courses for non-majors (known generically as “Particles for Poets”). But such rules of thumb are fallible, and the sort of eyeball renormalization I can do while spending fifteen or twenty minutes reading an admissions file isn’t as good as even a crude numerical conversion device would be.
There’s also the question of whether grade inflation makes students feel better about themselves and the learning process than would otherwise be the case. It shouldn’t be true if people were rational, but maybe it is. The grades I submit to the registrar are on a fairly generous curve, but the component grades for assignments and exams are on an absolute scale: on a ten-point question, an answer that’s half-right gets five points. Even though I carefully explain all this up front, an A student is still likely to panic when he gets back a first midterm with an 81, and explaining that the median was a 66 so an 81 is a very respectable score doesn’t always do as much as it should to calm those fears or salve that wounded pride. But on that point I’m firm: I’m not going to let compression at the top make it hard to tell who actually deserves an A as a final grade.
That points to a larger question. Educational institutions serve two functions: education, and branding/sorting. It’s not clear which is more important, either to the customers or to the wider society. And it’s not clear whether a grading system that gives the most valuable information about students is identical to the grading system that does the most to encourage learning. Probably not.
What’s frustrating is that, both at the institutional level and at the social level, we mostly allow the process to be driven by the opinions and incentives of individual faculty members, rather than having a serious debate about what we want the grading system to accomplish and then designing a system to optimally serve whatever mix of goals is chosen.
This is, I think, an instance of O’Hare’s Law, which states that any institution not subject to strong outside pressure will eventually be run for the comfort of those who get to make the decisions. It’s more comfortable not to argue about grading, so mostly we don’t.