Sunday, August 10, 2014

Princeton bags its misbegotten grading "rules"

Grade inflation is a problem we love to hate. Here's the problem--a problem shared with many aspects of higher education.

Grade point averages are very precise measurements--we compute them to three decimal places, sometimes four--but are all but meaningless, because they are averages of individual component grades given with little or no normalization or coordination. "A numerical soup of unreliable ingredients," as I called in Excellence Without a Soul. There are various reasons why grades have drifted steadily up (for more than a century, actually), but the main reason is that there is too little disincentive for faculty to be generous when given the choice. None of us really knows, in some ultimate sense, whether a student really deserves an A– or a B+, and if we don't know then the deans and academic committees sure don't.

A few years ago Princeton decided to take grade inflation squarely on by mandating that no more than 35% of the grades in any course could be A's. Except that that isn't what they did, because, reasonably enough, they thought there was no reason to punish very talented students for taking very challenging courses and doing well in them. So the 35% was not a mandate, it was a guideline.

The faculty has now revisited that policy and will probably abandon it. The proximal reason will be described as some as Princeton student complaints that they were being disadvantaged in graduate school applications relative their peers at other Ivies with squishier standards. But the report of the committee that looked into the experience makes clear that its actual rationale was less consumerist. Grades did go down at Princeton for awhile, but much of the increase decrease happened while the problem was under review and before the policy had been adopted. In other words, when the faculty were debating and discussing their grading practices, they got stricter; when they stopped talking about it, policy or no policy, grades started to drift up. The percentage cap was not an effective agent of grade deflation. The talking cure was.

Talk was one of the remedies I proposed last winter when the grade inflation issue flared up briefly at Harvard because of a question posed on the floor of the Faculty by Professor Mansfield. Here is what I suggested on this blog:
Require every department to have a discussion of grading once a year. Hand out the grades assigned by everyone in the department so they have to look at their own and their peers's practices while everyone is watching. Have someone from the administration go to the meeting to make sure it happens. No quotas, no rules, just information and a requirement for talking to each other about what grades are being given and why. The underlying idea here is to make the conversation more intimate, conducted not by a dean from on high but in a collegial way, by people the faculty have to interact with every day and whose respect they value.
We actually did this among the computer scientists. Can't say that it had any effect or that we have any particular problem to solve, but those who were there for the conversation certainly found it interesting. (There has been not another whisper about grade inflation from the university administration. I am not sure that is wrong-- I have always thought we had other educational issues to deal with that are a lot more consequential.)

So here is the bigger problem. Higher education is a soft sort of trade. We are in the business of creating mature, responsible men and women, who understand enough about themselves, the world from which they came, and the world in which they are living, to take responsibility for that world's future. There is no four-decimal place measure of readiness to do that, or of our success in fulfilling our obligation to graduate 22-year-olds who are readier to do that than the 18-year-olds we saw as freshmen four years earlier. We quantify things for various reasons, some of them good, but once a metric exists it becomes an end in itself. It is easier for students to sweat about the last decimal place in their GPA than to think about whether they are getting an education, and it is easier for the faculty to analyze and debate the grading data (the report has some terrifically interesting graphics) than to figure out what they are really doing with the four years of their students' lives over which they have almost complete control, especially in a nearly fully residential college. The same risk of distraction by numerical measures applies in lots of other areas. Really, who cares what percentage of students complete a MOOC? We never try to figure out how many finish the books we assign, and MOOCs aren't even assigned.

So, good for Princeton. (Ed Felton at Princeton has a good blog post on this.)


  1. Goodhart's law: When a measure becomes a target it ceases to be a good measure.

    If a school needs to impose rules from above to ward off grade inflation (or to increase grades- I know of a school that mandated that 35% of the grade be based on ``effort'' in order to boost grades of terrible students) then something has already gone terribly wrong.

    Having a department discuss the issue is a good idea- though I wonder if they will all end up concluding `we're doing fine of course'

    1. Thanks so much. I KNEW there was a law in there somewhere, I couldn't remember the name of it!

  2. Grades did go down at Princeton for awhile, but much of the increase happened while the problem was under review and before the policy had been adopted.

    1. Sorry, message sent too soon. I think you meant "decrease" not "increase".

    2. Right you are. Thanks, I fixed it.