Princeton is considering putting a cap on the number of A’s that professors could award to students, as a way of fighting grade inflation. Details are in Alyson Zureick’s story in today’s Daily Princetonian. To my knowledge, Princeton would be the first major university to take such a step. The proposal would have to be approved by a vote of the faculty before taking effect.
Grade inflation is a real problem, and it’s a hard one to fight. There are weak but steady pressures that push grades up over time. A professor, faced with a student on the borderline between two grades, finds it easier to give the higher grade; and at the end of a long semester of hard work by professor and student, it feels right to give that borderline student a tiny nudge upward. Students complain about low grades, and sometimes they can point to a grading error that justifies an upward adjustment; but rarely if ever do they complain about generous grades. These nudges and corrections slowly push the average grade upward.
I also think, notwithstanding the occasional grumbling of old-timers, that our students have gotten more capable over the years. If this is true, then grades naturally inflate, unless we grade the same work more harshly than we did in the past.
In recent years, Princeton’s strategy has been to report comparative statistics, telling each department how its grade distribution compares to others, and telling each professor how his grade distribution compares to his peers. Apparently that strategy has not been enough to stop grade inflation.
The new Princeton proposal would require each department to give no more than 35% A’s (including A+ and A-) in courses. It would be left to each department to decide how to stay under this cap.
I don’t know yet whether I’ll vote for or against this proposal. But I do know that if it passes, my department will have to set some policy for allocating our quota of A’s among our courses. Setting that policy will be no fun at all, even in a department as sane and collegial as mine.
UPDATE (10:45 AM): For more reaction, see today’s Boston Globe story by Marcella Bombardieri.
To anybody who has felt slighted in testing:
Seems to me that bright students taking math, physics, chemistry, biochem or other tests having objective parameters can all earn A’s. Subjective tests and hence grading is a different story. Assigning a quota of A grades is too simplistic.
Princeton Faculty Passes Grade Quota
Yesterday the Princeton faculty passed the proposed grade inflation resolution (discussed here), establishing a quota on A-level grades. From now on, no more than 35% of the course grades awarded by any department may be A-level grades, and no more tha…
The policy would cover undergraduate courses.
It is not completely clear from the story in the Daily Princetonian. Would this be for Undergraduate level courses only, Graduate level courses only, or both? In the universities I have had contact with, graduate level courses are graded on a different scale from undergraduate courses. Graduate students generally had grades from A->C with a C considered failing.
I think it’s great your university is considering this. At my school (University of Rochester) I don’t think this could even hit the table – we treat professors as gods and don’t really give them any restrictions (or even review). As one of the students who looks at grades seriously in the sense of understanding my performance (rather than merely as a tool for jobs/grad school) this kind of idea seems crucial. Grades, as they are now, are extremely difficult for me to interpret as a measure of performance. There are a few variables that change the context entirely:
I’m a computer science student – our classes are commonly accepted to be extremely difficult when compared to other to departments at the school, in particular humanities
I attend this University, and naturally vary from grading systems at other universities
Professors vary massively in their grade distribution
This means I can only understand my grade in the context of my professor, teaching my class, at my university. I can make semi sensible comparisons laterally across my department, and from year to year in a particular class. I’d like to be able to do more with it, though. Really, the only variable I see that would be truly difficult to eliminate (and in many ways is natural and acceptable) is the middle one, that some universities are outright “better” than others.
Gaming the mean system would work, until everybody else started gaming it too. Then the stronger students would cluster in the high-variance courses, and the weaker students in the low-variance courses, neutralizing the value of gaming.
The mean system seems like it would be very easy to game. If a statistically significant number of students participated in a grade-sharing scheme, the standard deviation would be easy to calculate. Then, in classes you think you will do well in, choose a prof with a large std dev. Otherwise, choose the small std dev. Or use the same data to choose electives in order to increase overall GPA.
Stanford Law School has long had an enforced mean system. Grades in each exam classes (paper classers are treated differently) must conform to a 3.4 mean (3.2 in my day, but inflation has struck). This leaves individual professors to decide the distribution. The students, in fact, often discuss whether a prof is a “clumper” or a “spreader” in efforts to game the system.
University of Toronto has had a somewhat similar system in place for a while, and it works quite well. You aren’t actively forbidden from handing out too many As and Bs, but if you exceed a certain quota, you have to write a justification for the unusual circumstances that prompted you to give out more grades, which then goes into your file (and if you accumulate a few of these justifications, it leads to raised eyebrows).
While fact-cheecking the (Rice-sourced, informal) claim that my alma mater didn’t have “that crappy grade inflation here”, I ran across http://www.gradeinflation.com/, which looked like as good a source of data as any other that I have seen. Could be useful for checking trends against layoffs, etc.
The other datapoint, which seems not too far off the mark in its description of the university, is: http://www.rice.edu/projects/reno/speeches/19960910_Gillisfaculty.html
However, one of the figures cited there (21% of grad seniors had B- or below, vs 11% at 30 other selective schools) seemed inconsistent with the average GPA quoted on at the grade inflation site.
You might check with one of your former students teaching there now for a reality check. I can confirm that it is possible
to get a C at Rice, or at least it was 20 years ago.
How was the distribution model invented? Were any studies done to understand what the distribution model should be, and the variables that affect it? Or is the model arbitrary?
Too often, I think these forced distributions ignore the filters that remove the bottom tail (layoffs in industry; tough sophomore-level classes in academia).