Some call it grade inflation. Others call it a wide assortment of other terms; I prefer “grade incommensurability”: a given letter grade means different things from different instructors in different departments. An opinion piece in Inside Higher Ed proposes a new statistical approach: simpler than the Achievement Index I and others proposed at UNC a year ago, but perhaps less statistically valid. What do y’all think? I find grade inflation a major barrier to good teaching in my own professional life but know others disagree. And I’m also certainly sensitive to the possibility that some cures are worse than the disease.
-
Pages
-
Categories
- books
- Canada
- contest
- cryptic
- dialogues
- disarray
- dork
- economics
- fun
- gender
- help wanted
- meta
- misadventures
- music
- online wonders
- personal
- politics
- procrastination
- professional
- religion
- science
- sports
- students
- teaching
- too much information
- travel
- Uncategorized
- what does this have to do w/ org theory?
- work and family
- writing
-
Archives
2 Comments
I don’t know–having been on the student and the instructor side of grading, I find all of these new-and-improved quantitative grade metrics pretty unconvincing. It seems to me that grades are incommensurable because courses and materials are incommensurable. To take the examples in the article: you will never find a completely objective way of comparing “Hip-Hop Hermeneutics” with String Theory, no matter how much statistics you throw at the problem.
I realize that some measure of achievement is unavoidable, but I think the low-resolution Stanford Law School system mentioned in the article is better than some fancy and convoluted scheme that only further mystifies the underlying fuzzy, subjective, and arbitrary qualities of grades.
As an aside, I was also annoyed that the author of the article began his call for a precise grading metric by raising the issue of batting averages in baseball, defending the use of three-decimal-point precision by saying that `a baseball `hit’ is a rather unambiguous thing’. That may be so, but anyone who’s worked on baseball statistics knows that the relationship between a hit and the underlying quantity of interest–a batter’s hitting ability–is very difficult to specify with any precision. Likewise, a right or wrong answer on a test is an unambiguous thing, but that doesn’t mean it measures student learning.
AP: At the selective schools, the preparation & ability of students who enter the school has been improving, so anyone at such a school who grades by a constant absolute standard will find his/her grades creeping up. This is coupled with the social pressure: if all the students in your class received all A’s in high school, they are less receptive to the idea that 50% of them should get a C in your class. (Think censored distributions.) Another factor is drop dates. In fact, virtually all of the “grade inflation” over time can be accounted for by more liberal drop policies. (Somebody I know actually published a note or column in Footnotes or the NYT about this, but I cannot remember who. In any event, I had figured this out for myself before he so published, and I would rather hope that almost anybody could if they thought about it a bit.) If everyone who is earning below a B drops your class by the 9th week of the term (or in some cases, below an A), it would be pretty asinine imho to impose a C-curve on the distribution of grades for the people who did not drop.
Also, as you get better at teaching certain kinds of skills, the average ability of a student to do what you want them to do will improve over your career.
I thought I was grade inflated because my undergrad grade distribution is centered on AB (our system has A, AB, B, BC, C grades) until I learned that many folks are giving over half As to undergrads.
None of which is a critique of the idea of improving on a GPA with the kind of statistical technique in the achievement index you linked to. The reason this is a potentially worthwhile exercise is that real-world rewards get attached to measures like GPA and one would like a system that seems fairer for such interpersonal comparisons than the one we have now. Whether either of the proposed ways of doing it would work takes a deeper investment of time than I’m willing to make. But I don’t see how either of these will solve the problem of between-school comparisons.
P: I agree with most of your points, except possibly about batting, as the output seems like it has agreed-upon indicators of quality. The underlying problem is trying to map a multi-dimensional space onto a line.