Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-L] ungrading



On 3/31/22 12:50 PM, Todd Pedlar wrote:

works a lot better in the humanities than I can ever imagine it
working in any scientific field...

It works better than you might imagine.

At Caltech:
a) Every course for the student's first two terms is
automatically pass/fail.
b) For each term thereafter, the student can pick two
elective courses and have them graded pass/fail. This
applies only to electives, i.e. courses that aren't
specifically required for the major.
c) Some other courses, including some of the most advanced
courses, are designated pass/fail by the instructor.

Yeah, I know, these are not typical students, but even so,
atypical data is still data.

They've done it that way for decades. Doesn't seem to be
a problem. There are plenty enough non-pass-fail courses
that it is straightforward to compute a GPA.

On the other hand, I've looked at thousands of transcripts
over the years, and GPA is definitely *not* what I am
looking for. The undergrad who got a C in general relativity
is going to float to the top a lot faster than one who got
an A in basket-weaving. Or to be more fair, let's compare
one out-of-major elective to another: The undergrad who got
a C in "Composition and Orchestration" is going to float up
a lot faster than one who got an A in "Music Appreciation".

Also, a C average from Princeton, where the grades are
famously not inflated, counts for a lot more than an A
average from some generic place where everybody gets As.

As a teacher, the grade may be useful to you and may be
useful to some of your students, but it is much less useful
to anybody on the outside.

I still have not figured out what grades are even supposed
to mean. What should I do with the guy who was smarter than
me on the way in and smarter than me on the way out? How
does that compare to the guy who knew absolutely nothing on
the way in and was modestly and narrowly competent on the
way out?

The grades sit on a one-dimensional number line, but they
ostensibly measure a multi-dimensional space. That's provably
impossible. There are theorems that say you cannot change
dimensionality in a way that is one-to-one and continuous
(e.g. Brouwer 1911).

-----

[I am replying to a message that came to me via two different
listservs. I'm not sure what is the proper etiquette in this
situation. For now, I'm not gonna lose sleep over it.]