I do not assign grades on an absolute scale because even after many
years as a professor it is not easy to
calibrate the difficulty of exams "just right". The typical
average letter grade in my classes is about a B/B+,
but in some cases it is higher when I feel that student's
performance exceeded my expectations.
Here is an example of grades from the last time I taught a large
undergraduate class (Physics 24, W2018).
The "combined" score (scale 0-100) was calculated based on the final
(50%), the midterm (35%), and the
homework (15%). The four distributions are shown below.
Then, based on the combined score, I assigned letters grades
"roughly" as follows:
- > 96 A+
- > 84 A
- > 80 A-
- > 76 B+
- > 70 B
- > 66 B-
- > 63 C+
- > 59 C
- > 56 C-
- > 54 D+
- > 50 D
I say roughly because I bumped up letter grades of several students
as follows:
- if the combined score was very close to a boundary, e.g, 79.9
- if the performance on the final was significantly better than
on the midterm
- if the combined score was hurt by a "poor" homework score
These were all judgement calls that go in favor of the students.
In this case the mean combined score was 78/100, which corresponds
to somewhere between B+ and A-.
(That class actually did "better than expected").