Students will want to know from the start of the quarter, and especially
when they get their test scores, what your "curve" will be; that is,
what distribution of grades you expect to give.
This is always hard to predict the first time you teach a course.
One way to satisfy both the students' desire for predictability and your need
for flexibility to evaluate performance later is to announce a tentative
grade scale at the start of the quarter that's rather demanding, and
say you will consider adjustments at the end of the quarter, but only
in the direction of making the the scale more generous.
Here is an example of such a grading policy:
"I will set the grade scale for the course at the end of the quarter.
My preliminary estimate is that the scale will be linear, with 4.0 = 90%
and 2.0 = 70%. The actual grading scale will be
no tougher than this preliminary estimate, so your course grade will be
at least as good as the preliminary scale indicates."
You might want to set 4.0 as high as 95% originally, or on the other hand
lower the 2.0 line to 65% or even 60%.
Talk to some recent instructors of your course to find out what their scales
were, and set your preliminary scale with percentages slightly higher
(especially at the 2.0 end) than the percentages they used for grades.
After the first test, you should give the students some score distribution data (median, etc.), and probably tell them at least the approximate 2.0 level. For instance, you might say, "A score of less than X indicates you're going to need to improve to get at least 2.0 in the course." To choose X, you should think about what score corresponds to work that shows at least a minimal grasp of the basic ideas tested, and some ability to work standard problems for this course through to the end (even if not completely correctly). The score distribution should inform your decision; for instance, if almost all the scores are low, maybe the test was too long or some problems were too novel to the class, and X should be correspondingly lower. But setting X should be a matter of judging student work, not just looking at a list of scores. See also the page on End of Term Issues for some additional discussion of course grades.
The Faculty Resource on Grading (FROG) includes sample grading guidelines that discuss how grades correspond to quality of performance, and reasonable grade distributions. The distribution for upper division courses roughly matches the following Math Department data.
Of 3863 students, mean grade was 2.98, and quartiles were
25th percentile = 2.6;
median = 3.2; and
75th percentile = 3.7.
This was about 0.1 above the corresponding numbers for 100 level classes,
and 0.1 below those for 400 level classes.
That said, classes do vary. If you think your class shows excellent command of the material, and they did well on tests that included nonroutine problems, you probably should have a median grade a little higher than 3.2. If you think the majority of your class has at best a grasp of only the most basic ideas in the course, and gave a mediocre performance on routine test problems, it's reasonable to have a median grade lower than 3.2.
Return to the
Index for the 3xx Instructor Guide.
Most recently updated on May 9, 2013