I have a variant on the usual grade-inflation problem ably discussed here. When I have control over the structure of my courses, I tend to favor a large number of short writing assignments for freshmen, and a smaller number of longer writing assignments for other undergrads. The tendency to increase the number of assignments has two valuable effects from the standpoint of evaluation: a larger sample from each student decreases variance and the effects of randomly landing on an assignment the student cannot execute as well, and more assignments mean a greater chance of improvement over the semester. But combined with my policy on late assignments, this creates two distinct sub-populations within the class: people who are unable or unwilling to do the assignments to the quality expected, and those who put in decent effort for the term. These two groups receive very different grades, as one might expect.

But this leads to a problem that average grade statistics don't really cover: a bimodal distribution where a few students get very low grades and a larger number receive decent-to-good grades. The average grade for a course I teach is around an 85, but the number doesn't give much useful information, since one group is around a 70 and the other is tightly bunched around 88-91. And thus the question: do I need to be a tougher grader on the people who are making progress over the semester? Should I accept the low grades as legitimate in the summary of my teaching experience, or should I treat them as outliers?

No comments: