The C-minus Rule

One of the requirements that has been introduced into the lower division courses is the ``C-minus Rule.'' The C-minus Rule states that all students must attain a C- or better in both the Lab (programming) and Lecture (theory) components of the introductory sequence in order to earn the C- required to advance. If one component or the other does not meet this requirement, then no grade better than a D+ shall be awarded to the student, even if numerically it would result in a higher grade.

This is a very necessary rule. In many ways it makes more sense for the introductory courses to be taught with separate lecture and lab grades, however the administrative overhead of such a change outweighs the benefit. Without actually breaking up the courses, this is the only way to ensure students are understanding the material presented in class (Lecture) and are proficient in their primary CS skill: programming (Lab).

However, this rule does have one unexpected side-effect: strongly bimodal distribution of grades, especially if the average is anywhere near the C range. This is logical: if a student is failing one portion, but not the other, it is more likely that the student is a C student anyway (during the summer course, there was one student who briefly was earning a B overall but had a D in lab and thus was dropped to a D+. All other students affected by this rule, through the session, had a C overall), now moving to a D. During the summer session, where lecture grades were generally high, as many as $40-50\%$ of the C students were being dropped to a D grade as a result of their lack of programming points. If you take any normally distributed set of grades A-F, and move half of the C's to D's, you get a bimodal distribution. The bi-modality that was noted in mid-course grades is, to a large extent, the result of the C-minus Rule.

Things that need to be kept in mind when dealing with the C-minus rule:

1. A 'C' cannot be considered average: If a grade of C is considered average, and everything below that is failing, then failure (D or F) rates approaching $50\%$ should be expected. With an extra $5\%$ leeway (either lowering the C-minus rule to work at the D+ level, or raising the average $5\%$) this behavior largely stops. I re-centered tests and quizzes to about the $80\%$ level, and this helped immensely. I strongly recommend adopting the Mudd standard: an ``average'' performance is about a B-.

2. When averages are low, bi-modality should be expected. Perhaps this is more useful in reverse: if a course with the C-minus rule in effect is witnessing bi-modality, the average scores are likely too low. Discounting the notion that a given set of students is significantly dumber than any previous set of students^B.6, it is likely that grading has been too harsh.