An experiment in extra credit

During the first summer session of 2003, the following grading formula/algorithm was published and used in CS14 -- see Appendix B on page :

• Take-home assignments assigned in lab, 20%.
• Style on those take-home assignments, 7%.
• Lab practicals, 8%.
• Final exam, 30%.
• Midterm exam, 20%.
• Quizzes and lecture-based homework (not programs), 15%.

Each of the ten lab's in-lab work is worth 10 points, and it's very easy to get nine points. The lab-assigned homework is a catalog of mostly 20-point assignments, each with its own due date. They have soft deadlines, with an additional 10% for each week early and a negative 10% for each week late. To get 100% of the on lab-assigned homework requires 200 points, i.e., the percentage is the max of 100 and

$$\sum 5 \times lab_i + \sum ( 10 + weekDue_j - weekSubmitted_j ) \times asst_j$$

where $i$ ranges from 1 to 10 and $j$ ranges over the entire catalog of take-home assignments. That gives each lab a weight of five and each assignment a weight that corresponds to the week it's due plus the number of weeks before the end of the quarter that it gets turned in. Also, when computing the lab portion (20%) of the grade, this scheme allows points for take-home assignments to be substituted for in-lab work.

The pros of this grading scheme are that:

• Missing an assignment or deadline isn't necessarily fatal.
• Students are motivated to finish their work, rather than submit whatever they've accomplished as of the deadline.
• Excellent students have additional work available and get credit (up to 100 points) for doing that work.
• Competent students who can't attend labs can survive the course.
• Students get to select assignments that they find interesting. (The impression of choice seems to be a motivator.)
• The grading workload is perhaps more uniformly distributed over the quarter.

The cons of this scheme are that it allows students to substitute unproctored take-home assignments for proctored labs^4.9 -- the lab portion of the course thus becomes worthless for documenting the student achievement. Also, soft deadlines are risky for weak students, who are not so intimidated into doing assignments in a timely fashion by soft deadlines as they would be by hard deadlines.^4.10