Spring 1999 CS302 Assignment #8

by Dan Garcia (ddgarcia@cs.berkeley.edu)

1. Design and justify a grading policy for your course.

The grading policy for CS5 is as follows:

There are ten weeks in the course, and each week there is an assignment due. Each assignment counts for the same amount, ten points. The final project is a multiple-week assignment. We've found that in order to prevent students from waiting until the last week to start the project, it's important to set intermediate deadlines for them. Thus, these deadlines will be graded as though they were the homework for that week. The finished final project is worth two homeworks. There is a midterm and a final exam; the midterm is worth one homework and the final is worth two homeworks. Each homework is worth 10 points, so we have a total of 10*10 (homeworks) + 20 (final project) + 10 (midterm) + 20 (final exam) = 150 points out of which the grade will be based. Each homework will be graded on the following scale:

 Grade  Comment for student (and for professor)
 1-5 Failing for one reason or another (usually because it wasn't done)
6 Some serious problems, but adequate for passing. You probably missed the point of the assignment (e.g. Did you have a story?), but I'll take it.
7 Good assignment. A little below average, but you had some good ideas.
8 Very good. You probably had two or three small problems, though.
9 Almost perfect. You probably missed one small point, but overall I was quite happy with what you turned in.
10 Perfect assignment. You completely understood what was necessary and executed it to perfection.
11 Above and beyond the call of the assignment. One of the best in the class this year.
12 One of the best I've ever ever seen. Wow wow and wow. Write home to mom.

The final grade will be determined from a scale which looks very similar to that proposed by McKeachie (1986) as mentioned by Davis in "Tools for Teaching":

147 and above (98 percent of 150) = A+
140 and above (93 percent of 150) = A
135 to 139 (90 percent) = A-

In particular, defend the following:

The grades will be based on an absolute standard. This gives the advantage of being fair to previous classes who also had to measure up to the same standard. Curve-based grading is useful if there are factors which would make the absolute standard unfair. For example, if it is the professor's first year, it may take some time to understand what the students are capable of. Or some years there may be hardships the students must overcome (new software, fewer computers) or perhaps even positive features (say, a higher-end animation program they have access to) that would change their grades relative to other years.

Since this course has been taught before and the software / equiptment is the same, the projects and grades of past years can serve as a good model for an absolute standard for the students this year.

The uniformity of the assignments' weight is meant to imply that all of the material throughout the course is of equal value, from the early discussions of the history of animation to the final discussions of the technical details of perspective transformations. The final project is a multiple-week (four weeks in total) assignment, but the students have had chunks of it graded already (models, sounds, rough animations, etc) so the final result is only worth two homeworks, rather than one large four-equivalent-homework grade. The weight of the exams is de-emphasized (the final is only worth two homeworks, the midterm one) because this course is more about teaching the students how to animate (how to design / model / light / render) than strictly abstract concepts. Their homework is the best measure of the skills we're trying to impart. However, there are still concepts they need to understand, and this is where the exams come in.

There are no extra credit assignments or "best M out of N scores" here, although the students can earn extra credit (up to 12 out of 10 per assignment) for doing really great work. The justification here is the need for some way to compensate students who clearly do work deserving of special merit. The fact that the grades are on an absolute scale means these extra credit points hurt no one.

Most of the assignments are done with partners; exams are completed independently. Partners receive the same credit for an assignment, and students are not allowed to work individually (machine shortage). Some component of the final project grade will be chosen by the students. For example, the students will independently be asked the following:

If you had one hundred extra bonus points to give to the other people in your team (not yourself) on the basis of how hard they worked on the assignment and deserved credit for doing more than their share of the work, how would you divide the points? Recall that the grades for this semester are on an absolute scale, so giving your teammates points does not affect your grade. Use this to reward the person(s) on your team who really deserve some extra credit.

The results from this question will be used to distribute a nominal amount of the points for the final project around.

2. Explain how you would decide a course grade for a student near a borderline in your grading scale, e.g. the best of the B+ students. If you would decide a borderline grade differently for different borderlines or for different distributions of scores comprising the final score total (say, someone doing very well on exams with a low homework score verses someone with the opposite distribution), explain how you would do so and defend your policy.

No student will ever be pulled below a borderline for any reason. All borderlines are treated equally. Students within a point or two below the borderline will be raised above the border if:

WWW Maven: Dan Garcia (ddgarcia@cs.berkeley.edu) Send me feedback

Teaching cs302