*By Joe Pickett, OCW Publication Director*

Student evaluations of courses are a time-honored end-of-semester ritual at most educational institutions. The utility of these surveys, forms, and write-ups has never been highly regarded, as the responses only come from a portion of the students, tend to praise instructors known for easy grading, and are too general to improve teaching in any meaningful way. Of course, they are of little use to the students who took the course because they have already moved on to other things.

In an ideal world, instructors would get specific feedback from students on different aspects of a course as the semester is rolling along. That way, the instructor could see what’s working well and what isn’t, what excites students and what doesn’t. This would enable improvements not just in future iterations of the course but also in the same semester, as the instructor could make adjustments for the particular mix of students in a given class.

Dr. Andrew Sutherland has actually implemented such a system in his course *18.783 Elliptic Curves*, a graduate-level course on elliptic curves with applications to number theory and cryptography. Sutherland explains how his survey system works on the course’s Instructor Insights page.

Each of the 13 problem sets assigned during the semester ends with a survey, which students are required to fill out. The survey is brief and asks students to rate each problem in the set by interest, difficulty, and time spent, and to rate each lecture in a similar way, including the quality and novelty of the material and the pace of presentation. Although the surveys are a requirement, Sutherland keeps the tone lighthearted. For example, the rating scale for interest ranges from 1 = “mind-numbing” to 10 = “mind-blowing.”

In response to problem set feedback, Sutherland has improved the course over several iterations. In particular, he has dropped or added problems, dropped or added hints, clarified points of confusion, and adjusted problems to make the connection clearer to the rest of the course content.

Deleted problems often get added to the course lecture notes. The lecture portion of the survey has led to a variety of adjustments, including the addition of review material and background topics and the slowing of the pace of delivery.

Sutherland reviews the surveys as he grades the p-sets, with a goal of turning everything around in a few days after the submission date. He often replies to students by email, sparking a dialogue. Sutherland has also used this survey strategy in another course, *18.782 Introduction to Arithmetic Geometry**.*

It might seem that gathering all this feedback requires a lot of dedication—and a thick skin! But as Sutherland sees it, the benefits are very much worth the effort:

“Perhaps the most useful thing I have learned as an educator from this dialogue is that it is very empowering (both for them and me) when students help to shape their own education. MIT students want to optimize everything and they have little patience for poor design. This can make them harsh critics, but it also makes them enthusiastic collaborators when they feel they are partners in the process.”