Today, I attended a seminar on physics education by Prof. Mazur of Harvard University.

He said that he had used to think that he himself was a very good lecturer, because he received 4.5 or 4.7 out of 5 on students' rating on his courses.

However, he realized that this was a misperception. There is a multiple-choice type test called FCI, which consists of about 30 questions which deal with Newtonian mechanics. This test focuses more on concepts rather than on problem solving skills. He had his students take this test before and after they took his course, and compared them. Surprisingly, he found out that there were not much differences on their performances between before and after the course.

Then, he said that the conventional physics problems checked only whether students are able to use physics formulas to plug in appropriate variables rather than whether they actually understood the concept.

As an example, he showed that students could easily solve electric circuit problems by using Kirchhoff's laws but failed to solve other electric circuit problems that focused more on concepts which should be easier, once students understand the concept, since no algebra or no formulas are needed. 52% of students solved the conceptual problems at the same proficiency as the conventional problem. 9% of students solved the conceptual problems better than the conceptual problems, and 39% of students solved the conceptual problem worse than the conventional problems. (About half of the students got 0 or 2 points out of 10 on the conceptual problems.)

So, his remedy to this problem was the following:

First, he assigns students pre-class reading. Then, in the lecture, he explains things in more details. Then, he asks the students to answer a conceptual problem. Of course, students use clickers for this purpose. Then, he shows how many students choose certain answers, and ask them to talk with one another about this conceptual problem, then ask the students choose the answer for the second time. Then, he finds, most students get the problem right.

The point is that the students who have just learned these materials can explain things to each other better than the professor, because the professor has learned it such a long time ago, and forgot what conceptual misunderstandings the students may have. He called this method of teaching "peer instruction."

During the question and answer session, somebody asked how he came up with such nice conceptual problems. The professor responded that he assigns pre-class reading, and asks them to solve three problems after doing the pre-class reading. The first two problems are very hard, so that they cannot just copy the answers from the pre-class reading, and the third problem asks the students to write what they found most confusing. He said that he made the conceptual problems based on what the students found most confusing.

Prof. Mazur also mentioned that he could raise the students' score significantly.

I think he is right. It's much more important to learn the concepts than to learn how to use physics formulas without thinking. I also think that standardized test such as SAT should ask such questions, but I think that it may be too hard to make a bunch of such conceptual questions.

He also mentioned that he never showed the students how to solve physics problems during the lecture, but it turned out that the students who learned in this way solved physics problems better than the ones who were just taught how to solve physics problems during the lecture.

He said that you cannot be a good marathon runner just by watching a dvd which recorded marathon runners running. He emphasized that one has to solve the physics problems himself.

You can check out the video of this seminar at the following website.

(9-28-09, Eric Mazur, Harvard University, 'Confessions of a Converted Lecturer')

His presentation slides are also available at