An Experiment in Flipped Physics

August 8, 2014

I carried out an experiment in teaching Physics in a flipped environment:  students listened to my recorded lectures on their own schedule, and we did conceptual problems in the classroom, using clickers to facilitate peer instruction.  This left time in class for microlectures on topics in Unsolved Problems in Physics and Astrophysics — the beating heart of Physics as Physicists experience it.  A pool of ungraded quantitative problems was available as a resource.    To encourage collaboration among students, grading was not curved.  Several lines of evidence suggest that this experiment was successful, including an anonymous clicker poll on the last day of class in which those who judged the flipped environment to be more effective than the traditional one outnumbered those who thought the converse by four to one.

The traditional lecture has always seemed to me to be an inefficient environment in which to teach and learn.   My own non-traditional educational background in Montessori schools, and my observations of the learning environment in my daughters’ Montessori school*, suggested to me that I propose to the Physics Department to carry out an experiment in non-traditional education in Physics.  In designing the course, I stole shamelessly from multiple sources, including Montessori pedagogy, Eric Mazur’s pioneering work at Harvard on Peer Instruction, and others.   No aspect of the course was entirely novel, although the combination of elements was new to the Berkeley Physics Department.

The course was Physics 8B:  roughly, Electromagnetism and Modern Physics for Pre-Meds.  About 150 students were enrolled, and we met MWF from 3-4.  These details are relevant to the design of the course. 

We used a Flipped Environment.  Rather than listening to lectures in the lecture hall and doing assigned homework problems outside of class,students listened to lectures that I recorded in the ETS microstudio,** and did assigned reading, before class.   Being vividly aware of my own rather limited attention span, I recorded the lectures in 15-20 minute segments.  This approach had several advantages.  Students could choose to listen to lectures on their own schedule, when they were most alert and receptive to new and challenging material.  (Speaking for myself, 3pm is the time of day when I am least attentive!)   Students could pause the lectures and return to them later if their attention was wandering.  If material was confusing, they could back up the recording and listen again.  

In class, we (mostly) did problems, facilitated by clickers.  The first question in class was always a graded, factual “check-in” question on the lecture or reading, for which the answer would be obvious if one had done the assigned listening and reading.  Most of class time was spent in doing 5-8 “clicker questions” (CQs).  Answers to graded clicker questions were 5% of the grade, but a minority of the CQs was graded:  these were CQs that reviewed previously introduced material in some sense or other.  But most CQs  — those that were on the frontiers of their understanding — were ungraded, so that the students would not be distracted by grade pressure while learning new material.   CQs were generally conceptual, not requiring a calculation, and some were entirely graphical.  Eric Mazur’s Peer Instruction is a wonderful resource for well-tested CQs in Physics. 

The best CQs were those in which roughly half of the students responded with the correct answer and the rest responded with some other answer.  This was the Golden Opportunity for Peer Instruction.  I would then announce:  “Someone near you disagrees with you.  Find that person, and convince them that you’re right.”  Chaos then ensued, which was delightful to watch.   Almost every time, the students would converge on the correct answer after a second poll.   I suspect that this method is effective because students are far more effective in explaining concepts to each other than I could be, since the material is all (or at least mostly) totally obvious to me, and I cannot really remember being confused about it.   For them, the confusion is fresh, and those who have worked their way out remember the path(s).    This also illustrates the importance of being wrong on the road to deep understanding.   My mantra was “Dare to be wrong!”   (I demonstrated this myself several times, sometimes even intentionally.  For this reason, this approach resonates with Richard Freist “You’re Invited to a Celebration of Mistakes!”)

Homework problems were available to students, but only as a resource.  They were suggested as a problem pool.   Many students reported to me that in their experience, when problems were assigned in Physics 8A, they not very useful for learning, because the answers could just be googled at the last minute.  But when they were viewed as a resource, they were more useful.   I explicitly stated that points on the exams sufficient to earn a “B” were based on the problem pool problems.

In my professional life, I find that I spend much more time in collaboration than in competition, and that effective collaboration requires experience, skill, and patience.    To be sure, competitive skills are important, but students are underserved in their preparation for professional life by the current balance, which is too heavily weighted toward competition, especially and notoriously among Pre-Meds.  To encourage collaboration, I did not curve the grade.    The students reported that, as a consequence, the learning environment was much more collegial, and that they had much more fun studying.  (Don’t discount the importance of having fun!)  I believe that this environment is better preparation for the real professional world, because it is a much better approximation of it. 

On the first day of class, I offered to write letters of recommendation to those students who earned an A in the class and who could demonstrate collaborative skills through letters from their peers.    This underlined the importance of collaboration.   I have followed through with this, and I am finding that the peer reviews make wonderfully quotable material for recommendations, in addition to my own comments, of course.

The students themselves had a strong influence on the evolution of the course.  I was explicit at the beginning of the course that this was an experiment, that there was no doubt that some things would go awry, and that their participation in the evolution of the course was vitally important.  This is an essential aspect of the Montessori environment.    At the suggestions of the students, for example, we changed the nature of the Problem Pool problems, introduced Piazza as a collaborative tool, and I recorded several examples of expert thinking in solving particularly challenging problems.     The students took the lead in organizing themselves into study groups through Facebook and by other means.

Because we were unconstrained by the need the cover specific material in live lectures, we could advance through the syllabus at a faster pace than would otherwise be practical.  As a consequence, we were able to reach the more sexy topics, such as some basics of particle physics.   Nevertheless, the vast majority of the material in Physics 8B is more than a century old, and so holds little of the excitement of working in current topics of Physics.   The flipped environment allowed me the time to present short (5-10 minute) microlectures on Unsolved Problems in Physics and Astrophysics, usually but not always connected with a topic that we were covering in class.  The students responded very positively to these digressions, which emphasized the fact that practicing physicists do not revisit the well-trodden material covered in the course, but work to understand unsolved mysteries.  Many students were astonished to learn that we humans do not understand the nature of 95% of the mass of the Universe.  

All of this may sound very nice, but the question of course is this:  did this experiment work?   I should be explicit about my criteria for success.  One goal, of course, was that the students learn some Physics.   The other goal (explicitly stated to the students) was to further their preparation for professional life by gaining experience in collaboration.  My assessment is that the experiment was successful at both levels.  Several lines of evidence point in this direction.   My observation of responses to CQs, especially in extensive review at the end of the course, showed an understanding of the material that surprised and delighted me.  The final exam (23 pages long!) was vetted by one of the GSIs, a Physics graduate student.   If I had graded his exam, his score would have been at about the median for the class.  But the most valuable opinion is probably that of the students themselves.  In an anonymous clicker poll on the last day of class I asked:  “As compared with the traditional approach, I have found this learning environment to be (less/equally/more) effective.”   The responses were 14%, 32%, and 54%, respectively.   Many also sent unsolicited reports to me of their view that collaboration through peer instruction, study groups, and the use of absolute standards in grading was very effective.

I treated the students as peers rather than as inferiors.  For example, I asked that they call me by my first name.    This certainly got their attention, and set an entirely different tone for the course.     I asked the students to respond to a questionnaire at the beginning of the course to learn something about each of them.  I also tried to learn the name of every student, and almost succeeded (after some embarrassing mixups!).    If the only thing that I knew about the students was their ability to do Physics, I might be rather underwhelmed.   But I found that the students were deeply impressive in their own spheres.   It was a great pleasure and a genuine privilege to be their guide through this very challenging material.    That I respected and appreciated them, no matter the grade they received in the course, seemed to surprise some.  (Their surprise, in turn, surprised me — shouldn’t this be normal?)  Some of the most deeply impressive students were those who, as a consequence of inadequate preparation or unfortunate personal circumstance, struggled mightily with the material but did not get an impressive grade in the end.  

Of course, my course bore little resemblance to a Montessori school, but the foundation of this course comes from the Montessori classroom.  In the pervasive and insidiously tenacious Lockean model of education, the student is a passive empty vessel into which the instructor pours knowledge.    My observations here and in my own children’s Montessori classrooms lead me to the conclusion that a more effective pedagogical approach is one in which the instructor is a guide: the student is the active learner, he or she participates in the development and evolution of the course, and takes responsibility in choices about how to use his or her time most effectively.   Maria Montessori said, “The environment itself will teach the child.”

 *Montessori Family School, just across Hearst Avenue from UC Berkeley

 **Many thanks to Tim Gotch of ETS for facilitating access to this wonderful facility

Ben Spike (Physics, U. C. Berkeley), Alissa Stolz (Montessori Family School), and Aaron Glimme (Berkeley High School) made important contributions to the development of this course.


Mazur, E.  (1996)  Peer Instruction:  A Users Manual.

Lillard, A. S. (2008)  Montessori:  The Science Behind the Genius.

Postman, N. (1996)  The End of Education: Redefining the Value of School

Lillard, A. & Else-Quest, N. (2006) Evaluating Montessori Education.  Science 313, 1893.

Deslauriers et al.  (2011) Improved Learning in a Large-Enrollment Physics Class. Science 332, 862.