What do students learn?

The physics department puts on a series of lectures during IAP. I went to two biophysics talks (I'm not a biophysicist, but it seems to be the hottest field in the soft condensed matter family at the moment), and one on interview skills.

The most interesting, however, was about teaching and learning in physics. The talk was by Dave Pritchard, an atomic physicist (incidentally, atomic physics is not closely related to nuclear physics - this point isn't really relevant to the rest of the post but I want to do my bit educating people about the shape of physics) who seems to have decided that his teaching and mentoring has the potential to contribute more to physics than his first-hand physics research (he hasn't won a Nobel, but people he has supervised have), and now focuses on educational research.

One of the most important points he was making is that it's important to measure what and how students learn, because if we try to guess, we are likely to be wrong. For instance, he mentioned a case where some homework was due at 10pm on Tuesday each week, and asked the audience how much they thought the students had done by the end of Sunday (it was internet-based homework, so they had these measurements). To make it easier, he asked people to raise their hands if they thought more than 10%: no hands went up. The answer was that people who copied homework less than 10% of the time (which mostly means 0%) had done 50% by this stage. Apparently only one person has ever got the question correct: a professor who is a lifer (i.e. MIT undergrad, PhD, professor). The rate is much lower for people who frequently copy.

Additionally, the amount of copying a student does on homework is by far the strongest predictor of their exam success: stronger than the diagnostic test, online homework or mid-term (written homework doesn't predict at all). Of course, it's extremely difficult to prove causation, but the arguments seem good - other things that can be measured (e.g. exam scores) are taken into account, and the students do better on questions similar to those they don't copy.

Another interesting statistic was a comparison of what teachers would want to teach and what students would want to learn, if they magically had 20% more time available for their physics classes. Teachers overwhelmingly favoured teaching students how to make sure answers made sense, rather than blindly believing their calculations, didn't have a particular inclination to look at applications of physics to the real world, and definitely didn't want to include more content. Students strongly wanted more content and applications, and saw no value in spending more time learning how to make sure answers made sense.

Regarding the making sense vs more content, I'm strongly with the teachers. Physics being performed in the real world is 80% about thinking well while applying very basic concepts that you learn in high school or early undergrad, and 20% very specific concepts that you learn as you establish yourself in the field. A physics education needs to be focused on the development of skills, and that only works if concepts are taught in the right structure (there's actually decent consensus on the general pattern a physics education should follow).

Application to the real world is an interesting one, and it suggests a deeper question regarding the purpose of studying physics. A small proportion of physics students are like me and plan a career in physics research: the standard university physics program is well-suited to us. Engineers are probably well-served by first year physics (and after that engineering departments teach their own physics courses). But what about all the people who study high school or first year physics, with no plans to pursue a physics-related career? I think the reason we study subjects beyond our specialisation is to encourage an ability to think, and to think even about less familiar information or topics. I think the teachers are right in their view of how to do this better. There are certainly areas where physics can be applied to the real world, but I don't think they're as dense as some other subjects (English, math and the economics I wish I had studied being the most applicable in my experience).

There were some other things in the talk: e.g. how much physics do students remember when they graduate. The researchers gave graduating students a test equivalent to a first-year physics final exam, and the results were unsurprising: physics majors did approximately as well as they had in first year (i.e. very well), closely related majors forgot about half of what they had learned, and unrelated majors forgot more or less everything. So there's a clear implication that we need to work out what we're trying to teach, and how to measure whether we're succeeding!