What can we learn from the success of problem-based learning?Posted: July 28, 2016
I have tended to steer clear of the controversy surrounding the effectiveness of problem-based learning in medical education. This is because I tend to apply the perspective of cognitive load theory which suggests that learning is constrained by the limits of working memory. We can overcome these limits by having sufficient knowledge in long-term memory to draw upon. This therefore predicts that breaking learning down into small, manageable chunks and explicitly teaching these chunks will be best for novices who have little to draw upon from long-term memory but that solving more complicated or realistic problems will be best for experts who already have a lot of domain knowledge.
My difficulty is in placing medical students on the novice-expert continuum. On the one hand, when they begin training they still know very little about medicine. On the other hand, they will have been highly successful school students who will posses a large amount of relevant biological, chemical and related knowledge. Add to this the fact that they are also likely to have a high level of general intelligence and that this is effectively a measure of working memory capacity. So medical students are likely to be constrained less by their working memories than most of us.
I was therefore fascinated to read a review of the evidence for problem-based learning in medicine written by Jerry Colliver in 2000. The discussion echoed many of those that I’ve been involved in when discussing education research but it also provided a few new insights. Not least of these is that Benjamin Bloom has suggested that the optimal form of teaching is one-to-one tutoring and has characterised the goal of education research as finding methods of instruction that approach the same level of effectiveness. I didn’t know this. Apparently, Bloom found an effect size of d=2.0 for one-to-one tutoring, placing a ceiling on what we may expect from education research, although this must be considered alongside our current understanding of the validity of effect sizes.
Colliver reviews a number of key studies that show problem-based learning to be more successful than traditional methods of medical education. But he is a critic of these studies. He notes, for instance, that many lack randomisation – students often self-selected into problem-based learning or the control condition and there were often systematic differences between the groups. Colliver references a paper that explores these differences.
Many of the outcome measures used in the studies showed little difference between conditions whereas some clearly favoured problem-based learning. Colliver discusses a measure of the students’ ability to relate to patients; a measure where problem-based learning students did well. Then he compares the two curricula. In one study:
“the PBL track ‘‘emphasized frequent contact with real or simulated patients for the dual purpose of practicing interpersonal, physical diagnosis and clinical skills,’’ whereas the traditional track ‘‘limited patient contact to supervised encounters with a small number of hospitalized patients as part of bedside tutoring groups in the first and second years.’’
So here is evidence, if it were needed, that students tend to learn the things you teach them and don’t tend to the learn the things you don’t teach them. I am thinking of making this my Principle Number One of Educational Psychology (posters and memes, please).
This has echoes in the kind of research that we see into inquiry-learning in high school science classes where students are tested on things such as their ability to formulate hypotheses; a key focus of many inquiry-learning programs but one that is not emphasised so much in traditional science classes where teachers instead tend to focus on teaching actual science.
Overall, Colliver is skeptical of both the experimental designs that are used to demonstrate the effectiveness of problem-based learning and the rather modest effect sizes that these studies generate. In his discussion, Colliver also critiques the educational principles and learning mechanisms on which the problem-based learning approach is based. He suggest that part of the theory is that learning something in context helps you recall and use that information in the same context later. However, the simulations used in problem-based learning are not real contexts so we haven’t actually achieved the learning-in-context aim.
Again, I think there are parallels to the learning theories that we tend to adopt in schools.
One of the strongest aspects of the Colliver paper lies in this rebuttal, the authors of which don’t seem able to address many of the issues that Colliver raises.