Concept maps are rubbishPosted: October 31, 2015
For many years, I have given the following advice to students who are preparing for an exam.
“There are two ways to revise; answering questions or turning information from one form into another. Both mean that you have to process information in some way. On the other hand, reading your notes or just copying them out can be done without really thinking and so these aren’t very effective.”
I don’t know what I based this advice upon, other than my own intuition. I was quite late to education research, as I suspect many teachers have been. So I was particularly interested when in 2013, Dunlosky et. al. published their findings about the most effective study strategies. Testing was in there – and that was the main message that seems to have come out from the various articles and blogs that have been written on the back of this paper (as an aside, it is also worth mentioning the recent debate as to whether testing works equally well for complex items as it does for simpler ones).
Rereading and similar strategies, such as highlighting, were found to be ineffective. However, the report is more ambiguous when it comes to my advice about, “turning information from one form into another.” By this, I meant making revision notes, flashcards or posters and Dunlosky et. al. found that ‘summarising’ had low utility because many students don’t do it very well.
One information processing strategy that is particularly popular amongst educational researchers and some teachers, and that is not directly addressed in the Dunlosky et. al. study, is the notion of constructing ‘concept maps’.
I speculate that concept maps are popular because they apparently mirror the ‘schema’ that psychological theories propose for the way that information is organised by the mind. You might expect that constructing a concept map of an area of study would require you to rehearse the different relationships between the concepts and so remember and retrieve that information better.
Which all sets the scene for a fascinating piece of research conducted in 2011 by Jeffrey “Testing Effect” Karpicke and Janell Blunt. They compared testing-based study strategies with concept map construction. In all of their experiments, testing was superior. This was true even when the final test – the “post-test” which we analyse to determine the effect of the experiment – required students to construct a concept map. So concept mapping is not even the best preparation for concept mapping.
This is quite a clever experimental variation because there is a general principle in education research that students tend to learn what you teach them. So, with a cunning experimental design, you can bias the results. If you read much research you start to see it everywhere. For example, imagine you were comparing an inquiry learning approach to gravity, where students perform experiments with balls on ramps, make hypotheses and so on with an explicit approach that teaches them about gravity. If your post-test focuses on hypothesising, results tables and so on then you’ll get a different outcome than if your post-test contains physics questions.
So the fact that Karpicke and Blunt found that testing beat concept maps under conditions most suited to concept maps is significant. However, one objection was raised; the students in the study had little experience in concept mapping and so perhaps they were not doing it very effectively.
Fortunately, we now have a 2015 replication of the Karpicke and Blunt study which also deals with the experience issue. Lechuga et. al. conducted their study in Spain and essentially found the same as Karpicke and Blunt. Crucially, they also assessed students who had prior experience with concept mapping. Retrieval was still more effective than concept mapping for this group, although by a smaller margin.
This illustrates how educational research should work. We have a replication and we have variations to the experimental design to take account of different explanations and hypotheses.
On this basis it is clear that investing time and resources in concept mapping is the wrong choice. It may be better to advise your students to test themselves and answer trial questions instead.