Is 'structured' or 'enhanced' discovery learning effective?
An argument in retreat
When I highlight the evidence arrayed against discovery learning, I often provoke an interesting response from its advocates. Apparently nobody claims that pure discovery learning is effective. However, modern and contemporary forms of discovery learning are highly structured and may even include the odd episode of explicit teaching. Therefore, in arguing against pure discovery learning, I am arguing against a position nobody holds — a straw man.
I am easily persuaded that adding structure, guidance and some explicit teaching to discovery learning will be more effective than pure discovery learning. I cannot imagine a teacher persisting with pure discovery for very long because it is such a poor strategy. As a young teacher, when students did not discover the scientific principle they were intended to discover from a well-structured investigation, I would back-fill with some exposition of what they should have found. So I am an old hand at structured discovery learning enhanced with episodes of explicit teaching.
However, if structured and guided discovery learning beats pure discovery learning, this is evidence for structure and guidance, not evidence for discovery learning. If we want to argue that the discovery component of structured and guided discovery learning adds value, then we need some evidence for that.
At this point, people often turn to a meta-analysis by Alfieri and colleagues that appears to show that ‘enhanced discovery’ beats ‘other forms of instruction’. I am not convinced about this.
One major problem with meta-analyses — there are several major problems — is that it is often unclear what the treatment is being compared with. This is necessarily opaque because once we have gone to the trouble of collecting a large number of studies investigating the effects of ‘enhanced discovery’ then not only is the way this term is defined likely to vary across the studies, the conditions it is compared to in these studies could be vastly different.
In the Alfieri et al. paper, the authors include 56 studies in their analysis of enhanced discovery. Seventeen of these have negative effect sizes, which is surprising given that in many cases, the enhanced discovery condition would have been novel and the focus of the attention of teachers, students and researchers, possibly creating a placebo effect. 20 studies involve adults. One study with a large effect size favouring ‘enhanced discovery’ involved university students completing an optional review of a chemistry topic. The experiment pitted a lecture against students working in groups on worksheet tasks. So this was not even an example of initial teaching and the lecture condition can hardly be described as an optimal example of explicit teaching.
Interestingly, one of the cited studies involved teaching students the scientific principle of controlling variables — a principle more education researchers should adhere to. By including it, Alfieri et al. must view this study as representing an example of ‘enhanced discovery’. Yet it was one of the 17 studies with a negative effect size and it found no advantage for enhanced discovery over something that looks a little more like what I would describe as explicit teaching. Moreover:
“When provided with explicit training within domains, combined with probe questions, children were able to learn and transfer the basic strategy for designing unconfounded experiments. Providing probes without direct instruction, however, did not improve children’s ability to design unconfounded experiments and make valid inferences.”
This is an important claim because advocates of discovery learning sometimes concede that it is less effective for learning basic facts and procedures but more effective at achieving higher goals such as the transfer of learning to new contexts. This experiment explicitly tested that hypothesis and it did not stand up.
It is therefore up to advocates of structured or guided discovery learning to move on from hypothesising its advantages to demonstrating them empirically.
I’ll wait.
Once we allow our language to mean anything that anybody wants it to mean, it becomes impossible to mean what we say.
--Wendell Berry, “In Distrust of Movements”
Words need to have meaning. Ideas such as “structured discovery” cause huge problems by clouding the issue. Structured discovery features muddled thinking both in the instructional design and in the way it is presented. I find this similar to what happened in reading with whole language and phonics-based approaches. Over time, the whole language approach was shown to be inferior. But its proponents avoided the arguments and persisted, renaming their approach “balanced literacy.” This attempt at compromise eventually failed, at great cost to students, because the underlying problems were never resolved. Instead, the proponents bowed to the research and included elements of phonics instruction, but only half-heartedly. Similarly, discovery learning has been discredited and explicit teaching is recommended as supported by research. Then the proponents of discovery learning make a few tweaks and rename their approach “structured discovery” or “guided discovery” and proceed as usual. But discovery learning in its original form is hard enough to pull off, so adding the difficulty of trying to integrate it with explicit teaching does not help. They never resolved the problem. But we can resolve it by rejecting inquiry learning and making instruction as explicit as possible. The only time inquiry learning should be used is when it is clearly appropriate for the material or the students, most often with complex material and advanced students.
I realize that some people are sincerely trying to improve instruction, but whether they intend it or not, their work can be misleading. For example, I came across a program that claimed to provide explicit instruction in spelling and reading. Upon closer inspection, the promotional materials referred to using “guided discovery.” I wrote to the authors to ask about the contradiction and was assured that the program uses explicit instruction. Then I read through the manual and found the familiar ideas and language of discovery learning in play. Students are expected to “discover” spelling patterns. The teacher does not directly instruct, but instead “facilitates” learning. But the word “explicit” isn’t ambiguous. Here’s a simple definition from Oxford Languages: “stated clearly and in detail, leaving no room for doubt.” The definition identifies one of the main problems with discovery learning. It leaves plenty of room for doubt and error. An inquiry approach may be useful in some circumstances, but to build it into the structure of an entire program, especially one intended for use with beginning or struggling learners, also violates a repeated claim of progressive educators, that they are matching instruction to the learner.
I cannot understand the persistent and widespread use of this approach, other than that, like so many progressive educational ideas, inquiry learning is attractive at first glance. However, in practice its faults become obvious. It seems that some people are trained in these ideas early on and cannot break free of them, despite their claims of scientific precision. If someone thinks that discovery learning is the best approach to take, they are free to use it, but they should say so, simply and directly. As reading teachers, we are told again and again that the best, evidence-based practice is to teach “explicitly and systematically.” We need to be able to proceed without misdirection and subterfuge.
Greg, you need to write an article about teacher professionalism and values.
If discovery learning can mean explicit teaching, the concepts becomes useless.
Its seems too many areas of teaching have similarly unclear concepts, balanced literacy can incorporate just enough phonics to be effective for example.
I am not sure how a profession can tolerate such loose definitions because it can only make the profession seem amateurish and bumbling.
Psychology was taken seriously when CBT was protocolised and there were clear outcome difference. Those that followed the protocol and those that did not showed clear outcomes differences. Clear definitions and approaches can only improve the status of teachers and their students.