Has a new study just proved that students like explicit teaching but it’s ineffective?

TL:DR – No

A paper was recently published in PNAS authored by Deslauriers, McCarty, Miller, Callahan and Kestin. I wasn’t going to blog about it, because I didn’t initially think it was of much consequence and Blake Harvard has done a pretty good job here. However, the fact that my brother-in-law alerted me to it should have signalled that this would be a paper with a wider reach than most papers of its kind. Sure enough, it’s since been popping up in my Twitter feed as proof of the effectiveness of constructivist teaching.

In short, groups of university students studying a mathematical physics course were randomised into one of two conditions. In the experimental condition, students first attempted to complete problems in groups before the teacher gave instruction in the standard problem solving methods. In the second ‘control’ group, students were given lectures where they filled in worked examples while the teachers completed them, but where they did no problem solving themselves.

This seems like an odd control condition and it appears that the researchers were aware of this:

“Typical class meetings consisted of chalkboard lectures enhanced with frequent physics demonstrations, along with occasional interactive quizzes or conceptual questions. In the instructional taxonomy of Stains this approach would likely be classified as interactive lecture, with lecturing as the primary mode, supplemented by student in-class activities.”

Why did they not use this as the control? It’s not clear.

The researchers found that the non-interactive lectures were less effective than the experimental condition and they also found that students enjoyed the lectures more than the experimental condition.

Neither of these results is surprising and makes the paper strangely ahistorical. For instance, it’s long been known that making lectures more interactive – even by just giving students ‘clickers’ to answer multiple choice questions – makes them more effective. I suspect this is largely due to the fact that interactivity requires students to pay attention.

This is one of the reasons I have been banging on for so many years about the need for explicit teaching to be interactive. This was a key finding of the process-product research of the 1960s that is summarised by Rosenshine’s Principles of Instruction. Principle three is “Ask a large number of questions and check the responses of all students.”

We should also not be surprised that students learn less from the condition they prefer the most. This has been a commonly replicated finding since at least Richard Clark’s 1982 paper on the subject which strangely does not appear as a reference. Instead, the researchers appear surprised by this finding. In my book, The Truth About Teaching, and elsewhere, I have regularly used this finding to cast doubt on the effectiveness of giving students choices over the learning process. Student choice is constantly advanced as something close to a panacea that will serve everything from tailoring learning to the individual needs of learners to meeting vague social justice aims.

There are two other points to note about the research. The experimental condition looks a lot like ‘productive failure’ ie students are asked to solve problems prior to being instructed in how to solve them. There is no reverse condition in which students are first instructed in how to solve the problems before practising solving problems themselves. For high element interactivity (complex concepts with novice learners) my own research suggests this would be superior.

And of course the other point for teachers to consider is how this result achieved with elite university students would generalise to a typical K-12 classroom.


6 thoughts on “Has a new study just proved that students like explicit teaching but it’s ineffective?

  1. Tyrone says:

    I agree the design is bizaare. There also appears to be a conflation of terms throughout the report. The control condition emphasized “fluency”, but fluency is normally used to describe learner behaviour not teacher behaviour as in this design.
    Further the experimental group received teacher questioning and assistance during the student problem solving, indicating a mixed approach to this element of the design.
    It then states the crucial variable between the groups is the level of student centred learning, where as it is actually the time spent actively engaging taught processes and concepts, which the experimental group got in both phases.
    I agree you alternate design would have been far more meaningful in its findings.

  2. Jane says:

    This study does make sense for higher education, where non-interactive lectures are still the norm and reformers are pushing “active learning”, and is quite well-designed, at least by the (admittedly low) standards of higher ed research. Coming from that background, it’s not obvious to me why adding a few more problems to the many hours of problems students already do each week in a physics class would have such a large effect. The fact that students tried several problems in an hour and the professor still had time to lecture implies that they only worked on each problem for a short time. Maybe a generation effect is in play.

      • Jane says:

        That’s probably because the researcher is one of the people trying to make university lectures more interactive. He is imitating others’ practices, not his own.

      • That’s an interesting bit of speculation, but I am sceptical because the experimental condition is also very different from business as usual. Wouldn’t a default uni course actually be a mix of lectures and seminars? The control condition here only got lectures and did not solve any problems at all.

  3. The control condition is then equivalent to students who don’t do their assigned problem sets.
    Many university courses have assigned problems that are worth only 10%. The idea is that is an incentive to do them but not so much that cheating will get them through.

    It should be easy to get data on how completing these correlates with exam results. Of course this compounds many factors as there is no random assignment when it is students choosing whether to complete the problem sets or not.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.