The difference in information in the study you referenced was a worked example vs a worked example that included explanations (didn’t appear to matter very much). All work was individual – some were asked to “self-explain” steps in the worked example & others to “paraphrase”.

The difference in information in the Schwartz study was a Tell & Practice lesson for marking deviations on a histogram vs whatever ideas the students exchanged on how to decide which grade or which athlete did better.

Here is one study that carefully controls for the exact content students are exposed to and still finds a benefit for inquiry.

That’s from 2010, and I’m not aware of any prior studies that were as carefully controlled. I haven’t followed the literature as closely in the last 5 years, so I don’t know if this result has been replicated or refuted elsewhere. For those who want the tl;dr version, here’s the abstract:

“Self-explaining is a domain-independent learning strategy that generally leads to a robust understanding of the domain material. However, there are two potential explanations for its effectiveness. First, self-explanation generates additional content that does not exist in the instructional materials. Second, when compared to comprehension, generation of content increases understanding and recall. An in vivo experiment was designed to distinguish between these potentially orthogonal hypotheses. Students were instructed to use one of two learning strategies, self-explaining and paraphrasing, to study either a completely justified example or an incomplete example. Learning was assessed at multiple time points and levels of granularity. The results were consistent, favoring the generation account of self-explanation. This suggests that examples should be designed to encourage the active generation of missing content information.”

I am not criticizing the “Invention” approach. Just the opposite. And, I am certainly not arguing that DI is always better for … However, I do not see this study as providing evidence that Teach & Practice limits transfer. It is easy to think of content for the Tell and Practice lesson that would be of no help (calculating “r”) and content that would be very helpful (something just like the embedded worked example). How was the content for the Tell & Practice lesson chosen and how did that choice affect the results?

It would be interesting to record the student ideas during the invention phase, and then provide some or all of that information to the other group using Tell and Practice. This way there is less confounding of the form of instruction by the differences in information.

If the first three areas didn’t provide concrete (and repeated examples of what you claim is unlikely), my own practice unarguably did. I was, after all, the instructor; I did engage in reflective practice; and I did find my instruction to be wanting, and not just marginally so or so that I could say later that I was “self-critical.”

If you want a conversation, I can readily flesh out the above with examples. If you want to “win” an argument, I won’t waste my time writing anecdotes that you feel compelled to ignore. Just let me know what you’re looking for. And consider doing it with a real name and putting some of your own practice out here for discussion. I’ve been teaching and working in education since 1973 and my record is anything but unblemished. But it’s all grist for the mill.

Finally, please explain how the existence of teachers of “extraordinary number” who do something obviates the existence of those who have a different, perhaps conflicting experience. And let me know if my hammer example was too subtle for you or if there was another reason you chose to ignore it, as I fear you’re going to ignore what I’ve written here in yet another stab at “winning.”

But that has nothing to do with Explicit Instruction.

These straw men you set up are easy for you to knock down, but they don’t represent my case in the slightest.

Why is it that giving a thoroughly worked example for 5 minutes, before expecting the students to start doing them themselves, is “passive”. Yet letting them watch a 5 minute video as part of a 3-act routine is “active”? Because, to me, watching a video is just as passive as watching a teacher. (No offence to the 3-act, because I can see it’s place — but not because it is somehow magically “active”.)

My belief is that students learn Maths best by doing Maths. If they enter my room and get 60 Algebra problems done in a lesson, how is that passive? (And, no, not 60 of the same problems, they build in complexity and challenge.)

Yet I get them on some “discovery” based learning and they spend 10 minutes confused as to what is required, then 10 minutes getting their heads around the first problem, and only 20 problems solved overall, and that’s “active”?

There is very good teacher- centered instruction but it is done by people who are likely chafing against the collar of its philosophical assumptions and inherent limitations.

I’d like some evidence that good teachers are actually teaching in a way they think is unsatisfactory — because, on the face of it, that seems unlikely. In my experience, it is falsified by an extraordinary number of teachers who use teacher-centred instruction quite happily and successfully.

@l hodge, I hope you’ve managed to catch the breath that was taken from you by Schwartz’s piece!

In any case, I don’t find your quotation unrepresentative of current practice. Check out what Sweller himself used as the experimental condition in his original paper. It’s more blunt than Schwartz.

Maybe Schwartz’s calculation of variance article would be more to your liking. Maybe it won’t. In general this seems like an unfalsifiable position:

Explicit instruction is always better for all novices for helping them develop new knowledge. If a study shows that another kind of instruction was more effective, that’s because they used the wrong kind of explicit instruction. You’ll know you’ve used the right kind of explicit instruction because the study will demonstrate that explicit instruction was more effective.

Are you familiar with Ken Koedinger’s work? He has been exploring your exact question for some time. He doesn’t have a definitive answer yet, but if you look at his 2012 article in Cognitive Science, you’ll see Table 4 , which provides his theoretical framework for that question.

And if you want his summary of the empirical evidence so far, it’s in Table 6 with further explanation below the table.