People aren’t flipping out for flipped classrooms like they used to

Flipped learning, a new strategy for teaching that flips the traditional idea of classroom lecture followed by homework on its head so that students watch the lecture before class, then use class time for active learning.  This has seemed to produce excellent results. We discussed it as one of the solutions that Stanford offered during the pandemic and even pointed to it as a potential solution to ChatGPT woes.

But a new meta-analysis breaks down the reality of how flipped learning is being implemented and whether it really is that useful. The study looked at 173 studies of flipped learning and summarized their conclusions across K-16. 

One traditional problem with converting to flipped learning has been the prep time for the teachers to write and record a lesson and then edit it into a video format for the students. But something this meta-analysis explored was just how many different styles of flipped learning there are, and how some classes have skipped over that hurdle, mostly by assigning other people’s videos. This was always a possibility, but it changes the role of the teacher dramatically, and comes with the disadvantage that the content might not line up perfectly with the coursework.

Ultimately the meta-analysis authors concluded the wide variety of flipped learning was a problem and that the solution lay in a more rigorous form of flipped learning. Something they call “fail, flip, fix, and feed”. 

The idea is to first test students with problems and let them “fail” which will make them curious about the answer. This also gives the instructor a chance to diagnose if a student is already fairly well versed in the material or might need extra help. The teacher then provides the instructional content ahead of class which will explain how to arrive at the correct answer, same as with a “flipped” class. Third, the instructor “fixes” misconceptions of the class in a format that can be similar to a traditional lecture tailored based on the results from the initial tests. Finally the students and instructors receive “feedback” on the students’ understanding, usually through another test, possibly the same one issued at the beginning so the students can see their own progress.

This “test first” concept is one that is popular in a lot of EdTech products. If you have ever used Duolingo to try to learn a language, almost certainly the very first thing you encounter is a question about which word translates to “man”. In fact repeated, low stakes testing like this is a learning tool itself, and is the foundation that something like flash cards is built on. For topics where the underlying content is based on patterns, like Math or Physics, testing before a lecture can be a powerful way to instill some instinctive understanding without having to learn the details of theorems.

Many hours are spent in geometry classes around the US memorizing that “side-angle-side” is sufficient to prove congruency, but there are lots of students that could repeat “side-angle-side” without knowing how to implement it.

The same is not true for history. A quiz that teaches students that the “Defenestration of Prague” took place in 1618 or that it led to the Thirty Years War doesn’t also give students a deeper understanding of the Catholic-Protestant conflicts that underpinned the event. History is a particularly narrative and story-driven subject and requires lectures in a way that the other subjects don’t always. But it still could be augmented by some testing at the beginning. Like a murder mystery which starts with a context-light question. Why was Benoit Blanc invited to this island retreat, and by whom? Constructing a History lesson specifically like a mystery could lead to more curiosity and even better motivation to learn.

The difference in teaching subjects also sparks a discussion about how different pedagogy works for different subjects, something that was a little outside the purview of the meta-analysis. Flipped learning would have to be different for history than for math. What does in-class “active learning” for history look like? It is likely to be discussions and debates, probably in small groups. Whereas math will be problem sets, which could still be in small groups or alone. 

The other intersection here is “mastery learning”, and what a flipped classroom could look like for learning personalization. If an instructor is assessing individuals to get feedback on their level of understanding, can they take that opportunity to circle back on something that was already taught and re-teach it to the one or few students who are struggling? Like the rest of these new teaching styles, incorporating mastery learning into the “fail, flip, fix, feed” model will take a motivated instructor and some extra work. But it also fits perfectly and the system could even be amended to something like “fail, flip, fix, feed, flawless” indicating that you repeat the cycle until the student’s understanding is flawless.

Flipped classrooms may not have been the silver bullet to learning, but if there has been one thing that is clear since reading Failure to Disrupt, nothing will be a silver bullet. We must, rate, review, and refine our own teaching systems to continuously learn how they can be even a little bit better as we strive to make learning better for all students.