Boundary conditions for the flipped classroom: Evidence from statistical programming

We propose a novel explanation that accounts for a variety of findings reported in the flipped classroom literature. The flipped classroom reverses the sequence of learning where students first independently learn out-of-class, then subsequent class time can better develop higher-order learning objectives in Bloom's Taxonomy. However this assumes students have remembered and understood content during pre-class study. We investigate this boundary condition in a statistical programming course, which is difficult to learn due to its abstract nature, and hence more sensitive to teaching quality. Using a pre-post-test design, we compare learning gains from the flipped classroom compared to alternative teaching methods: traditional and experiential classes. We find that students in the flipped classroom have smaller learning gains compared to students in the other treatments, and this is because learning gains decrease as flipped students do extra self-study: a finding not found in other treatments. These findings are consistent with our account: students pick up misunderstandings along the way which becomes more entrenched with self-study, and is more difficult for the teacher to rectify in subsequent class. This paper highlights the need – whatever the teaching method – to develop students' understanding and to correct misunderstandings before it is too late.