Abstract:

Many mathematically intensive engineering courses lose students when instruction begins with abstract formulas that only later connect to applications. Here, we report the study of a backwards design process implemented in a large-enrollment introductory electric circuits course to address these issues. We reframed the core learning objectives of the course and refined the design of two learning modules, each with a different pedagogical approach. The first approach was that of Problem-Solving before Instruction (PS-I), where learners first engage (or struggle) with problem-solving activities, structured to different degrees, and only then are presented with the canonical solution. This approach is also known to produce a productive failure, an approach shown to support the transfer of learning. The second approach is that of a flipped classroom, where students first learn the canonical solution (usually on their own using recordings or textbooks) and then experience problem-solving, in a guided way, with the instructor or with peers. This approach allows Instruction-before-Problem-solving (I-PS) and was shown to support procedural knowledge, often required in engineering education.

In the current work, we therefore compare students' performances, experiences, and participation comparing two learning modules - one that uses PS-I and the other that uses I-PS in an intermediate large-enrollment engineering course. This redesign allows the evaluation of the impact of the different designs for learning on the transfer of learning, student attendance, and engagement, as well as the scalability of these pedagogies.

Our research questions are:
(RQ1) Which approach better promotes understanding and transfer of learning: PS-I or I-PS?
(RQ2) Do these approaches scale effectively to a large-enrollment setting?
To answer RQ1, we will compare a single cohort (n>400) on the two learning units in a single course run (within-subjects). To answer RQ2, we compare the PS-I module implementation of the 2025 large-enrollment course run with that of the smaller 2024 pilot run in a between-subjects design.

Evaluation integrates common assessment items, a retrospective post-then-pre survey of perceived learning, cognitive effort, and confidence, and classroom observations. Data collection is underway (completion planned for Feb 2026, before the conference). However, preliminary results of the 2024 run suggest that PS-I improved attendance, engagement, and supported the understanding and transfer of material. In the current work, we will assess how well PS-I scales and whether these benefits are maintained when compared to I-PS. The presentation will report additional results and share a replicable implementation blueprint: activity templates, facilitation moves, and timing strategies, to support adoption in large-enrollment, mathematically intensive courses while balancing coverage demands with deeper, transferable learning.

Keywords:

Engineering education, problem-solving, backwards design, scalability.