Peer instruction: a link to the exam

Abstract

In computer science, the active learning pedagogical practice of Peer Instruction (PI) has been shown to improve final exam performance, reduce student failure rates, and improve student retention. PI consists of two major parts: group discussion and follow-up instructor intervention. We expect that PI performance as a whole will correlate with final exam performance, but it is unclear whether or how each piece of PI is involved in these relationships. In this work, we use isomorphic questions to isolate the effects of peer discussion and instructor intervention, and examine scores on a final exam and its code-writing and code-tracing questions. We find that both pieces of PI correlate with the final exam as a whole, code-tracing question (similar to PI questions), and code-writing question (not similar to PI questions). This is further evidence that both PI components are important to the success of PI.