The First Workshop on AI-supported Education for Computer Science (AIEDCS)

We explore the sequences of affective states that students experience during their first encounter with computer programming. We conducted a study where 29 students with no prior programming experience completed various programming exercises by entering, testing, and running code. Affect was measured using a retrospective affect judgment protocol in which participants annotated videos of their interaction immediately after the programming session. We examined sequences of affective states and found that the sequences Flow/Engagement ↔ Confusion and Confusion ↔ Frustration occurred more than expected by chance, which aligns with a theoretical model of affect during complex learning. The likelihoods of some of these frequent transitions varied with the availability of instructional scaffolds and correlated with performance outcomes in both expected but also surprising ways. We discuss the implications and potential applications of our findings for affect-sensitive computer programming education systems.