How Pairing by Code Similarity Influences Discussions in Peer Learning

Shiyu Xu, Ashley Zhang, and Steve Oney

Peer learning, as a form of collaborative learning, has been widely used in programming courses as a means of promoting active learning and enhancing students' programming skills. However, it is challenging for instructors to group students effectively so that they can have fruitful conversations. We conducted a study with 15 students from an introductory programming course to investigate whether and how grouping students with similar or different solutions affects the discussions that take place within groups. The findings indicate that pairing students by the similarity of their code might influence students' learning and coding skills. Specifically, students who were paired with people that had different solutions had, on average, more engaging conversations and were more likely to write more diverse solutions in the future. The results also highlight the need for tools to facilitate the pairing process in programming courses in order to optimize the learning outcomes for students.