Factors Affecting Students’ Performance in Distributed Pair Programming

Abstract

Pair Programming has been shown to increase productivity and code quality not only in professional software development but also in the context of programming education. The provision of broadband Internet access gave rise to Distributed Pair Programming (DPP) enabling two programmers to collaborate remotely. To gain insight into the benefits of DPP, we performed an empirical study on an object-oriented programming course where 62 students carried out assignments through a DPP platform. The goal of the study is to investigate, in the context of DPP, whether prior programming skills (assessed at the level of student, his or her partner and pair) and pair compatibility are related to student performance. To further examine the effect of DPP on learning outcomes, we have studied whether a pair’s performance on DPP assignments is related to the students’ grade. The findings indicate that the student’s actual skill and the pair’s actual skill affect his or her performance in an object-oriented programming course. The results also suggested that there is no association between pair compatibility and his or her own performance. Finally, pair performance on DPP assignments is related to the individual student performance in the final exams. Such evidence can be used to guide instructors when planning DPP assignments and especially when forming student pairs.