Pair versus Solo Programming -- An Experience Report from a Course on Design of Experiments in Software Engineering
This provides incremental evidence for educators and practitioners on the trade-offs between pair and solo programming in software development training.
The paper evaluated the duration and effort of pair versus solo programming in a software engineering course experiment, finding that pair programming reduced duration by 28% while solo programming reduced effort by 30%, with results closely matching a 1998 study.
This paper presents an experience report about an experiment that evaluates duration and effort of pair and solo programming. The experiment was performed as part of a course on Design of Experiments (DOE) in Software Engineering (SE) at Autonomous University of Yucatan (UADY). A total of 21 junior student subjects enrolled in the bachelor's degree program in SE participated in the experiment. During the experiment, subjects (7 pairs and 7 solos) wrote two small programs in two sessions. Results show a significant difference (at alpha=0.1) in favor of pair programming regarding duration (28% decrease), and a significant difference (at alpha=0.1) in favor of solo programming with respect to effort (30% decrease). With only a difference of 1%, our results regarding duration and effort are practically the same as those reported by Nosek in 1998.