2023 ASEE Annual Conference & Exposition

Pair programming is an active learning technique where two students share a single screen to complete a project synchronously. The practice confers several benefits; however, its full potential is undercut when pair breakdown occurs. The strategies for optimal pairing of students to alleviate this breakdown, allowing students to effectively collaborate and communicate, are multi-faceted and remain an area with a large gap in knowledge. Strategies can be further complicated by unexpected events requiring a shift from traditional learning environments. Our work identifies strategies to optimize student success by examining pair performance in two sections of an upper-level computer science course at a public university where a majority of students (69%) chose to pair program remotely. Data on several key factors were gathered and analyzed for their effect on pairs: programming confidence and experience, gender, preferences toward deadlines, communication style, and leadership style. These factors were examined for their effect on assignment and exam scores using backward stepwise regression. We found that paired students with similar programming confidence performed 11% (p=.036) higher on assignments, while students in pairs with dissimilar communication styles scored 14% (p = .006) higher than those whose styles were similar. On exams, being in a pair with similar, but not the same, preference toward others leading resulted in a 10% higher average score (p=.014). Some factors impacted male and female students differently. Male students in pairs with similar preference toward others leading scored 11% higher on exams (p=.016), while female students in pairs with the same preference toward deadlines scored on average 13% higher on exams (p=.032). These findings show that similarity in some factors (confidence), while diversity in others (working styles, communication styles) are needed to optimize student success in pair programming teams and support women in computer science.