Evaluating Solo vs Pair Programming in an Online Setting for Introductory Programming Students.

Many studies have shown the efficacy of pair programming for students learning to program. However, most of these studies have taken place in an in-person environment, where the driver and navigator are physically sharing a keyboard and screen and can communicate verbally and non-verbally. With the increase in online learning, especially during the COVID-19 pandemic. It is important to know whether these results generalize to an online environment. In this work, we develop a methodology to replicate existing pair programming research in a remote context. Students can fulfill the same driver and navigator roles and share access to a single IDE. However, communication is limited to video chat, and participants can never physically interact. This will allow us to replicate various studies, evaluating the efficacy, perceptions, impacts, and perceptions of solo vs. pair programming. An initial study of 116 students enrolled in an introduction to programming course validated our experimental setup and showed that pair programming positively impacted the completion and correctness of programming exercises in an online environment. With 67.3% of pair programming submissions passing at least one test case, vs. 55.3% in the solo programming condition, and 63.5% of pair programming submissions passing all test cases vs. 45.0% of solo submissions (p < 0.02). This work validates our experimental design and shows promise that future work will be able to replicate many additional pair-programming studies in an online environment.