Impact of Pair Programming Dynamics and Profiles to Pair Success

Document Type


Publication Date



In this paper, we investigate the convergence and leader-follower patterns of pairs of novice programmers as they traced and debugged fragments of code and the impact of these patterns against the success of programming pairs. We performed a dual eye tracking experiment, recorded their fixations and computed for the recurrence rate and the diagonal recurrence profile using Cross-Recurrence Quantification Analysis (CRQA). Results showed that programming pairs who chat and work on the same program together and converge more often significantly perform better than those who do not. The highly proficient but poorly acquainted pairs, despite of poor convergence, still perform the best. On the other hand, low proficiency pairs that are highly acquainted have the weakest performance even if they converge very well. Findings also revealed that there is a significant difference as to who is leading who. The more successful participants within pairs are leading the less successful participants majority of the time. This study is important because it provides information about the dynamics that may likely occur in a pair programming setup.