Exploring Pair Programming in Distributed Object-Oriented Team Projects

Author(s): Baheti, P., Williams, L., Gehringer, E., Stotts, D.
Venue: OOPSLA Educator's Symposium 2002
Date: 2002

Type of Experiement: Quasi-Controlled Experiment
Sample Size: 132
Class/Experience Level: Graduate Student
Participant Selection: self-selected class participation
Data Collection Method: Survey, Code Metric, Project Artifact(s)


Link: http://collaboration.csc.ncsu.edu/laurie/Papers/EdSym_DistPP_FinalSubmis...

How Pair Programming Was Used: 5 week team project, 4 types of teams: Pair Programming or not, Collocated or Distributed

In this experiment, the authors are trying to determine if the practice of pair programming is beneficial to distributed pairs. They allowed students to choose their team, and the team type as they wished. Distributed pair programming teams used a variety of software to perform the practice of pair programming, but essentially they used a screen sharing program, and some sort of audio chat. Students used an online tool called Bryce to record metrics of software development, but due to mistakes by many teams, only the data on LOC/hr was able to be used in analysis. The authors found that distributed teams had a slightly greater productivity, but it was not statistically significant. The same thing was true for student grades. Student feedback indicated that communication and cooperation was better in pair programming teams than others, as might be expected. With regard to distributed pair programming specifically, students did not find that the technology was a hindrance. The authors conclude that more work needs to be done to be able to draw statistically significant conclusions, however they did find that it was a feasible method for team projects, as students were comparably successful, and did not have significant problems.