: Julia Schenk, Lutz Perchelt< Stephan SalingerVenue
: International Conference Software Engineering Date
: 2013Type of Experiement
: Controlled ExperimentSample Size
: 2Class/Experience Level
: Undergraduate StudentData Collection Method
The Authors stated that pair programming could be useful to “reduce elapsed time, to reduce defect density, to improve program design, to make sure more than one person is familiar with each part of the code, to increase the amount of knowledge available when solving a task, to increase focus and keep up discipline, to accelerate learning, and to build within-team trust.”
The authors main goal from this paper was to compare the Local Pair Programming(LPP) vs Distributed Pair Programming (DPP), and they based their research on studying one pair of programmers. They noticed several differences between those two methods.
The method the authors used is to have two programmers do a DPP while both of them and their screens were being videoed. They used Grounded Theory Methodology(GTM) to analyze the data they collected from the experiment. They noticed that sometime the observer would assume the position of the driver, and he would just do the job himself. Another observation was that both programmers use the freedom of editing the code and work on different areas in order to achieve the same goal.
From their experiment the authors concluded that there are two main difference between LPP and DDP. The first main difference authors mentioned is that in local PP programmers send a lot of cues from posture, gesture, gaze direction, facial expression, and vocal expressions while in DPP either these cues are partially transmitted or they don’t exist. The other difference is that in DDP both programmers can become solo programmer and work on solving the same problem on their own machine while they can easily switch back to PP when needs be. On the other hand in LPP only one programmer is able to do the typing.
Overall they authors believe that DPP is more effective that LPP.