Investigating the Impact of Personality and Temperament Traits on Pair Programming: A Controlled Experiment Replication

This paper presents the partial results of an extended replicated pair programming experiment, conducted at the Alexander Technological Educational Institute of Thessaloniki, Greece during the two semesters of the year 2009-10. The entire study involved 160 second semester students attending a Software Engineering advanced programming course with pair programming and test-driven development as teaching methodology. A series of formal experiments were conducted to investigate, as in the first experiment, the impact of developer personalities and temperaments on pair performance. In this paper we report partial results and we focus on pair performance in terms of communication, time to complete assignments and overall score. The main variables were statistically analyzed in order to test differences between two personality groups: A control group with homogeneous personalities in pairs and an experimental one with heterogeneous personalities. The findings reported here concern only the first winter semester and confirm that pairs consisting of heterogeneous personalities and temperaments performed better than pairs with the same personality and temperament type. However, time to complete assignments was not statistically different between the two groups.