Title:Adopting Softer Approaches in the Study of Repository Data: A Comparative Analysis

Abstract:Context: Given the acknowledged need to understand the people processes enacted during software development, software repositories and mailing lists have become a focus for many studies. However, researchers have tended to use mostly mathematical and frequency-based techniques to examine the software artifacts contained within them. Objective: There is growing recognition that these approaches uncover only a partial picture of what happens during software projects, and deeper contextual approaches may provide further understanding of the intricate nature of software teams' dynamics. We demonstrate the relevance and utility of such approaches in this study. Method: We use psycholinguistics and directed content analysis (CA) to study the way project tasks drive teams' attitudes and knowledge sharing. We compare the outcomes of these two approaches and offer methodological advice for researchers using similar forms of repository data. Results: Our analysis reveals significant differences in the way teams work given their portfolio of tasks and the distribution of roles. Conclusion: We overcome the limitations associated with employing purely quantitative approaches, while avoiding the time-intensive and potentially invasive nature of field work required in full case studies.