Developer coordination in software engineering projects

Software development has evolved to become more complex and distributed. This highlights the importance of coordination between developers, which is known to impact development performance. Measuring developer coordination is complex but being able to understand its role in software engineering projects requires an accepted way of measuring it. One approach used to conceptualize and measure coordination is known as ‘Socio-Technical Congruence’ (STC), developed by Cataldo et al. (2006), which is the fit between required coordination and actual coordination measured as the proportion of actual coordination activities that occurred relative to the number of coordination activities that should have taken place (that is, required coordination). Research to-date suggests that STC is a useful predictor of development task performance in software projects. However, STC is relatively new and has been empirically evaluated in only limited software engineering contexts. Also, it has focused on explicit developer coordination. Coordination also happens implicitly and other project contexts may yield additional insights. Therefore, the opportunity and need exists to further investigate STC as a measure of developer coordination in software engineering projects. Consequently, the aims of this dissertation are threefold: 1) to validate the use of STC as a measure of explicit coordination to predict task performance in a contrasting project setting and test out its boundaries; 2) to re-examine developer coordination in the literature, focusing on group awareness as an explicit and implicit coordination mechanism for overcoming challenges of complexity and distribution in software development projects; and, based on this re-examination, 3) to extend and validate the application of STC to measure implicit developer coordination as well as explicit coordination. The first aim was fulfilled by performing a differentiated replication study of Cataldo et al.’s (2006) seminal research to investigate STC in new software engineering contexts, namely in eight open source development projects of different sizes (the original study used a single large in-house commercial project). In contrast to the original study, findings from this research indicated that STC is only effective as a measure of explicit developer coordination in large projects (that is, projects with 50 or more developers). Results for projects smaller than 50 developers showed variable, mostly non-significant results. This unexpected finding is discussed and explained, based on the nature of coordination in different sized groups. The second aim was fulfilled by exploring developer coordination and the role of group awareness in facilitating coordination in the literature. Following an initial overview of prior research on developer coordination, two Systematic Literature Review (SLR) studies were conducted. The first SLR was used to identify coordination issues and causes of coordination difficulty in software engineering projects, particularly in software engineering projects. A key finding of this study related to the role of group awareness in coordination in software engineering projects. Group awareness helps development team members to coordinate explicitly and implicitly. A second SLR study was then conducted to investigate further the role of group awareness as a coordination mechanism. A key finding from this study was that group awareness arises primarily through familiarity of group tasks, group members and the group work context. Together, these studies provided strong theoretical support for using implicit coordination in calculating STC as a measure of development task performance. Finally, the third aim was fulfilled by empirically re-examining the eight open source projects using a measure of implicit developer coordination for STC, motivated by group awareness (in contrast to the explicit coordination measure used in the replication study, following the original study). The study found support for STC as a measure of implicit coordination, not only in the large projects but the small ones as well. In extending application of the STC measure of developer coordination, the results highlight the important role of explicit and implicit coordination as a predictor of task performance and suggest that the relevance of STC may be broader when applied to implicit coordination. This dissertation contributes to the body of knowledge on developer coordination in software engineering. First, it validates and qualifies the use of explicit coordination-based STC as a predictor of development task performance in a replication study in a contrasting setting. Second, it synthesizes current knowledge on coordination challenges and response strategies, and the role of explicit and implicit group awareness as a coordination mechanism in group work contexts. Third, it validates the use of an implicit coordination-based measure of STC as a predictor of development task performance in the same contrasting setting and clarifies the boundaries of STC’s use in projects of different sizes. The studies find that STC is effective as a measure of explicit developer coordination only in large projects, where project complexity makes explicit coordination more relevant.

Grantor dc:publisher

UNSW, Sydney

Year dc:date

2014

Author dc:creator

• Mohd Fauzi, Shukor Sanim

• Software Engineering
• Socio-Technical Congruence
• Coordination
• Explicit Coordination
• Implicit Coordination

Statement dc:rights

• open access
• CC BY-NC-ND 3.0
• free_to_read

Licence

• https://purl.org/coar/access_right/c_abf2
• https://creativecommons.org/licenses/by-nc-nd/3.0/au/

Language dc:language

EN

Identifier

https://doi.org/10.26190/unsworks/18665

OAI identifier oai:identifier

oai:unsworks.library.unsw.edu.au:1959.4/55420