Sampling Affects of Software Developers to Understand Individual & Team Performance

Abstract

Background Software development is human-centred and consists of intellectual activities and teamwork which requires people skills. However, work towards improving individual and team performance has put much effort into improving technology and processes, and less on human factors such as moods or feelings. Affect can be used as an umbrella term for moods and emotions, and this thesis adheres to the dimensional approach of affects, in which valence, arousal, and dominance describes people’s state of feeling. Objective The purpose of this thesis is to understand the impacts of valence, arousal, and dominance on individual and team performance of software developers. Method Experience sampling method (ESM) was chosen for collecting data on affects as it was created to study what people do, feel and think in their natural settings. Data on affects and individual performance was collected with ESM and analysed using manual interpretation and linear mixed-effects model (LMM). Team performance data was gathered using self-assessment surveys and compared with the affect results from the ESM study. Data were analysed using manual interpretation and Kendall’s tau-b correlation. The study was conducted in an industrial setting consisting of 28 developers in 4 teams from Volvo Car Retail Solutions. Results Results showed that valence have a significant impact on individual performance. For team performance, manual interpretation indicated a close relationship between valence and team performance. Conclusions We demonstrated how performance and human factors could be measured, and showed how data could be analysed using a LMM, manual interpretation, and Kendall’s correlation. Results showed a significant correlation between affects and performance. The results introduces a new perspective by including both individual and team performance in an industrial setting.