[Paper Review] Mining Valence, Arousal, and Dominance - Possibilities for Detecting Burnout and Productivity?

Mika Mäntylä, Bram Adams+3 more|arXiv (Cornell University)|Jan 1, 2016

Sentiment Analysis and Opinion Mining46 references83 citations

TL;DR

This study applies the Valence-Arousal-Dominance (VAD) emotional dimension model to 700,000 Jira issue reports and 2 million comments from open source projects to detect emotional patterns linked to burnout and productivity. It finds that higher issue priority increases Arousal, resolution boosts Valence—especially for reporters—and prolonged resolution times correlate with rising Arousal and falling Valence, suggesting VAD metrics can serve as early indicators of emotional strain in software teams.

ABSTRACT

Similar to other industries, the software engineering domain is plagued by psychological diseases such as burnout, which lead developers to lose interest, exhibit lower activity and/or feel powerless. Prevention is essential for such diseases, which in turn requires early identification of symptoms. The emotional dimensions of Valence, Arousal and Dominance (VAD) are able to derive a person's interest (attraction), level of activation and perceived level of control for a particular situation from textual communication, such as emails. As an initial step towards identifying symptoms of productivity loss in software engineering, this paper explores the VAD metrics and their properties on 700,000 Jira issue reports containing over 2,000,000 comments, since issue reports keep track of a developer's progress on addressing bugs or new features. Using a general-purpose lexicon of 14,000 English words with known VAD scores, our results show that issue reports of different type (e.g., Feature Request vs. Bug) have a fair variation of Valence, while increase in issue priority (e.g., from Minor to Critical) typically increases Arousal. Furthermore, we show that as an issue's resolution time increases, so does the arousal of the individual the issue is assigned to. Finally, the resolution of an issue increases valence, especially for the issue Reporter and for quickly addressed issues. The existence of such relations between VAD and issue report activities shows promise that text mining in the future could offer an alternative way for work health assessment surveys.

Motivation & Objective

• To explore whether emotional dimensions—Valence, Arousal, and Dominance (VAD)—can be automatically mined from software development communication to detect early signs of burnout and productivity decline.
• To assess whether VAD metrics correlate with issue report characteristics such as type, priority, and resolution time.
• To compare the predictive power of VAD against discrete emotions, sentiment, and politeness in explaining defect resolution time.
• To investigate bidirectional causality between emotional states and issue resolution dynamics in real-world software repositories.

Proposed method

• Applied a general-purpose VAD lexicon of 13,915 English words to extract emotional scores from 700,000 Jira issue reports and 2 million comments.
• Used statistical analysis and effect size estimation to examine relationships between VAD scores and issue metadata (e.g., type, priority, resolution time).
• Treated issue resolution as a temporal event and analyzed VAD evolution across the lifecycle of an issue report.
• Assessed the impact of VAD on defect resolution time using regression models, comparing it to sentiment, emotion, and politeness metrics.
• Validated findings through consistency checks with psychological theory, including the Yerkes-Dodson law and the role of control (Dominance) in burnout.
• Addressed potential confounding factors by assuming causal relationships based on empirical patterns and by ensuring data was collected without developer awareness.

Experimental results

Research questions

• RQ1 RQ1: How do VAD metrics relate to issue report characteristics such as type (e.g., Bug vs. Feature) and priority level?
• RQ2 RQ2: How do VAD scores evolve over time as an issue progresses from creation to resolution?
• RQ3 RQ3: Can VAD scores explain the time taken to resolve a defect, and how do they compare to other affective measures?
• RQ4 RQ4: Which issue characteristics predict high or low VAD scores in developer comments?

Key findings

• Higher issue priority (e.g., from Minor to Critical) is significantly correlated with increased Arousal, indicating heightened emotional activation.
• Issue type impacts Valence: Bug reports are associated with lower Valence (less pleasurable), while feature requests and improvements show higher Valence.
• Issue resolution is linked to a significant increase in Valence, with the strongest effect observed for the issue reporter rather than the assignee.
• Prolonged resolution time correlates with increased Arousal and decreased Valence, suggesting growing stress and emotional fatigue over time.
• VAD metrics explain a meaningful portion of defect resolution time, with effect sizes (e.g., d = 0.389) substantially higher than in prior emotion mining studies.
• The study reveals a bidirectional relationship: emotional states influence issue resolution dynamics, and unresolved issues in turn shape emotional responses, particularly increasing Arousal and decreasing Valence over time.