Abstract
Online brand communities (OBCs) are vital strategic assets for firms, facilitating direct interaction and consumer feedback. A positive emotional atmosphere is essential for community health and member retention, yet the dynamic evolution of these emotions is often misunderstood. This study investigates emotional contagion among three sources: original posters, other community members, and official firm feedback. By analyzing longitudinal data from the MIUI forum using machine learning and Vector Autoregressive (VAR) modeling, the research examines both emotional valence (mean mood) and volatility (mood stability). The findings reveal that while firm feedback is a primary driver of community sentiment, these emotional contagions are remarkably fleeting, typically dissipating within 72 hours.
Introduction
Maintaining a healthy "vibe" in an Online Brand Community is more than just a PR exercise; it is essential for long-term customer loyalty. However, emotions in these digital spaces are not static. They ripple through the community as members interact with each other and with the brand. Despite its importance, the dynamic nature of how these emotions spread—a process known as emotional contagion—has remained under-explored. This paper fills that gap by mapping the multidimensional emotional interactions between posters, peers, and official staff.
Methodology
The researchers developed a multi-stage analytical pipeline to track emotional "ping-pong" over time:
- Sentiment Analysis: Utilizing the Jieba library for Chinese word segmentation and a dictionary-based machine learning approach, the team calculated scores for both valence (positivity/negativity) and volatility (fluctuation of sentiment) across 17,622 posts and nearly 100,000 replies.
- Vector Autoregressive (VAR) Modeling: This time-series method captured the causal, reciprocal relationships between the emotional states of the three sources.
- Impulse Response Function (IRF): This allowed for the simulation of "shocks"—such as a sudden burst of official positivity—to observe how the community reacts over a 10-day window.
Data and Model Reliability
The study utilized data from the MIUI forum, Xiaomi’s official user community. To ensure the data reflected standard emotional contagion rather than external crises, the researchers selected a stable period between major software updates. To ensure the findings were statistically sound, the researchers confirmed the stability of the VAR model.
Key Findings
The Power of the Firm
Variance decomposition analysis revealed that after a source's own emotional history, the firm's official feedback is the most significant external driver of member sentiment. Specifically, official responses explained approximately 8.8% of the variance in member post valence—the highest contribution from any external source. This underscores the high ROI of active, positive community management.
Volatility is Contagious
The study found that it isn't just what people feel that spreads, but how stably they feel it. High emotional volatility in one group (e.g., erratic feedback from other members) can trigger instability in the original poster’s sentiment. Managing the "calmness" of a community is as important as managing its positivity.
The Fleeting Nature of Contagion
Perhaps the most critical finding for community managers is the short lifespan of emotional shocks. Whether a shock is positive or negative, its impact peaks quickly and typically returns to baseline within three days.
Practical Implications
For brands, these results offer two clear directives:
- Proactive Shaping: Firms can and should proactively influence community mood by carefully designing their official feedback. Because the firm acts as an emotional anchor, its "voice" carries significant weight in stabilizing and boosting member sentiment.
- Consistency is Key: Because emotional contagion is fleeting, community management cannot be a sporadic effort. To maintain a positive atmosphere, firms must engage consistently. A single positive interaction will not sustain a community for a week; frequent, daily engagement is required to keep the "vibe" alive.
Conclusion
By distinguishing between emotional sources and examining both valence and volatility, this study provides a granular map of digital emotional contagion. While the fleeting nature of these interactions presents a challenge for consistency, it also offers a "reset" button for communities facing temporary negativity, provided the firm remains an active and positive participant in the conversation.
Summary updated