Sharing the load: contagion and tolerance of mood in social networks
Research output: Contribution to journal › Article › peer-review
Colleges, School and Institutes
- University of Oxford
The relations between self and others are fluid and constantly changing but exert a profound influence on our identity and emotional experiences. Indeed, human emotions are frequently and intensely social, and the people with whom we interact can alter our momentary mood. But does emotional ‘contagion’ extend over prolonged periods of hours to days, and if so, how does it propagate through interconnected groups? Answering this question is empirically challenging, since mood similarity in connected individuals can arise through multiple mechanisms (social influence, social selection, and shared external causation), making causal inferences hard to draw. We address this challenge using temporally high-resolution, longitudinal data from two independent, bounded social networks during periods of high communal activity and low external contact. Adolescent study participants (N=79) completed daily mood (n=4724) and social interaction (n=1775) ratings during residential performance tours of classical music lasting five to seven days. Analyses using statistical network models show that in both networks, adolescent musicians became reciprocally more similar in mood to their interaction partners. The observed contagion effect was greater for negative than for positive mood. That is, while one may ‘catch’ a friend’s bad mood, the friend may feel less negative in the process. These results suggest a mechanism for emotional buffering and the ‘cost’ of social support. We found no evidence for social selection based on mood. Indeed, participants were remarkably tolerant of their peers’ mood fluctuations, and showed no evidence of altering their patterns of social interaction accordingly.
|Early online date||28 Dec 2020|
|Publication status||E-pub ahead of print - 28 Dec 2020|
- Social networks, Emotion, Social influence, Mood contagion, Stochastic actor-oriented models