ABSTRACT
Participation in communities is essential to individual mental and physical health and can yield further benefits for members. With a growing amount of time spent participating in virtual communities, it's increasingly important that we understand how the community experience manifests in and varies across these online spaces. In this paper, we investigate Sense of Virtual Community (SOVC) in the context of live-streaming communities. Through a survey of 1,944 Twitch viewers, we identify that community experiences on Twitch vary along two primary dimensions: belonging, a feeling of membership and support within the group, and cohesion, a feeling that the group is a well-run collective with standards for behavior. Leveraging the Social-Ecological Model, we map behavioral trace data from usage logs to various levels of the social ecology surrounding an individual user's participation within a community, in order to identify which of these can be associated with lower or higher SOVC. We find that features describing activity at the individual and community levels, but not features describing the community member's dyadic relationships, aid in predicting the SOVC that community members feel within channels. We consider implications for the design of live-streaming communities and for fostering the well-being of their members, and we consider theoretical implications for the study of SOVC in modern, interactive online contexts, particularly those fostering large-scale or pseudonymized interactions. We also explore how the Social-Ecological Model can be leveraged in other contexts relevant to Computer-Supported Cooperative Work (CSCW), with implications for future work.
