Disentangling the Network Structure of Online Social Support: A Multilayer Network Analysis of a Twitter Long COVID...

Online social support has been shown to be an important resource for people experiencing chronic illnesses. Although numerous studies have examined online support communities, few have conducted in-depth analyses grounded in social network theory and methodologies. To address this gap, this study investigates the types of support and the overall patterning of such exchanges (i.e., the network structure) on Twitter (now X), focusing on a community of people who seek and provide support for Long COVID. Using an exhaustive keyword-based corpus of Long COVID tweets (N = 19,196) collected from August 1-31, 2021, we identified two types of social support shared in this community-informational and emotional-through human annotation combined with supervised machine learning. Conducting descriptive and stochastic social network analysis, we examined how exchanges of each type of support were structured. We focused on the density, reciprocity, and centralization of support ties-three structural features shown to be related to specific types of support. Our results showed that informational support was more prevalent than emotional support. Additionally, these informational and emotional support networks tended to be both reciprocal and concentrated around a small number of central users who received considerable support (i.e., were centralized). Our findings suggest that online health communities hosted on platforms such as Twitter may be better suited for informational support. We also observed that central actors play an important role in facilitating these exchanges. These findings provide practical insights for effective social network interventions.