U-Belong Social Network Data: Egocentric Network Interviews and Survey Measures, 2023-2024

This deposit comprises two related egocentric social network datasets collected as part of the UK MRC–funded U-Belong project, which examines social belonging, social connection, and mental health among students in UK higher education. Together, these datasets capture both the structure and composition of students’ personal social networks and students’ perceptions and experiences of those networks, enabling detailed cross-sectional and longitudinal network analyses.

The first component (“From Familiar Faces”) consists of structured egocentric social network interview data collected from university students at up to two time points. Participants identified individuals (“alters”) in their personal social networks and provided information on relationship type using broad, pre-specified categories (e.g. friend, family member, romantic partner, classmate, flatmate, university staff), emotional closeness, frequency and mode of contact, and perceived support. Participants also reported whether alters knew one another, allowing the construction of alter–alter ties and the derivation of network structure measures such as network size, density, interconnectedness, and composition over time.

The second component comprises complementary survey-based social network measures. These data focus on aggregate and compositional characteristics of students’ social networks, including perceived network size, diversity, availability of support, and patterns of social connection across different relational contexts. These measures are designed to be analytically compatible with the interview-based network data, enabling comparison between detailed egocentric network structures and broader self-reported network characteristics.

Across both components, the datasets include participant-level variables (with pseudonymous identifiers enabling longitudinal linkage), alter-level variables (where applicable), tie-level indicators, and derived network measures. No names or free-text descriptions of network members are included. Relationship types are restricted to broad categorical descriptors, and temporal information has been minimised to reduce disclosure risk.