Time: Thursday, 26/June/2025: 10:00am - 11:40am Session Chair: Jennifer Lawlor |
Location: Room 202 |
Using hybridised weighted centrality measure to identify cliques and subgraphs of a community structure
DELTA STATE POLYTECHNIC, OTEFE-OGHARA, Nigeria
Identification of key nodes in complex networks is the driving force that controls or informs of the situation of such a network. Most real-world network systems are shown to be graphs of weighted networks (i.e. networks with link-weights and/or node-weights). In this paper, considerations are given to the weighted traditional centrality measures of degree, betweenness and closeness of graphs and their shortcomings identified/addressed in comparism with hybridised centrality measures which combined isolated centrality (with the traditional centralities), for the purpose of identifying the most influential nodes. This has a good impact on the dissemination of information and helps with the identification of the subgraphs/cliques and in turn the community structure. However, despite some few disadvantages of these hybridised centrality methods, such as high cost of computation and reliance on tuning parameters, the new method is seen to have an improvement of close to 50% over the traditional methods.
Advancing Methodology of Chosen Family and Kinship in Social Network Analysis for LGBTQ+ Health Equity
1Department of Health Management and Policy, University of Michigan School of Public Health, United States of America; 2Department of Medicine, Health, and Society, Vanderbilt University, United States of America
Purpose: This study employs a novel, theoretically-driven social network analysis methodology to capture the complex, non-traditional kinship experiences of queer and trans adults. We examine how chosen family ties may differ from other relationship ties regarding closeness, social support, and mental health among queer and trans individuals.
Data: We use two egocentric network datasets: (1) Queer and Trans Vietnamese American Advocates Network (QTVAAN; N=38 egos, 628 alters), and (2) LGBTQ+ Social Networks, Aging, and Policy Study (QSNAPS; Wave 3; N=981 egos, 13,412 alters).
Methods: We use descriptive statistics to identify group differences in closeness, relationship length, and social support across chosen family, family of origin, and friend only relationship ties; and multi-level mixed-effects negative binomial regression models (alter-level characteristics nested within ego outcomes) to assess the impact of percent of chosen family ties within a network on mental health.
Findings: In both datasets, "chosen family" ties represent ~10% of all ties and were predominantly co-identified as “friends”; they were long-term, averaging 15-20 years. Egos in both datasets reported the highest proportions of feeling close to chosen family alters (at least 70% of these ties considered very close). There were no differences in providing instrumental support by relationship type in both datasets. Mixed-effects model results showed non-significant associations between percent of chosen family ties in network with ego distress, social anxiety, and social well-being among QTVAAN respondents. Despite the null effects on mental health outcomes, there are underlying social dynamics of long-term closeness among chosen family ties that warrant further investigation.