Conference Agenda

Background: Transactional sex is a critical public health concern given its associations with HIV/STI transmission and sexual violence. Little scholarship has focused on social networks as drivers of transactional sexual behavior.

Methods: We collected sociocentric network data from all adult men (age ≥18 years) who resided in eight villages in southwest Uganda. Network ties were elicited through five locally-adapted name generator questions on health, financial, emotional, food sharing, and leisure relations. Our outcome was self-reported engagement in transactional sex (i.e., exchanging money or gifts for sex) in the last year. Explanatory variables were having at least one male alter who engaged in transactional sex, total out-degree, perpetration of violence against women, and adverse childhood experiences. We used a multivariable logit regression model with cluster-correlated robust standard errors to estimate the association between transactional sex and these explanatory variables, adjusting for age, age at sexual debut, HIV status, marital status, and education.

Results: Of 791 participants, 93 (12%) engaged in transactional sex; and 240 (44%) had at least one male alter who had engaged in transactional sex. Participants with an alter who engaged in transactional sex were more likely to engage in transactional sex themselves (adjusted odds ratio [AOR]=1.91, 95% CI 1.13-3.23). Perpetration of violence against women (AOR=2.37, 95% CI 1.49-3.78) and adverse childhood experiences (AOR=1.12, 95% CI 1.01-1.25) were also associated with engaging in transactional sex.

Conclusion: Peer behavior may be an important driver of men’s engagement in transactional sex, suggesting the need for network-based, sexual health interventions.

There has been substantial research on the impact of social relationships on health. However, most studies focus on isolated aspects of individuals’ social networks, e.g., size, contact frequency, composition or perceived support. And most are conducted in high-income settings. We therefore examined personal network data from 1176 rural South Africans, mostly young adults, and used a network typology approach to simultaneously account for structural (size, density, overall support) and compositional (gender, location, kin) network characteristics. Using Wards-linkage hierarchical clustering we identified four distinct network types: single-alter low support, few-alters low support, few-alters moderate support and many-alters high support. We then used regression models adjusting for socio-demographics to assess the association of network types with key sexual and mental health outcomes: HIV and HSV-2 serostatus; risky sexual behaviour; self-reported knowledge of, misinformation on and capacity for safe sexual practices; and mental health and wellbeing. We found that while many-alter high support individuals were not less likely to be living with HSV-2 or HIV, they had the best mental wellbeing and were at lowest risk for almost all sexual health risk factors. In contrast, individuals with low support networks had significantly higher sexual risk behaviours and poorer mental health. These findings highlight the potential for identifying at-risk youth based on social network characteristics, even in a population with limited overall social support and very high health burden. They also suggest that social networks may play an important role in mitigating sexual and mental health harm amongst youth in resource constrained settings.

Food insecurity disproportionately affects pregnant and postpartum women, particularly those in structurally marginalized communities. Social networks are known to buffer against food insecurity, yet their role during crises remains understudied, particularly in low- and middle-income country contexts. This study explores the lived experiences of food insecurity among indigenous Tharu women in Nepal during the COVID-19 pandemic, with a focus on the structural and qualitative aspects of their social networks. Drawing on semi-structured interviews with ten women, we employed thematic analysis to examine how social networks shaped pathways into, experiences of, and coping strategies for food insecurity. Findings highlight how pre-existing structural vulnerabilities, including gendered power relations, economic dependency, and restrictive cultural norms, were amplified during the pandemic, increasing reliance on social ties that, in some cases, failed to provide support. Women with strong kinship-based networks exhibited greater resilience, leveraging relational wealth, the collective resources embedded in their social ties, to mitigate food insecurity. However, disrupted networks and negative social ties, particularly within patrilocal households, exacerbated food insecurity and emotional distress. Our results underscore the importance of incorporating social network-based interventions into maternal and child health programs, particularly during periods of crisis. Applying a qualitative social network approach, this study contributes to understanding how intersecting vulnerabilities shape women’s ability to navigate food insecurity and highlights the critical role of social networks as both protective and constraining forces.

This study examines the structural similarities and differences between opinion leader and friendship networks in a 2024 school-based smoking intervention in the Philippines and Indonesia. It explores how these networks may shape the spread of intervention messages, as intervention participants are embedded within them. Prior research in the US and Europe suggests that selecting either popular or trusted students as peer leaders effectively reduces health-risk behaviours. However, it remains unclear whether these strategies function similarly across different cultural contexts and how peer leader networks compare structurally to organic friendship networks.

Two key questions guide the study: (1) How do the structural properties of peer leader and friendship networks differ? (2) How do these patterns vary across schools and countries? ERGMs are used to model tie formation based on network properties, homophily in student attributes, and individual nodal characteristics such as gender and smoking behaviours.

Preliminary findings suggest that network structures show modest variation between peer leader and friendship networks within schools. Students with extremely high in-degree do not form significantly more ties in either network. While both friendship and peer leadership networks exhibit similar densities, peer leader network contains more closed triads. Smoking and gender homophily are key drivers of friendship ties, whereas peer leader nominations in the studied schools are less influenced by these factors. This indicates that while peer leadership networks share some structural features with friendship networks, their formation mechanisms differ, offering insights for optimising intervention message dissemination in schools.

For at least two centuries, the idea that population (or network) size would be a necessary asset for the acquisition and maintenance of complex socio-cultural knowledge at the population level has been generally accepted (Simmel, 1902; Carneiro, 1986). In the past 20 years, fields ranging from archeology (Shennan, 2001; Powel et. al. 2009; Marquet, et al., 2012; Vaesen, et.al., 2016; Collard et al., 2005; Henrich, 2017) to psychology (Caldwell, et. al, 2010; Kempe, et al, 2014; Muthukrishna et al, 2014) have explored this topic with experimental, modelling, observational, and social network analysis designs. This work has sustained an ongoing debate about the relevance of population size, including as it relates to how complex tasks or knowledge might actually be defined in diverse settings. Here, we use uncommon data from a field experiment that involved the implementation of a maternal and child health intervention in a long-duration RCT (Christakis & Airoldi, 2024). We also implemented a participatory methodology in which local community experts (N=15) were part of a consensus-building workshop (Brown & Isaacs, 2005) and ranked the knowledge outcomes included in the RCT as “easy” or “hard to remember” and as “complex” or “simple to learn.” Hence, using data for 8,050 people in 88 villages of varying sizes (20 to 400) in Honduras, we tested the relationship between population size and health knowledge uptake at the population level with respect to 18 diverse health attitudes and beliefs (ranging from newborn care to respiratory illness prevention, and of varying levels of complexity), measured over 8 years. After controlling for demographic, social network, and village characteristics, we find that population size did not affect knowledge uptake or persistence, regardless of the level of complexity of the outcomes. In the context of traditional, rural, and resource-limited populations living in a Low- and Middle-Income Country, our findings contradict the notions supported by social learning theory, diffusion of innovation theory, and demographic assessments that assert that having a larger pool of individuals to share newly acquired knowledge and beliefs would enable populations to adopt and sustain more complex and difficult forms of knowledge. Our work adds to the ongoing debate regarding population size and cultural evolution through a real-world experiment in the context of face-to-face social networks.

Background

Assessing the overlap between venue and HIV molecular networks can guide targeted interventions, potentially disrupting transmission.

Methods

From 07/2019-12/2022 persons newly diagnosed/living with HIV in Mexico City participated in a survey (asking where they met sex partners) and donated a blood sample. HIV partial pol sequences obtained via Sanger sequencing were used to infer a molecular network (pairwise genetic distance threshold <1.5% using Seguro HIV-TRACE). Individual risk scores were classified by sexual risk (sexual minority identity, drug use during sex, and <100% condom use). A value of ‘1’ was assigned to each (index averaging 0-3). Degree, eigenvector centrality, and compactness were characterized to assess venue co-affiliation network structure. Chi-squared tests and ANOVAs were used to compare the population by cluster status.

Results

Among 7502 participants, 50.9% were in HIV clusters, and 14.7% named venues. Participants in HIV clusters were significantly more likely to name venues (χ2(1)=24.3,p<0.001). Participants who named internet cafes had the highest mean risk score (Mean=1.67,SD=0.50), followed by nightclubs/bars (Mean=1.65,SD=0.59) and cabinas (Mean=1.59,SD=0.64). Sex work streets, community centers, and markets had the largest proportion of people in HIV clusters. The overall median degree, eigenvector centrality, compactness scores for venues were 15, 0.05, and 0.01. Public transportation had the highest degree and eigenvector centrality (indicating it is both popular and connects people), and community centers, centro commercial, and internet cafes had the highest compactness score (indicating nominations by an insular group).

Conclusion

Engaging both public transportation and internet cafes in HIV prevention interventions would be highly effective.