Network Effects on Health Behaviors: Examining the Effects of Norms, Social Support, and Social Capital

Abstract

Health behaviors have a significant effect on health and mortality, and one’s social network is strongly associated with behaviors, for better (positive encouragement to eat healthier) or worse (peer pressure to drink excessively). This dissertation aims to understand how the mechanisms of peer influence and network social context affect health behaviors across the life course. I examine the effects of networks across a wide demographic and substantive range – from college students’ sleep behaviors in Chapters Two and Three to older adults’ preventive health behaviors in Chapter Four. This breadth allows me to examine multiple social mechanisms (peer influence, social support and network position/contexts) that speak simultaneously to many open questions in the social embeddedness of health. My results confirm that health behavior is socially conditioned and that our understanding of people’s own health behaviors requires understanding their network activity.Chapter Two examines the effect of peer influence on sleep duration, bedtime, waketime, and the daily variance of each measure among college students. Sleep duration, timing, and variability have significant health impacts; irregular and reduced sleep are associated with cardiovascular risk, obesity, mental health outcomes, and overall mortality (Luyster et al. 2012, Bei et al. 2016). Despite documented effects of peers on sleep in specific populations such as middle-schoolers, the effects of peers on sleep timing and duration amongst college students is poorly understood. Linear mixed-effect models show a strong positive peer effect across all metrics, even accounting for school/work schedule constraints and overall peers’ sleep behaviors. For example, for every standard deviation increase in sleep duration among one’s peers for that night, the participant sleeps an additional 10 minutes. This effect is similar but varies in size across the metrics, suggesting that sleep timing and duration is socially conditioned by peers. The study implies that interventions need to recognize the consistent effect that peer influence holds, net of schedule and homophily effects.Chapter Three examines how network structure affects sleep behavior through the competing demands framework, social support, and stress. Both sleep insufficiency and irregular timing correlate with poor health. College students consistently sleep less and with more irregular timing than recommended; however, a majority of sleep research is focused on adolescents due to the importance of early forming of sleep habits. Educational interventions among this population have largely been unsuccessful; however, if peer-led interventions are to be attempted, the method for choosing a ‘change agent’ must be examined in the context of sleep behavior. Degree (reflecting high social obligations), eigenvector (reflecting embeddedness in a core influential group), and betweenness centrality (reflecting bridging between two groups) are often chosen due to their large reach within a network, but the high demands, expectations, and stress as a result of their positions may lead to worse sleep behaviors, making them an ineffective change agent. Longitudinal data from 246 college students and their peers are used to identify the effect of network structure on sleep duration, bedtime, waketime, and sleep regularity. I find that students with higher betweenness centrality have lower sleep duration and lower values on the Sleep Regularity Index (SRI), and that those with larger networks go to bed earlier and wake up later than those with smaller networks, indicating healthier sleep behavior. The effects of eigenvector centrality on sleep regularity differ based on sex: for men it leads to lower regularity and later bedtimes, while for women it leads to higher regularity and earlier bedtimes. While we have little data on the exact mechanisms driving these differences, I speculate that group identity or differing strategies in managing social obligations play a role in driving this relationship. These results suggest that an individual’s position in their network influences their sleep behaviors and that future research on sleep habits needs to account for network embeddedness to better disentangle peer processes associated with heightened youth social involvement from individual health practices.Chapter Four examines how social capital influences health behaviors in older adults, while accounting for potential reverse causality. Health behaviors are crucial for older adult’s health and reduction of secondary disabilities, but are understudied. Some aspects of networks such as social capital have been shown to influence health behaviors in older adults, but most studies are cross-sectional and do not take into account causal direction. Using exploratory factor analysis and auto-regressive cross-lagged panel models, I test the causal direction of social capital and health behaviors. I find that earlier measures of network social capital lead to increased physical activity and vitamin usage in later waves, as well as a reduction in smoking. These relationships largely hold when put in the reverse causal direction, although the effect for vitamins disappears and an additional effect for binge drinking (leading to decreased social engagement) emerges. These results suggest that older adults with high social capital networks have better preventive health behaviors and are less likely to smoke, and that health behaviors hold relatively similar reverse causal effects on networks.