The formation and influence of online health social networks on social support, self-tracking behavior and weight loss outcomes

The current dissertation provides an examination of online health social networks in intentionally designed health-related social media. It investigated the theoretical mechanisms that drive the formation of online health social networks, examined the joint dynamics of network selection and network influence on individual health outcomes, and tested a model of ego network structure, social support and self-tracking behavior. The dissertation is situated in an entrepreneurial online weight management social networking site, FatSecret. It employed a four-month longitudinal study design, and collected unobtrusive behavioral data extracted from the site and self-reported data from an online survey with the users of FatSecret. ❧ Drawing on MTML framework, the first study found that demographic homophily including age and gender similarity, inbreeding homophily in the form of group affiliation, and health-related homophily including initial health status and health progress significantly predicted the selection of online health buddies. Similarity in health goal was not a significant predictor due to its strong correlation with initial health status in the specific health issue of weight loss. A person’s frequency of updating personal health information was marginally significant in predicting tie formation. By conducting a SIENA analysis, the study also found empirical support for a social influence effect among health buddies, such that an individual’s weight outcome tended to become similar to the average of the individual’s health buddies’ weight outcomes. The second study drew from literature on structural and functional social support, the buffering model of social support and self-regulation depletion studies. By conducting SEM analyses, the study found that both the size and triadic closure of an individual’s ego network predicted perceived social support for weight loss in FatSecret. Then, perceived social support predicted more active self-tracking behavior, and self-tracking behavior predicted improved health progress. Post hoc analysis showed that there was a negative and significant effect of health progress on self-tracking behavior, such that improved health progress at an earlier time point would reduce the amount of self-tracking at a later time point. Finally, the implications of designing effective web-based weight loss interventions by better organizing and engineering peer-to-peer social networks were discussed.

本博士论文针对刻意构建的健康类社交媒体中的在线健康社交网络展开系统性研究。本研究探究了驱动在线健康社交网络形成的理论机制，考察了网络选择与网络影响对个体健康结果的联合动态作用，并检验了自我中心网络（ego network）结构、社会支持与自我追踪行为的关联模型。
本研究以创业型在线体重管理社交平台FatSecret为研究场景，采用为期四个月的纵向研究设计，采集了从该平台提取的非介入式行为数据，以及针对FatSecret用户开展的线上问卷所收集的自我报告数据。
❧ 本研究依托多特质-多方法（MTML, Multi-Trait Multi-Method）框架开展第一项研究，结果表明：人口统计学同质性（涵盖年龄与性别相似性）、以群体归属为表现形式的近亲同质性，以及包含初始健康状态与健康进展的健康相关同质性，均能显著预测在线健康伙伴的选择。由于在体重管理这一特定健康议题中，健康目标相似性与初始健康状态存在较强的相关性，因此其并未表现出显著的预测效应。个体更新个人健康信息的频率对社交联结形成的预测作用仅呈现边际显著性。通过SIENA分析，本研究还为健康伙伴间的社会影响效应提供了实证支持：个体的体重结果会逐渐趋近于其健康伙伴体重结果的平均值。
第二项研究借鉴了结构与功能型社会支持、社会支持缓冲模型以及自我调节损耗领域的相关研究成果。通过结构方程模型（SEM, Structural Equation Model）分析，本研究发现，个体自我中心网络的规模与三元闭合特性均能显著预测其在FatSecret平台上感知到的体重管理相关社会支持。感知到的社会支持可正向促进更频繁的自我追踪行为，而自我追踪行为则能改善健康进展。事后分析结果显示，健康进展对自我追踪行为存在显著的负向影响：早期阶段的健康改善会降低后期阶段的自我追踪行为频次。
最后，本论文探讨了通过优化点对点社交网络的组织与构建，设计高效的线上体重管理干预方案的实践启示。