Comparing Social Network Influence on Alcohol Use Using Affiliation Data

6.项目摘要/摘要 提出的研究目标是扩展对友谊关系的单模网络数据分析 (按参与者的网络)到双模式(按事件的参与者从属关系/双方)网络数据，以对 共同参加校本组织活动对青少年饮酒的影响。第一个目标是 通过直接建模来确定基于从属关系的网络影响力的新概念的效用 (使用多个网络自相关模型)双模/从属关系暴露对饮酒的影响 控制其他竞争同龄人的行为主要是基于亲密朋友或同龄人的影响 占据相似的网络位置。第二个目标是确定多项指标的相对贡献 酒精使用的网络效应的形式，以及这些网络效应是如何作为风险因素共同作用的。这个 R00阶段的研究将集中在双模网络的结构特性和动力学 与酒精有关的从属网络的结构分析。与此相关的第三个目标是估计 具有几何加权的双模指数随机图模型(ERGM)的新规范 测试共生关系对青少年饮酒影响的技术，并将扩展到纵向 通过随机建模单模和双模网络的协同演化来检验 与饮酒有关的友谊和联系网络进化中的交叉依赖 基于参与者的建模框架。第四个目标是探索推动联合动力的社会机制。 与饮酒有关的以单模朋友网络为条件的从属网络结构。 解决这些问题将有助于确立PI的长期职业目标，即确立自己作为一名 双模网络专家，这与主要导师的不同(托马斯·瓦伦特) 以同伴对物质使用的影响研究中单模网络分析的主要方法 关于青少年饮酒的问题。对K99和R00两个阶段使用的数据的研究 青少年健康纵向研究(ADD Health)和南加州大学收集的数据。不仅是因为 拟议的研究有可能为以学校为基础的酒精预防计划提供洞察 基于社会影响力模型的实施，提出了运用两种模式的方法 网络数据不需要收集网络数据(社会计量)。因此，这种方法论具有 可应用于各种研究并用于公共卫生各个领域的潜力， 传统上一直不能使用网络方法。

6. Project Summary/Abstract The goal of the proposed research is to extend one-mode network data analysis of friendship relations (actor-by-actor network) to two-mode (actor-by-event affiliation/bipartite) network data to model the influence of co-participation in school-based organized activities on adolescent alcohol use. The first aim is to determine the utility of a new conceptualization of affiliation-based network influence by directly modeling (using multiple Network Autocorrelation Models) the effect of two-mode/affiliation exposure on drinking behavior controlling for other competing peer influences primarily those based on close friends or peers occupying similar network position. The second aim is to determine the relative contributions of multiple forms of network effects on alcohol use, and how these network effects operate together as risk factors. The research during the R00 phase will focus on structural properties and the dynamics of two-mode network structural analysis of affiliation networks in relation to alcohol. Related to this, the third aim is to estimate the new specification of two-mode Exponential Random Graph Model (ERGM) with geometric weighting technique to test the effects of co-affiliation on adolescent alcohol use, and it will be extended to longitudinal analysis by stochastically modeling the co-evolution of one-mode and two-mode networks to examine the cross-dependencies in the evolution of friendship and affiliation networks in relation to alcohol use within an actor-based modeling framework. The fourth aim is to explore social mechanisms that drive joint dynamics of affiliation network structures conditioned on one-mode friends' networks in relation to alcohol use. Addressing these questions would help establish the PI's long-term career goal of establishing herself as a specialist in two-mode network, which is separate and distinct from the primary mentor's (Thomas Valente) main approach of analyzing one-mode networks in the study of peer influence on substance use centered on adolescent alcohol use. Research for both K99 and R00 phase use data from both the National Longitudinal Study of Adolescent Health (Add Health) and from USC-collected data. Not only does the proposed study have the potential to provide insight into how school-based alcohol prevention program based on social influence model should be implemented, the proposed approach of using two-mode network data does not require the collection of network data (sociometrics). Therefore this methodology has the potential to be applied to a wide variety of studies and used in various fields of public health that traditionally have not been able to use network methods.