Statistical Models for Three-Way Social Network Dynamics

点击翻译按钮获取中文摘要

This research project will develop models for the dynamics of social networks in which relationships involve three social entities. Conventional models for social network data focus on two-party (sender-receiver) relationships. Complementing these models with an extra dimension is critical to understanding social phenomena ranging from network perception biases (perceiver-sender-receiver relations) to gossip (sender-receiver-target relations). The methods to be developed will be applicable to many substantive fields, including education, political science, psychology, and sociology. A large-scale study on the dynamics of gossip among children and adolescents and its consequences on health and performance outcomes will constitute one of the major applications of this project. Through established collaborations, the project will directly influence policy and help to construct successful prevention strategies for social exclusion. The development of publicly available software and the creation of training materials and cross-disciplinary workshops will ensure broad access to the new models. Graduate students in the social sciences and statistics will be trained and mentored as part of the research process. The results of this research will be incorporated into the undergraduate course setting.The models for the dynamics of three-way network data will be formulated in the stochastic actor-oriented modeling framework. These continuous-time models will enable researchers to study individuals' network decisions and the micro-level social mechanisms governing network change. Project results will include estimation procedures for both time-stamped relational event data and for longitudinal network panel data. To account for individual heterogeneity, a random effects model for the study of three-way network dynamics will be developed. Using the newly developed methods, the project will explore both substantive and methodological questions of interest. For example, the assumption that individuals base their relational behavior on a shared cognition of a 'true' network is standard in social network models. At the same time, it is questionable since people's network perceptions are known to differ. Simulation studies using the new methods will evaluate the impact of this assumption.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.