Hierarchical Network Models: Mediation and Influence

The purpose of the project is to lay the foundation for the development of hierarchical network models that feature social networks as the mediator between an intervention and an outcome. The social network within each school provides insight into the mechanisms that affect individual outcomes and acts as a powerful mediating variable between the intervention and outcome, especially in large-scale interventions. Social networks are particularly informative for studies whose aim is to change the professional social structure of schools, whether it is an increase in teacher collaboration, a push toward small learning communities, or a change in other resource sharing relationships. Methods exist for estimating the effects of an intervention on a social network, and methods exist for estimating the influence of a social network on an outcome, but methods have not been developed for modeling social networks as mediators.

The research is based on results from a previous IES-funded grant (Hierarchical Network Models in Education Research). The team will use an extensive set of simulation studies to compare a two-step estimation procedure to full Markov Chain Monte Carlo estimation and to investigate parameter recovery of various effect size magnitudes, goodness-of-fit measures, the impact of model misspecification, the impact of missing data, and the impact of network size and density on parameter recovery. The team will also use real data to demonstrate the models, once they are developed. The results will be disseminated through trainings, software release, and peer-reviewed conference presentations and journal manuscripts.