Social Space Project

This project titled Striding Towards Multidimensional Place and Social Space; Advancing Statistical Modeling Approaches for Social and Spatial Spillover in Latent Variable Models is a Swiss Science Foundation grant funded project (grant number 216293).

This project stems from work I began during my doctoral studies and subsequently published:
Roman, Z. J., & Brandt, H. (2021). A latent auto-regressive approach for Bayesian structural equation modeling of spatially or socially dependent data. Multivariate behavioral research58(1), 90-114.

Generally speaking, the goal of this project is to increase the flexibility of latent variable models to accommodate non-random (dependent) samples. This concept results in more flexibility in research designs for applied researchers of many disciplines. In the process of statistically controlling for dependence we are able to gain unique insights in the source of dependence. Common sources of dependence which can be modeled are spatial and social dependence. Spatial dependence occurs when cases (e.g., people, regions, schools, etc.)

This project aims to make spatial/social network SEM more accessible to researchers by further developing the framework established in Roman and Brandt (2021) and making the tools accessible to researchers. Multi-group and longitudinal adaptations are proposed to maximize the empirical applications of the models.

The models will be developed, evaluated with simulation studies, and typified with empirical examples
utilizing existing spatial data and scraped social media data. The project culminates in the development of open source (R package) implementations of the models in an effort to increase the accessibility of the approach to applied researchers and to improve the scientific impact of the project.