Speaker
Details
The Princeton R Group (“Princeton tigeRs”) hosts informal monthly lunch events where we hear from scholars about interesting applications of their work in R.
Invited Speaker: Ben Rosche, Sociology
Modeling Aggregation Processes with Multilevel Regression: Applications in Spatial and Network Analysis
This workshop starts with an introduction to multilevel regression and then introduces the R package 'rmm' for modeling multilevel problems where higher-level outcomes (e.g., output of a university) are influenced by lower-level units (e.g., university departments). Building on Snijders (2016), the proposed model incorporates a parametrizable aggregation function into the regression equation to model such micro-to-macro links. The model supports various aggregation functions (e.g., min, max, mean, sum) as it employs Bayesian estimation, which eliminates the need for analytically tractable likelihood functions. The method demonstrates that aggregation functions can be empirically derived when their specific forms are a priori unknown. For example, in network and spatial analysis, weight matrices define the relative influence of neighboring on focal units. The model and accompanying R package enable researchers to estimate these weights instead of specifying fixed values.
Lunch will be provided. Register today!
Directions to JRR, Room A97:
If entering JRR from Washington Road or entering by the Simpson Atrium Cafe, either walk down one level or take the elevator down to Level A. Map of JRR & Louis Simpson.