Improving Prediction and Causal Inference with Graphical Methods and Models

September 1, 2003

Funding: The National Science Foundation

Summary: This project explores the utility of various graphical models and methods, in particular causal graph theory, social networks and random graphs for relational data, in improving prediction and causal inference in empirical political and social sciences. Graphical models are naturally suited for conceptualizing and representing relationships, and graphical methods provide promising tools for studying structural properties of political and social systems. The project also seeks to extend the methods to accommodate special features of social science data, such as functional complexity and rareness of certain events.