University of Massachusetts Amherst

Search Google Appliance

Links

Causal Inference with Graphical Models Workshop

ISSR Summer Methodology Workshop
June 26 to June 27
UMass Amherst, 107 Bartlett Hall

 

Causal Inference with Graphic Models Workshop


Description: Inferring causality is central to many quantitative studies in social science.  A large number of analytical methods have been developed to infer causal dependence from observational data, including propensity score matching, instrumental variable designs, interrupted time-series designs, and many others.  Unfortunately, the assumptions and limitations of these methods can be difficult to explain and reason about.  

This tutorial introduces participants to causal graphical models, a powerful formalism developed within computer science and statistics that simultaneously provides: 1) a unifying formal framework for understanding and explaining specific methods for causal inference; 2) a practical tool for representing and reasoning about the implications of particular causal models; and 3) powerful algorithmic methods for learning complex causal models from data and reasoning about their implications.  This tutorial assumes only a basic understanding of probability and statistics and no knowledge of programming.  Participants familiar with experimental and quasi-experimental designs will gain a new understanding of the benefits and assumptions of these methods, and participants without that knowledge will learn about multiple methods for inferring causality within a single unifying framework.
 

Instructor: David Jensen is a Professor in the College of Information and Computer Sciences at the University of Massachusetts Amherst.  He serves as the Associate Director of the Computational Social Science Institute, an interdisciplinary effort at UMass to study social phenomena using computational tools and concepts.  His current research focuses on machine learning and data science for analyzing large social, technological, and computational systems.  In particular, his work focuses on methods for constructing accurate causal models from observational and experimental data, with applications to social science, fraud detection, security and systems management.  His research is supported by many organizations, including the National Science Foundation, Defense Advanced Research Projects Agency, and Intelligence Advanced Research Projects Activity.
 

Learn more and Register