May 27, 2017

Causal Interpretation and Identification of Conditional Independence Structures

September 27 - October 8, 1999


Organizing Committee:

Phil Dawid, University College London
Glenn Shafer, Rutgers University


Graphical models have been traditionally seen as ways of describing and manipulating probabilistic conditional independence. However, they are often given informal causal interpretations, and there now exist mathematical ways of making these precise and manipulating them.

This research seminar will aim to extend understanding of these methods, and to explore the scope of graphical modelling as a tool for causal inference and analysis. Attention will also focus on the nature of the relationship between causal interpretations and conditional independence properties of graphical representations, the two main themes of the overall program.

During the week, September 27 - October 1, 1999, fairly informal activities are planned.

During the second week, October 4 - 8, 1999, working discussions are planned on the following topics:

  • Monday, October 4, 1999
    Causal Thinking in Data Analysis. How might we put more and better causal thinking in a second course in statistics?
  • Tuesday, October 5, 1999
    Observation and Experiment. What are the different strategies for teasing causal information out of observational studies?
  • Wednesday, October 6, 1999
    What Use Counterfactuals?
  • Thursday, October 7, 1999
    Causal Logic. Causal information is sometimes too fragmentary to be organized as a causal model. What then?
  • Friday, October 8, 1999
    The Causal Interpretations of Graphical Models. Is there more than one?


M. Eerola Rolf Nevanlinna Institute J. Robins Harvard School of Public Health
M. Forster University of Wisconsin D. Rubin Harvard University
P. Holland University of California R. Scheines Carnegie Mellon University
S. Lauritzen Aalborg University R. Shachter Stanford University
W. Oldford University of Waterloo M. Studeny Academy of Science, Czech Republic
J. Pearl University of California L. Wasserman Carnegie Mellon University