The Logic of Cause and Effect: Unifying
Counterfactual, Graphical and Structural Models
Judea Pearl, UCLA
Friday, April 18 at 12:30pm in Smith 120
A systematic handling of causality requires a
mathematical language in which causal relationships
receive symbolic representation, clearly distinct from
statistical associations.
Two such languages have been proposed in the past:
path analysis and structural equations models,
used extensively in economics and the social sciences,
and Lewis-Neyman-Rubin's counterfactual (or potential-response)
model, used sporadically in philosophy and statistics.
Each of these two languages emphasizes different
aspects of the causal inference process and each has
encountered conceptual difficulties and strong opposition;
path analysis and structural equations
because they have been greatly misused and inadequately
formalized and the counterfactual framework because
it has been only partially formalized and, more
significantly, because it rests on esoteric and
seemingly metaphysical relationships (among
counterfactual variables) that bear no apparent
connection to ordinary understanding of cause-effect
processes.
I will propose a formal model,
based on DYNAMIC structural equations,
that unifies the two languages
above, explicates their conceptual and mathematical
bases and resolves their technical difficulties.
A simple rule enables us to translate a problem back
and forth, between the structural and counterfactual
representations, and choose the one appropriate
for analysis.
References for Judea Pearl talks
J. Pearl
"On the Foundations
of Structural Equation Models."
Technical Report R-244-s, UCLA, Computer Science Dpt
November 1996.
Pearl, J., ``On
the Testability of Causal Models
with Latent and Instrumental Variables,'' (with
appendix: Graphs, Structural Eqautions and Counterfactuals)
In P. Besnard and S. Hanks (Eds.), Uncertainty in
Artificial Intelligence 11\fR, Morgan Kaufmann,
San Francisco, CA, 435--443, 1995.
Pearl, J., ``On the Identification of Nonparametric
Structural Models,'' Technical Report (R-207),
Revision IV, November 1995.
To appear in Lecture Notes Series:
Latent Variable Modelling with Application
to Causality, Springer-Verlag.
J. Pearl, "Causal
diagrams for empirical research"
(with discussion), Biometrika 82(4),
Dec. 1995, pp. 669-710.
Other relevant papers can be found
here.