The Graphical Identification for Total Effects by using Surrogate Variables
arXiv:1207.1392
Abstract
Consider the case where cause-effect relationships between variables can be described as a directed acyclic graph and the corresponding linear structural equation model. This paper provides graphical identifiability criteria for total effects by using surrogate variables in the case where it is difficult to observe a treatment/response variable. The results enable us to judge from graph structure whether a total effect can be identified through the observation of surrogate variables.
Appears in Proceedings of the Twenty-First Conference on Uncertainty in Artificial Intelligence (UAI2005)