Surrogate Outcomes and Transportability
share
Identification of causal effects is one of the most fundamental tasks of causal inference. We study a variant of the identifiability problem where the experimental distribution of interest is partially known. This corresponds to a real-world setting where experiments were conducted on a set of variables, which we call surrogate outcomes, but the variables of interest were not 10 measured. We label this problem as surrogate outcome identifiability and show that the concept of transportability provides a sufficient criteria for determining surrogate outcome identifiability for a large class of problems.
READ FULL TEXT