The Graphical Identification for Total Effects by using Surrogate Variables

TL;DR

This paper develops graphical criteria to identify total causal effects using surrogate variables in linear structural equation models when direct observation of variables is challenging.

Contribution

It introduces new graphical identifiability criteria that determine when total effects can be inferred from surrogate variables in causal graphs.

Findings

Criteria for identifiability of total effects using surrogates

Applicable to models with unobservable variables

Guidelines for causal inference in complex systems

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.