Causal Inference on Networks under Misspecified Exposure Mappings: A Partial Identification Framework

Estimating treatment effects in networks is challenging, as each potential outcome depends on the treatments of all other nodes in the network. To overcome this difficulty, existing methods typically impose an exposure mapping that compresses the treatment assignments in the network into a low-dimensional summary. However, if this mapping is misspecified, standard estimators for direct and spillover effects can be severely biased. We propose a novel partial identification framework for causal inference on networks to assess the robustness of treatment effects under misspecifications of the exposure mapping. Specifically, we derive sharp upper and lower bounds on direct and spillover effects under such misspecifications. As such, our framework presents a novel application of causal sensitivity analysis to exposure mappings. We instantiate our framework for three canonical exposure settings widely used in practice: (i) weighted means of the neighborhood treatments, (ii) threshold-based exposure mappings, and (iii) truncated neighborhood interference in the presence of higher-order spillovers. Furthermore, we develop orthogonal estimators for these bounds and prove that the resulting bound estimates are valid, sharp, and efficient. Our experiments show the bounds remain informative and provide reliable conclusions under misspecification of exposure mappings.

💡 Research Summary

The paper tackles a fundamental challenge in causal inference on networks: each unit’s outcome can depend on the treatment assignments of all other units, violating the classic Stable Unit Treatment Value Assumption (SUTVA). To make the problem tractable, the literature typically introduces an exposure mapping g that compresses the high‑dimensional vector of neighbor treatments into a low‑dimensional summary z (e.g., the proportion of treated neighbors). However, the exposure mapping must be specified a priori, and in many real‑world settings the true mechanism g* is only partially understood. Misspecifying g leads to biased point estimates of direct and spillover effects, and the standard identification assumptions (network interference and unconfoundedness) no longer hold.

The authors propose a partial‑identification framework that does not require the researcher to know the exact exposure mapping. Instead, they model the discrepancy between the assumed mapping g and the true but unknown mapping g* through a propensity‑ratio bound:

\