🤖 AI Summary
This paper addresses the concurrent presence of three systematic biases in causal inference: interference (where an individual’s treatment affects others’ outcomes), unmeasured confounding, and lack of transportability across populations. We propose the first unified weighted sensitivity analysis framework that jointly quantifies the impact of all three biases on causal effect estimation. Our approach introduces interpretable sensitivity parameters and employs a weighting-based estimation strategy that accommodates unmeasured confounding while explicitly modeling interference structures and constraints on cross-population extrapolation. Empirical evaluations across multiple real-world settings demonstrate that the method robustly assesses bias magnitude and enhances the credibility of causal estimates. It provides an interpretable, scalable tool for causal inference in complex, dependent environments—such as social networks and public health interventions—where traditional assumptions of independence and identifiability fail.
📝 Abstract
In many applications of causal inference, the treatment received by one unit may influence the outcome of another, a phenomenon referred to as interference. Although there are several frameworks for conducting causal inference in the presence of interference, practitioners often lack the data necessary to adjust for its effects. In this paper, we propose a weighting-based sensitivity analysis framework that can be used to assess the systematic bias arising from ignoring interference. Unlike most of the existing literature, we allow for the presence of unmeasured confounding, and show that the combination of interference and unmeasured confounding is a notable challenge to causal inference. We also study a third factor contributing to systematic bias: lack of transportability. Our framework enables practitioners to assess the impact of these three issues simultaneously through several easily interpretable sensitivity parameters that can reflect a wide range of intuitions about the data.