Estimating Causal Effects under Image Confounding, Applied to Aid Evaluation

• In Event: Methodological Advances in Causal Inference

Abstract

Observational studies require adjustment for confounding factors that are correlated with both the treatment and outcome. In the setting where the observed variables are tabular quantities such as average age or income in a neighborhood, many tools have been developed for addressing such confounding. However, in many parts of the developing world, features about local communities may be scarce. In this context, map and satellite imagery can play an important role, serving as a proxy for the confounding variables otherwise unobserved. In this paper, we study confounder adjustment in this non-tabular setting, where patterns or objects found in images (such as maps or satellite imagery) contribute to the confounder bias. Using the evaluation of anti-poverty aid programs in Africa as our running example, we formalize the challenge of performing causal adjustment with such unstructured data---what conditions are sufficient to identify and estimate causal effects, how to perform estimation, and how to quantify the ways in which certain aspects of the unstructured image object are most predictive of the treatment decision. In Monte Carlo experiments, we analyze how causal inference approaches such as Inverse Propensity Weighting (IPW) are sensitive to misspecification of the confounder-generating image pattern mechanism. We also quantify the role of resolution, which controls the kinds of confounding objects that can be reconstructed. Finally, we use our methodology to estimate the effects of the policy interventions on poverty in African communities from satellite imagery.

Author

• Connor T Jerzak, University of Texas at Austin