Primary Submission Category: Proximal Causal Inference
Proximal Inference for Hidden Outcomes
Authors: Helen Guo, Ilya Shpitser, Elizabeth Ogburn,
Presenting Author: Helen Guo*
Nonparametric proxy variable methods have emerged as a powerful approach for addressing hidden variables in causal inference. One influential line of this work recovers latent factors which compose the target functional via eigendecomposition tasks. Within this framework, we identify the full law under discrete hidden outcomes and develop influence-function–based estimators for causal effects, resulting in Neyman-orthogonal estimation with desirable efficiency properties. Although methods for handling missing and mismeasured outcomes exist, this work provides what we believe to be the first proximal inference results for hidden outcomes. We illustrate our estimation approach through simulation studies.
Â