A0752
Title: Adaptive estimators for causal effects under network interference
Authors: Fei Fang - Duke University (United States) [presenting]
Alexandre Belloni - Duke University (United States)
Alexander Volfovsky - Duke University (United States)
Abstract: The estimation of causal effects is increasingly relevant in different applied fields. We consider a causal inference problem in the presence of interference. The focus is on observational studies where interference across units is governed by a known network interference. However, the radius (and intensity) of interference is unknown and can be dependent on the observed treatment assignments in the relevant subnetwork. We study causal estimators for the average direct treatment effect given the network interference. The proposed estimators build upon a Lepski-like procedure that searches over the possible relevant radius/assignment patterns. In the process, we also obtain estimators for the radius of the interference that can be dependent on the treatment assignment of neighbors. Thus it creates an adaptive estimation of the network interference structure. We establish oracle inequalities and corresponding adaptive rates for the direct treatment effect estimators. The adaptive network interference can be defined over the labelled subgraphs themselves or on features of these, which recover many assumptions previously used in the literature. We present theoretical examples and numerical simulations that illustrate the performance of the proposed estimators.
