Treatment effect estimation with observational network data using machine learning

Causal inference methods for treatment effect estimation usually assume independent units. However, this assumption is often questionable because units may interact, resulting in spillover effects between them. We develop augmented inverse probability weighting (AIPW) for estimation and inference of...

Full description

Saved in:
Bibliographic Details
Main Authors: Emmenegger Corinne, Spohn Meta-Lina, Elmer Timon, Bühlmann Peter
Format: Article
Language:English
Published: De Gruyter 2025-04-01
Series:Journal of Causal Inference
Subjects:
Online Access:https://doi.org/10.1515/jci-2023-0082
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850265850535215104
author Emmenegger Corinne
Spohn Meta-Lina
Elmer Timon
Bühlmann Peter
author_facet Emmenegger Corinne
Spohn Meta-Lina
Elmer Timon
Bühlmann Peter
author_sort Emmenegger Corinne
collection DOAJ
description Causal inference methods for treatment effect estimation usually assume independent units. However, this assumption is often questionable because units may interact, resulting in spillover effects between them. We develop augmented inverse probability weighting (AIPW) for estimation and inference of the expected average treatment effect (EATE) with observational data from a single (social) network with spillover effects. In contrast to overall effects such as the global average treatment effect, the EATE measures, in expectation and on average over all units, how the outcome of a unit is causally affected by its own treatment, marginalizing over the spillover effects from other units. We develop cross-fitting theory with plugin machine learning to obtain a semiparametric treatment effect estimator that converges at the parametric rate and asymptotically follows a Gaussian distribution. The asymptotics are developed using the dependency graph rather than the network graph, which makes explicit that we allow for spillover effects beyond immediate neighbors in the network. We apply our AIPW method to the Swiss StudentLife Study data to investigate the effect of hours spent studying on exam performance accounting for the students’ social network.
format Article
id doaj-art-25ec5c8be1584caaa3b836f9775fe7e3
institution OA Journals
issn 2193-3685
language English
publishDate 2025-04-01
publisher De Gruyter
record_format Article
series Journal of Causal Inference
spelling doaj-art-25ec5c8be1584caaa3b836f9775fe7e32025-08-20T01:54:19ZengDe GruyterJournal of Causal Inference2193-36852025-04-01131591310.1515/jci-2023-0082Treatment effect estimation with observational network data using machine learningEmmenegger Corinne0Spohn Meta-Lina1Elmer Timon2Bühlmann Peter3Seminar for Statistics, ETH Zurich, Zurich, SwitzerlandSeminar for Statistics, ETH Zurich, Zurich, SwitzerlandDepartment of Humanities, Social and Political Sciences, ETH Zurich, Zurich, SwitzerlandSeminar for Statistics, ETH Zurich, Zurich, SwitzerlandCausal inference methods for treatment effect estimation usually assume independent units. However, this assumption is often questionable because units may interact, resulting in spillover effects between them. We develop augmented inverse probability weighting (AIPW) for estimation and inference of the expected average treatment effect (EATE) with observational data from a single (social) network with spillover effects. In contrast to overall effects such as the global average treatment effect, the EATE measures, in expectation and on average over all units, how the outcome of a unit is causally affected by its own treatment, marginalizing over the spillover effects from other units. We develop cross-fitting theory with plugin machine learning to obtain a semiparametric treatment effect estimator that converges at the parametric rate and asymptotically follows a Gaussian distribution. The asymptotics are developed using the dependency graph rather than the network graph, which makes explicit that we allow for spillover effects beyond immediate neighbors in the network. We apply our AIPW method to the Swiss StudentLife Study data to investigate the effect of hours spent studying on exam performance accounting for the students’ social network.https://doi.org/10.1515/jci-2023-0082dependent datainterferenceobserved confoundingsemiparametric inferencespillover effects62d2062g20
spellingShingle Emmenegger Corinne
Spohn Meta-Lina
Elmer Timon
Bühlmann Peter
Treatment effect estimation with observational network data using machine learning
Journal of Causal Inference
dependent data
interference
observed confounding
semiparametric inference
spillover effects
62d20
62g20
title Treatment effect estimation with observational network data using machine learning
title_full Treatment effect estimation with observational network data using machine learning
title_fullStr Treatment effect estimation with observational network data using machine learning
title_full_unstemmed Treatment effect estimation with observational network data using machine learning
title_short Treatment effect estimation with observational network data using machine learning
title_sort treatment effect estimation with observational network data using machine learning
topic dependent data
interference
observed confounding
semiparametric inference
spillover effects
62d20
62g20
url https://doi.org/10.1515/jci-2023-0082
work_keys_str_mv AT emmeneggercorinne treatmenteffectestimationwithobservationalnetworkdatausingmachinelearning
AT spohnmetalina treatmenteffectestimationwithobservationalnetworkdatausingmachinelearning
AT elmertimon treatmenteffectestimationwithobservationalnetworkdatausingmachinelearning
AT buhlmannpeter treatmenteffectestimationwithobservationalnetworkdatausingmachinelearning