Causal Inference Under Approximate Neighborhood Interference

Leung, Michael P.

doi:10.3982/ecta17841

Cited by 29 publications

(54 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is analogous to familiar notions of spatial or temporal weak dependence, except the distance between observations is measured using path distance. Their paper and Leung (2022) verify ψ ‐dependence for a variety of network models used in practice.…”

Section: Asymptotic Theorymentioning

confidence: 88%

Network Cluster‐Robust Inference

Leung

2023

ECTA

Self Cite

View full text Add to dashboard Cite

Since network data commonly consists of observations from a single large network, researchers often partition the network into clusters in order to apply cluster‐robust inference methods. Existing such methods require clusters to be asymptotically independent. Under mild conditions, we prove that, for this requirement to hold for network‐dependent data, it is necessary and sufficient that clusters have low conductance, the ratio of edge boundary size to volume. This yields a simple measure of cluster quality. We find in simulations that when clusters have low conductance, cluster‐robust methods control size better than HAC estimators. However, for important classes of networks lacking low‐conductance clusters, the former can exhibit substantial size distortion. To determine the number of low‐conductance clusters and construct them, we draw on results in spectral graph theory that connect conductance to the spectrum of the graph Laplacian. Based on these results, we propose to use the spectrum to determine the number of low‐conductance clusters and spectral clustering to construct them.

show abstract

Section: Asymptotic Theorymentioning

confidence: 88%

Network Cluster‐Robust Inference

Leung

2023

ECTA

Self Cite

View full text Add to dashboard Cite

show abstract

“…As for how to choose probes that don’t interfere with each other, we can treat it as an experimental design choice. For instance, non-interference tends to be satisfied with higher probability if the probes are performed in pairs far away in the graph [50]. Finally, we refer the reader to [18] for a thorough analysis of experimental design choices that result in non-interfering probes.…”

Section: Interventional Lifting For Link Predictionmentioning

confidence: 99%

Causal lifting and link prediction

Cotta,

Bevilacqua,

Ahmed

et al. 2023

Proc. R. Soc. A.

View full text Add to dashboard Cite

Existing causal models for link prediction assume an underlying set of inherent node factors—an innate characteristic defined at the node’s birth—that governs the causal evolution of links in the graph. In some causal tasks, however, link formation is path-dependent : the outcome of link interventions depends on existing links. Unfortunately, these existing causal methods are not designed for path-dependent link formation, as the cascading functional dependencies between links (arising from path dependence ) are either unidentifiable or require an impractical number of control variables. To overcome this, we develop the first causal model capable of dealing with path dependencies in link prediction. In this work, we introduce the concept of causal lifting, an invariance in causal models of independent interest that, on graphs, allows the identification of causal link prediction queries using limited interventional data. Further, we show how structural pairwise embeddings exhibit lower bias and correctly represent the task’s causal structure, as opposed to existing node embeddings, e.g. graph neural network node embeddings and matrix factorization. Finally, we validate our theoretical findings on three scenarios for causal link prediction tasks: knowledge base completion, covariance matrix estimation and consumer-product recommendations.

show abstract

“…This is also the focus of many previous studies(Eckles et al 2016, Chin 2019, Cortez et al 2022a Leung (2022),Belloni et al (2022). have explored the choice of optimal n, which is beyond the scope of our study.…”

mentioning

confidence: 94%

“…A basic example of exposure mapping is the fractional q neighborhood exposure, which categorizes based on whether the proportion of treated neighbors exceeds a specific threshold (Ugander et al 2013, Eckles et al 2016. There is an expanding body of literature that extends the conventional causal inference framework to accommodate network interference (Bowers et al 2013, Eckles et al 2016, Athey et al 2018, Basse et al 2019, Forastiere et al 2021, Leung 2022, Hu et al 2022, Yu et al 2022). We will further discuss and contrast these various approaches to better understand network interference with our own work later in this paper.…”

Section: Introductionmentioning

confidence: 99%