A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates
A powerful tool for the analysis of nonrandomized observational studies has been the potential outcomes model. Utilization of this framework allows analysts to estimate average treatment effects. This article considers the situation in which high-dimensional covariates are present and revisits the s...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
De Gruyter
2019-09-01
|
| Series: | Journal of Causal Inference |
| Subjects: | |
| Online Access: | https://doi.org/10.1515/jci-2018-0024 |
| _version_ | 1818650900522074112 |
|---|---|
| author | Ghosh Debashis Cruz Cortés Efrén |
| author_facet | Ghosh Debashis Cruz Cortés Efrén |
| author_sort | Ghosh Debashis |
| collection | DOAJ |
| description | A powerful tool for the analysis of nonrandomized observational studies has been the potential outcomes model. Utilization of this framework allows analysts to estimate average treatment effects. This article considers the situation in which high-dimensional covariates are present and revisits the standard assumptions made in causal inference. We show that by employing a flexible Gaussian process framework, the assumption of strict overlap leads to very restrictive assumptions about the distribution of covariates, results for which can be characterized using classical results from Gaussian random measures as well as reproducing kernel Hilbert space theory. In addition, we propose a strategy for data-adaptive causal effect estimation that does not rely on the strict overlap assumption. These findings reveal under a focused framework the stringency that accompanies the use of the treatment positivity assumption in high-dimensional settings. |
| first_indexed | 2024-12-17T01:57:34Z |
| format | Article |
| id | doaj.art-599ba7e0f70b442885c0227440466692 |
| institution | Directory Open Access Journal |
| issn | 2193-3677 2193-3685 |
| language | English |
| last_indexed | 2024-12-17T01:57:34Z |
| publishDate | 2019-09-01 |
| publisher | De Gruyter |
| record_format | Article |
| series | Journal of Causal Inference |
| spelling | doaj.art-599ba7e0f70b442885c02274404666922022-12-21T22:07:55ZengDe GruyterJournal of Causal Inference2193-36772193-36852019-09-017273536010.1515/jci-2018-0024A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional CovariatesGhosh Debashis0Cruz Cortés Efrén1Department of Biostatistics and Informatics, 144805Colorado School of Public Health, 80045Aurora, CO, United StatesDepartment of Biostatistics and Informatics, 144805Colorado School of Public Health, 80045Aurora, CO, United StatesA powerful tool for the analysis of nonrandomized observational studies has been the potential outcomes model. Utilization of this framework allows analysts to estimate average treatment effects. This article considers the situation in which high-dimensional covariates are present and revisits the standard assumptions made in causal inference. We show that by employing a flexible Gaussian process framework, the assumption of strict overlap leads to very restrictive assumptions about the distribution of covariates, results for which can be characterized using classical results from Gaussian random measures as well as reproducing kernel Hilbert space theory. In addition, we propose a strategy for data-adaptive causal effect estimation that does not rely on the strict overlap assumption. These findings reveal under a focused framework the stringency that accompanies the use of the treatment positivity assumption in high-dimensional settings.https://doi.org/10.1515/jci-2018-0024average causal effectcovariate balancefunctional datamachine learningpositivity |
| spellingShingle | Ghosh Debashis Cruz Cortés Efrén A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates Journal of Causal Inference average causal effect covariate balance functional data machine learning positivity |
| title | A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates |
| title_full | A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates |
| title_fullStr | A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates |
| title_full_unstemmed | A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates |
| title_short | A Gaussian Process Framework for Overlap and Causal Effect Estimation with High-Dimensional Covariates |
| title_sort | gaussian process framework for overlap and causal effect estimation with high dimensional covariates |
| topic | average causal effect covariate balance functional data machine learning positivity |
| url | https://doi.org/10.1515/jci-2018-0024 |
| work_keys_str_mv | AT ghoshdebashis agaussianprocessframeworkforoverlapandcausaleffectestimationwithhighdimensionalcovariates AT cruzcortesefren agaussianprocessframeworkforoverlapandcausaleffectestimationwithhighdimensionalcovariates AT ghoshdebashis gaussianprocessframeworkforoverlapandcausaleffectestimationwithhighdimensionalcovariates AT cruzcortesefren gaussianprocessframeworkforoverlapandcausaleffectestimationwithhighdimensionalcovariates |