Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning
A wide range of machine-learning-based approaches have been developed in the past decade, increasing our ability to accurately model nonlinear and nonadditive response surfaces. This has improved performance for inferential tasks such as estimating average treatment effects in situations where stand...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2022-12-01
|
Series: | Entropy |
Subjects: | |
Online Access: | https://www.mdpi.com/1099-4300/24/12/1782 |
_version_ | 1797459235442786304 |
---|---|
author | Vincent Dorie George Perrett Jennifer L. Hill Benjamin Goodrich |
author_facet | Vincent Dorie George Perrett Jennifer L. Hill Benjamin Goodrich |
author_sort | Vincent Dorie |
collection | DOAJ |
description | A wide range of machine-learning-based approaches have been developed in the past decade, increasing our ability to accurately model nonlinear and nonadditive response surfaces. This has improved performance for inferential tasks such as estimating average treatment effects in situations where standard parametric models may not fit the data well. These methods have also shown promise for the related task of identifying heterogeneous treatment effects. However, the estimation of both overall and heterogeneous treatment effects can be hampered when data are structured within groups if we fail to correctly model the dependence between observations. Most machine learning methods do not readily accommodate such structure. This paper introduces a new algorithm, stan4bart, that combines the flexibility of Bayesian Additive Regression Trees (BART) for fitting nonlinear response surfaces with the computational and statistical efficiencies of using Stan for the parametric components of the model. We demonstrate how stan4bart can be used to estimate average, subgroup, and individual-level treatment effects with stronger performance than other flexible approaches that ignore the multilevel structure of the data as well as multilevel approaches that have strict parametric forms. |
first_indexed | 2024-03-09T16:48:32Z |
format | Article |
id | doaj.art-1136357fb60e4960a89e7b0014b5afba |
institution | Directory Open Access Journal |
issn | 1099-4300 |
language | English |
last_indexed | 2024-03-09T16:48:32Z |
publishDate | 2022-12-01 |
publisher | MDPI AG |
record_format | Article |
series | Entropy |
spelling | doaj.art-1136357fb60e4960a89e7b0014b5afba2023-11-24T14:42:52ZengMDPI AGEntropy1099-43002022-12-012412178210.3390/e24121782Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine LearningVincent Dorie0George Perrett1Jennifer L. Hill2Benjamin Goodrich3Code for America, San Francisco, CA 94103, USADepartment of Applied Statistics, Social Science, and the Humanities, New York University, New York, NY 10003, USADepartment of Applied Statistics, Social Science, and the Humanities, New York University, New York, NY 10003, USADepartment of Political Science, Columbia University, New York, NY 10025, USAA wide range of machine-learning-based approaches have been developed in the past decade, increasing our ability to accurately model nonlinear and nonadditive response surfaces. This has improved performance for inferential tasks such as estimating average treatment effects in situations where standard parametric models may not fit the data well. These methods have also shown promise for the related task of identifying heterogeneous treatment effects. However, the estimation of both overall and heterogeneous treatment effects can be hampered when data are structured within groups if we fail to correctly model the dependence between observations. Most machine learning methods do not readily accommodate such structure. This paper introduces a new algorithm, stan4bart, that combines the flexibility of Bayesian Additive Regression Trees (BART) for fitting nonlinear response surfaces with the computational and statistical efficiencies of using Stan for the parametric components of the model. We demonstrate how stan4bart can be used to estimate average, subgroup, and individual-level treatment effects with stronger performance than other flexible approaches that ignore the multilevel structure of the data as well as multilevel approaches that have strict parametric forms.https://www.mdpi.com/1099-4300/24/12/1782BARTStancausal inferencemachine learningheterogeneous treatment effectsmultilevel data |
spellingShingle | Vincent Dorie George Perrett Jennifer L. Hill Benjamin Goodrich Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning Entropy BART Stan causal inference machine learning heterogeneous treatment effects multilevel data |
title | Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning |
title_full | Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning |
title_fullStr | Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning |
title_full_unstemmed | Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning |
title_short | Stan and BART for Causal Inference: Estimating Heterogeneous Treatment Effects Using the Power of Stan and the Flexibility of Machine Learning |
title_sort | stan and bart for causal inference estimating heterogeneous treatment effects using the power of stan and the flexibility of machine learning |
topic | BART Stan causal inference machine learning heterogeneous treatment effects multilevel data |
url | https://www.mdpi.com/1099-4300/24/12/1782 |
work_keys_str_mv | AT vincentdorie stanandbartforcausalinferenceestimatingheterogeneoustreatmenteffectsusingthepowerofstanandtheflexibilityofmachinelearning AT georgeperrett stanandbartforcausalinferenceestimatingheterogeneoustreatmenteffectsusingthepowerofstanandtheflexibilityofmachinelearning AT jenniferlhill stanandbartforcausalinferenceestimatingheterogeneoustreatmenteffectsusingthepowerofstanandtheflexibilityofmachinelearning AT benjamingoodrich stanandbartforcausalinferenceestimatingheterogeneoustreatmenteffectsusingthepowerofstanandtheflexibilityofmachinelearning |