Interpretable Almost Matching Exactly With Instrumental Variables

M. Usaid Awan, Yameng Liu, Marco Morucci, Sudeepa Roy, Cynthia Rudin, Alexander Volfovsky
Proceedings of The 35th Uncertainty in Artificial Intelligence Conference, PMLR 115:1116-1126, 2020.

Abstract

Uncertainty in the estimation of the causal effect in observational studies is often due to unmeasured confounding, i.e., the presence of unobserved covariates linking treatments and outcomes. Instrumental Variables (IV) are commonly used to reduce the effects of unmeasured confounding. Existing methods for IV estimation either require strong parametric assumptions, use arbitrary distance metrics, or do not scale well to large datasets. We propose a matching framework for IV in the presence of observed categorical confounders that addresses these weaknesses. Our method first matches units exactly, and then consecutively drops variables to approximately match the remaining units on as many variables as possible. We show that our algorithm constructs better matches than other existing methods on simulated datasets, and we produce interesting results in an application to political canvassing.

Cite this Paper

BibTeX
@InProceedings{pmlr-v115-awan20a, title = {Interpretable Almost Matching Exactly With Instrumental Variables}, author = {Awan, M. Usaid and Liu, Yameng and Morucci, Marco and Roy, Sudeepa and Rudin, Cynthia and Volfovsky, Alexander}, booktitle = {Proceedings of The 35th Uncertainty in Artificial Intelligence Conference}, pages = {1116--1126}, year = {2020}, editor = {Adams, Ryan P. and Gogate, Vibhav}, volume = {115}, series = {Proceedings of Machine Learning Research}, month = {22--25 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v115/awan20a/awan20a.pdf}, url = {https://proceedings.mlr.press/v115/awan20a.html}, abstract = {Uncertainty in the estimation of the causal effect in observational studies is often due to unmeasured confounding, i.e., the presence of unobserved covariates linking treatments and outcomes. Instrumental Variables (IV) are commonly used to reduce the effects of unmeasured confounding. Existing methods for IV estimation either require strong parametric assumptions, use arbitrary distance metrics, or do not scale well to large datasets. We propose a matching framework for IV in the presence of observed categorical confounders that addresses these weaknesses. Our method first matches units exactly, and then consecutively drops variables to approximately match the remaining units on as many variables as possible. We show that our algorithm constructs better matches than other existing methods on simulated datasets, and we produce interesting results in an application to political canvassing.} }
Endnote
%0 Conference Paper %T Interpretable Almost Matching Exactly With Instrumental Variables %A M. Usaid Awan %A Yameng Liu %A Marco Morucci %A Sudeepa Roy %A Cynthia Rudin %A Alexander Volfovsky %B Proceedings of The 35th Uncertainty in Artificial Intelligence Conference %C Proceedings of Machine Learning Research %D 2020 %E Ryan P. Adams %E Vibhav Gogate %F pmlr-v115-awan20a %I PMLR %P 1116--1126 %U https://proceedings.mlr.press/v115/awan20a.html %V 115 %X Uncertainty in the estimation of the causal effect in observational studies is often due to unmeasured confounding, i.e., the presence of unobserved covariates linking treatments and outcomes. Instrumental Variables (IV) are commonly used to reduce the effects of unmeasured confounding. Existing methods for IV estimation either require strong parametric assumptions, use arbitrary distance metrics, or do not scale well to large datasets. We propose a matching framework for IV in the presence of observed categorical confounders that addresses these weaknesses. Our method first matches units exactly, and then consecutively drops variables to approximately match the remaining units on as many variables as possible. We show that our algorithm constructs better matches than other existing methods on simulated datasets, and we produce interesting results in an application to political canvassing.
APA
Awan, M.U., Liu, Y., Morucci, M., Roy, S., Rudin, C. & Volfovsky, A.. (2020). Interpretable Almost Matching Exactly With Instrumental Variables. Proceedings of The 35th Uncertainty in Artificial Intelligence Conference, in Proceedings of Machine Learning Research 115:1116-1126 Available from https://proceedings.mlr.press/v115/awan20a.html.

Related Material