Designing Transportable Experiments Under S-admissability
My Phan, David Arbour, Drew Dimmery, Anup Rao
Abstract
We consider the problem of designing a randomized experiment on a source population to estimate the Average Treatment Effect (ATE) on a target population. We propose a novel approach which explicitly considers the target when designing the experiment on the source. Under the covariate shift assumption, we design an unbiased importance-weighted estimator for the target population’s ATE. To reduce the variance of our estimator, we design a covariate balance condition (Target Balance) between the treatment and control groups based on the target population. We show that Target Balance achieves a higher variance reduction asymptotically than methods that do not consider the target population during the design phase. Our experiments illustrate that Target Balance reduces the variance even for small sample sizes.
BibTeX
@InProceedings{pmlr-v130-phan21a,
title = { Designing Transportable Experiments Under S-admissability },
author = {Phan, My and Arbour, David and Dimmery, Drew and Rao, Anup},
booktitle = {Proceedings of The 24th International Conference on Artificial Intelligence and Statistics},
pages = {2539--2547},
year = {2021},
editor = {Banerjee, Arindam and Fukumizu, Kenji},
volume = {130},
series = {Proceedings of Machine Learning Research},
month = {13--15 Apr},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v130/phan21a/phan21a.pdf},
url = {https://proceedings.mlr.press/v130/phan21a.html},
abstract = { We consider the problem of designing a randomized experiment on a source population to estimate the Average Treatment Effect (ATE) on a target population. We propose a novel approach which explicitly considers the target when designing the experiment on the source. Under the covariate shift assumption, we design an unbiased importance-weighted estimator for the target population’s ATE. To reduce the variance of our estimator, we design a covariate balance condition (Target Balance) between the treatment and control groups based on the target population. We show that Target Balance achieves a higher variance reduction asymptotically than methods that do not consider the target population during the design phase. Our experiments illustrate that Target Balance reduces the variance even for small sample sizes. }
}