Modeling Covariate Transition for Efficient Estimation of Longitudinal Treatment Effects in Randomized Experiments

Naoki Chihara, Tatsushi Oka, Yasuko Matsubara, Yasushi Sakurai, Shota Yasui
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:19546-19580, 2026.

Abstract

We present a regression-adjustment framework designed for the estimation of longitudinal treatment effects in randomized experiments under static regimes. While regression-adjustment methods are useful for variance reduction in randomized experiments by using pre-treatment covariates, they usually focus only on average effects, from which we cannot obtain valuable insights into when the effects appear and how long they continue. To address this issue, we consider intermediate outcomes and evolving post-treatment covariates over time, and we represent such dynamic trajectories using transition kernels. Furthermore, we establish the asymptotic normality and the semiparametric efficiency bound for our estimator, enabling more powerful statistical inference. Simulation studies and empirical analysis using A/B test data from a streaming platform in Japan show the practical advantages of our method.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chihara26a, title = {Modeling Covariate Transition for Efficient Estimation of Longitudinal Treatment Effects in Randomized Experiments}, author = {Chihara, Naoki and Oka, Tatsushi and Matsubara, Yasuko and Sakurai, Yasushi and Yasui, Shota}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {19546--19580}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chihara26a/chihara26a.pdf}, url = {https://proceedings.mlr.press/v306/chihara26a.html}, abstract = {We present a regression-adjustment framework designed for the estimation of longitudinal treatment effects in randomized experiments under static regimes. While regression-adjustment methods are useful for variance reduction in randomized experiments by using pre-treatment covariates, they usually focus only on average effects, from which we cannot obtain valuable insights into when the effects appear and how long they continue. To address this issue, we consider intermediate outcomes and evolving post-treatment covariates over time, and we represent such dynamic trajectories using transition kernels. Furthermore, we establish the asymptotic normality and the semiparametric efficiency bound for our estimator, enabling more powerful statistical inference. Simulation studies and empirical analysis using A/B test data from a streaming platform in Japan show the practical advantages of our method.} }
Endnote
%0 Conference Paper %T Modeling Covariate Transition for Efficient Estimation of Longitudinal Treatment Effects in Randomized Experiments %A Naoki Chihara %A Tatsushi Oka %A Yasuko Matsubara %A Yasushi Sakurai %A Shota Yasui %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chihara26a %I PMLR %P 19546--19580 %U https://proceedings.mlr.press/v306/chihara26a.html %V 306 %X We present a regression-adjustment framework designed for the estimation of longitudinal treatment effects in randomized experiments under static regimes. While regression-adjustment methods are useful for variance reduction in randomized experiments by using pre-treatment covariates, they usually focus only on average effects, from which we cannot obtain valuable insights into when the effects appear and how long they continue. To address this issue, we consider intermediate outcomes and evolving post-treatment covariates over time, and we represent such dynamic trajectories using transition kernels. Furthermore, we establish the asymptotic normality and the semiparametric efficiency bound for our estimator, enabling more powerful statistical inference. Simulation studies and empirical analysis using A/B test data from a streaming platform in Japan show the practical advantages of our method.
APA
Chihara, N., Oka, T., Matsubara, Y., Sakurai, Y. & Yasui, S.. (2026). Modeling Covariate Transition for Efficient Estimation of Longitudinal Treatment Effects in Randomized Experiments. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:19546-19580 Available from https://proceedings.mlr.press/v306/chihara26a.html.

Related Material