Predictive State Propensity Subclassification (PSPS): A causal inference algorithm for data-driven propensity score stratification

Joseph Kelly, Jing Kong, Georg M. Goerg
Proceedings of the First Conference on Causal Learning and Reasoning, PMLR 177:352-372, 2022.

Abstract

We introduce Predictive State Propensity Subclassification (PSPS), a novel learning algorithm for causal inference from observational data. PSPS combines propensity and outcome models into one encompassing probabilistic framework, which can be jointly estimated using maximum likelihood or Bayesian inference. The methodology applies to both discrete and continuous treatments and can estimate unit-level and population-level average treatment effects. We describe the neural network architecture and its TensorFlow implementation for likelihood optimization. Finally we demonstrate via large-scale simulations that PSPS outperforms state-of-the-art algorithms – both on bias for average treatment effects (ATEs) and RMSE for unit-level treatment effects (UTEs).

Cite this Paper

BibTeX
@InProceedings{pmlr-v177-kelly22a, title = {Predictive State Propensity Subclassification ({PSPS}): A causal inference algorithm for data-driven propensity score stratification}, author = {Kelly, Joseph and Kong, Jing and Goerg, Georg M.}, booktitle = {Proceedings of the First Conference on Causal Learning and Reasoning}, pages = {352--372}, year = {2022}, editor = {Schölkopf, Bernhard and Uhler, Caroline and Zhang, Kun}, volume = {177}, series = {Proceedings of Machine Learning Research}, month = {11--13 Apr}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v177/kelly22a/kelly22a.pdf}, url = {https://proceedings.mlr.press/v177/kelly22a.html}, abstract = {We introduce Predictive State Propensity Subclassification (PSPS), a novel learning algorithm for causal inference from observational data. PSPS combines propensity and outcome models into one encompassing probabilistic framework, which can be jointly estimated using maximum likelihood or Bayesian inference. The methodology applies to both discrete and continuous treatments and can estimate unit-level and population-level average treatment effects. We describe the neural network architecture and its TensorFlow implementation for likelihood optimization. Finally we demonstrate via large-scale simulations that PSPS outperforms state-of-the-art algorithms – both on bias for average treatment effects (ATEs) and RMSE for unit-level treatment effects (UTEs).} }
Endnote
%0 Conference Paper %T Predictive State Propensity Subclassification (PSPS): A causal inference algorithm for data-driven propensity score stratification %A Joseph Kelly %A Jing Kong %A Georg M. Goerg %B Proceedings of the First Conference on Causal Learning and Reasoning %C Proceedings of Machine Learning Research %D 2022 %E Bernhard Schölkopf %E Caroline Uhler %E Kun Zhang %F pmlr-v177-kelly22a %I PMLR %P 352--372 %U https://proceedings.mlr.press/v177/kelly22a.html %V 177 %X We introduce Predictive State Propensity Subclassification (PSPS), a novel learning algorithm for causal inference from observational data. PSPS combines propensity and outcome models into one encompassing probabilistic framework, which can be jointly estimated using maximum likelihood or Bayesian inference. The methodology applies to both discrete and continuous treatments and can estimate unit-level and population-level average treatment effects. We describe the neural network architecture and its TensorFlow implementation for likelihood optimization. Finally we demonstrate via large-scale simulations that PSPS outperforms state-of-the-art algorithms – both on bias for average treatment effects (ATEs) and RMSE for unit-level treatment effects (UTEs).
APA
Kelly, J., Kong, J. & Goerg, G.M.. (2022). Predictive State Propensity Subclassification (PSPS): A causal inference algorithm for data-driven propensity score stratification. Proceedings of the First Conference on Causal Learning and Reasoning, in Proceedings of Machine Learning Research 177:352-372 Available from https://proceedings.mlr.press/v177/kelly22a.html.

Related Material