Generic Machine Learning Inference on Heterogeneous Treatment Effects in Randomized Experiments, with an Application to Immunization in India
Open accessThe DOI matches its Crossref record; 14 of the 14 references it cites were found in a registry or an index.
Abstract
We propose strategies to estimate and make inference on key features of heterogeneous effects in randomized experiments. These key features include best linear predictors of the effects using machine learning proxies, average effects sorted by impact groups, and average characteristics of most and least impacted units. The approach is valid in high dimensional settings, where the effects are proxied (but not necessarily consistently estimated) by predictive and causal machine learning methods. We post-process these proxies into estimates of the key features. Our approach is generic, it can be used in conjunction with penalized methods, neural networks, random forests, boosted trees, and ensemble methods, both predictive and causal. Estimation and inference are based on repeated data splitting to avoid overfitting and achieve validity. We use quantile aggregation of the results across many potential splits, in particular taking medians of p-values and medians and other quantiles of confidence intervals. We show that quantile aggregation lowers estimation risks over a single split procedure, and establish its principal inferential properties. Finally, our analysis reveals ways to build provably better machine learning proxies through causal learning: we can use the objective functions that we develop to construct the best linear predictors of the effects, to obtain better machine learning proxies in the initial step. We illustrate the use of both inferential tools and causal learners with a randomized field experiment that evaluates a combination of nudges to stimulate demand for immunization in India.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Estimating causal effects of treatments in randomized and nonrandomized studies
Rubin, D. B. (1974) · Journal of Educational Psychology · doi:10.1037/h0037350 · Cited by 9,436 · cited 6×
“Let Y (1) and Y (0) be the potential outcomes in the treatment state 1 and the non-treatment state 0 (see Neyman (1923), Rubin (1974)).”
InfluentialCausal Inference for Statistics, Social, and Biomedical Sciences: An Introduction
Imbens, G. W., & Rubin, D. B. (2015) · doi:10.1017/cbo9781139025751 · Cited by 4,210
Estimation and Inference of Heterogeneous Treatment Effects using Random Forests
Wager, S., & Athey, S. (2018) · Journal of the American Statistical Association · doi:10.1080/01621459.2017.1319839 · Cited by 2,917
Semiparametric Difference-in-Differences Estimators
Abadie, A. (2005) · The Review of Economic Studies · doi:10.1111/0034-6527.00321 · Cited by 2,286
Recursive partitioning for heterogeneous causal effects
Athey, S., & Imbens, G. (2016) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.1510489113 · Cited by 1,586
Algorithmic Learning in a Random World
(2005) · doi:10.1007/b106715 · Cited by 1,119
Banerjee, A. V., Duflo, E., Glennerster, R., & Kothari, D. (2010) · BMJ · doi:10.1136/bmj.c2220 · Cited by 508
Self-citationOn regression adjustments to experimental data
Freedman, D. A. (2008) · Advances in Applied Mathematics · doi:10.1016/j.aam.2006.12.003 · Cited by 417 · cited 2×
“Covariates are not necessary and even linear regression is not the best way to analyze the data (Freedman, 2008; Imbens and Rubin, 2015).”
Algorithmic Learning in a Random World
Vovk, V., Gammerman, A., & Shafer, G. (2022) · doi:10.1007/978-3-031-06649-8 · Cited by 296
Who Should Be Treated? Empirical Welfare Maximization Methods for Treatment Choice
Kitagawa, T., & Tetenov, A. (2018) · Econometrica · doi:10.3982/ecta13288 · Cited by 277
An Exact and Robust Conformal Inference Method for Counterfactual and Synthetic Controls
Chernozhukov, V., Wüthrich, K., & Zhu, Y. (2021) · Journal of the American Statistical Association · doi:10.1080/01621459.2021.1920957 · Cited by 176
Self-citationDebiased machine learning of conditional average treatment effects and other causal functions
Semenova, V., & Chernozhukov, V. (2021) · The Econometrics Journal · doi:10.1093/ectj/utaa027 · Cited by 167 · cited 7×
“Semenova and Chernozhukov (2021) developed a related HT strategy that applies series/sieve learners to the denoised HT-transformed outcome, but it explicitly relies on consistent estimation of the regression function, unlike our approach.”
InfluentialSelf-citationUddin, M. J., Shamsuzzaman, M., Horng, L., Labrique, A., Vasudevan, L., Zeller, K., Chowdhury, M., Larson, C. P., Bishai, D., & Alam, N. (2016) · Vaccine · doi:10.1016/j.vaccine.2015.11.024 · Cited by 155
Wakadha, H., Chandir, S., Were, E. V., Rubin, A., Obor, D., Levine, O. S., Gibson, D. G., Odhiambo, F., Laserson, K. F., & Feikin, D. R. (2013) · Vaccine · doi:10.1016/j.vaccine.2012.11.093 · Cited by 149
Cited by
Machine Learning Methods That Economists Should Know About
Athey, S., & Imbens, G. W. (2019) · Annual Review of Economics · doi:10.1146/annurev-economics-080217-053433 · Cited by 1,163
Metadata
- Authors
- Chernozhukov, V., Demirer, M., Duflo, E., & Fernández-Val, I.
- Venue
- National Bureau of Economic Research
- Publisher
- National Bureau of Economic Research
- Published
- 2018-06
- OpenAlex
- W3114809603
- Cited by
- 190
- References
- 14
- Type
- report
- Language
- en
- Open access
- green
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.