Decomposing Treatment Effect Variation

DOI10.1080/01621459.2017.1407322zbMATH Open1418.62194arXiv1605.06566OpenAlexW2963256194WikidataQ61450147 ScholiaQ61450147MaRDI QIDQ75398FDOQ75398

Peng Ding, Luke Miratrix, Avi Feller, Luke W. Miratrix, Peng Ding, Avi Feller

Publication date: 21 May 2016

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: Understanding and characterizing treatment effect variation in randomized experiments has become essential for going beyond the "black box" of the average treatment effect. Nonetheless, traditional statistical approaches often ignore or assume away such variation. In the context of randomized experiments, this paper proposes a framework for decomposing overall treatment effect variation into a systematic component explained by observed covariates and a remaining idiosyncratic component. Our framework is fully randomization-based, with estimates of treatment effect variation that are entirely justified by the randomization itself. Our framework can also account for noncompliance, which is an important practical complication. We make several contributions. First, we show that randomization-based estimates of systematic variation are very similar in form to estimates from fully-interacted linear regression and two stage least squares. Second, we use these estimators to develop an omnibus test for systematic treatment effect variation, both with and without noncompliance. Third, we propose an $R^2$-like measure of treatment effect variation explained by covariates and, when applicable, noncompliance. Finally, we assess these methods via simulation studies and apply them to the Head Start Impact Study, a large-scale randomized experiment.

Full work available at URL: https://arxiv.org/abs/1605.06566

Recommendations

• Randomization inference for treatment effect variation
• Randomization Inference in a Group–Randomized Trial of Treatments for Depression
• A note about the identifiability of causal effect estimates in randomized trials with non-compliance
• Rerandomization and regression adjustment
• On equivalencies between design-based and regression-based variance estimators for randomized experiments

zbMATH Keywords

heterogeneous treatment effectidiosyncratic treatment effect variationnoncompliancerandomization inferencesystematic treatment effect variation

Mathematics Subject Classification ID

Nonparametric hypothesis testing (62G10) Applications of statistics to biology and medical sciences; meta analysis (62P10) Foundations and philosophical topics in statistics (62A01)

Cites Work

• Estimating treatment effect heterogeneity in randomized program evaluation
• Model assisted survey sampling.
• On the application of probability theory to agricultural experiments. Essay on principles. Section 9. Translated from the Polish and edited by D. M. Dąbrowska and T. P. Speed
• A paradox from randomization-based causal inference
• Randomization-based causal inference from split-plot designs
• Robust regression: Asymptotics, conjectures and Monte Carlo
• Semiparametric instrumental variable estimation of treatment response models.
• Agnostic notes on regression adjustments to experimental data: reexamining Freedman's critique
• Compared to what? Variation in the impacts of early childhood education by alternative care type
• Interpretation of interaction: a review
• Sharp bounds on the variance in randomized experiments
• Instrumental variables: an econometrician's perspective
• Evaluating marker-guided treatment selection strategies
• Heterogeneous impacts in PROGRESA
• Recursive partitioning for heterogeneous causal effects
• Principal Stratification in Causal Inference
• Identification of Causal Effects Using Instrumental Variables
• Title not available (Why is that?)
• Title not available (Why is that?)
• Interaction
• Title not available (Why is that?)
• Instrumental Variables Regression with Weak Instruments
• Making The Most Out Of Programme Evaluations and Social Experiments: Accounting For Heterogeneity in Programme Impacts
• Estimation and Inference of Heterogeneous Treatment Effects using Random Forests
• Regression-assisted inference for the average treatment effect in paired experiments
• Assessing Treatment‐Selection Markers using a Potential Outcomes Framework
• Reduced Sensitivity to Hidden Bias at Upper Quantiles in Observational Studies with Dilated Treatment Effects
• Causal Inference for Statistics, Social, and Biomedical Sciences
• Sensitivity Analysis for m‐Estimates, Tests, and Confidence Intervals in Matched Observational Studies
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Observational studies.
• Randomization Inference for Treatment Effect Variation
• A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity

Cited In (10)

• Design-Based Ratio Estimators and Central Limit Theorems for Clustered, Blocked RCTs
• Comment: Invariance and causal inference
• Sample size requirements for detecting treatment effect heterogeneity in cluster randomized trials
• Flexible Sensitivity Analysis for Observational Studies Without Observable Implications
• Causal mediation analysis with latent subgroups
• Estimating heterogeneous causal effects in observational studies using small area predictors
• hettx
• Modern approaches for evaluating treatment effect heterogeneity from clinical trials and observational data
• Robust machine learning for treatment effects in multilevel observational studies under cluster-level unmeasured confounding
• Assumption-Lean Analysis of Cluster Randomized Trials in Infectious Diseases for Intent-to-Treat Effects and Network Effects

This page was built for publication: Decomposing Treatment Effect Variation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q75398)