Abstract:This paper studies inference in randomized controlled trials with covariate-adaptive randomization when there are multiple treatments. More specifically, we study in this setting inference about the average effect of one or more treatments relative to other treatments or a control. As in Bugni et al.(2018), covariate-adaptive randomization refers to randomization schemes that first stratify according to baseline covariates and then assign treatment status so as to achieve “balance” within each stratum. Importantly, in contrast to Bugni et al. (2018), we not only allow for multiple treatments, but further allow for the proportion of units being assigned to each of the treatments to vary across strata. We first study the properties of estimators derived from a “fully saturated” linear regression, i.e., a linear regression of the outcome on all interactions between indicators for each of the treatments and indicators for each of the strata. We show that tests based on these estimators using the usual heteroskedasticity-consistent estimator of the asymptotic variance are invalid in the sense that they may have limiting rejection probability under the null hypothesis strictly greater than the nominal level; on the other hand, tests based on these estimators and suitable estimators of the asymptotic variance that we provide are exact in the sense that they have limiting rejection probability under the null hypothesis equal to the nominal level. For the special case in which the target proportion of units being assigned to each of the treatments does not vary across strata, we additionally consider tests based on estimators derived from a linear regression with “strata fixed effects,” i.e., a linear regression of the outcome on indicators for each of the treatments and indicators for each of the strata. We show that tests based on these estimators using the usual heteroskedasticity-consistent estimator of the asymptotic variance are conservative in the sense that they have limiting rejection probability under the null hypothesis no greater than and typically strictly less than the nominal level, but tests based on these estimators and suitable estimators of the asymptotic variance that we provide are exact, thereby generalizing results in Bugni et al. (2018) for the case of a single treatment to multiple treatments. A simulation study and an empirical application illustrate the practical relevance of our theoretical results.

Assignment to treatment group on the basis of a covariate

ASSIGNMENT TO TREATMENT GROUP ON THE BASIS OF A COVARIATE

THE ASSESSMENT OF A TREATMENT EFFECT WHEN A COVARIATE IS OBSERVED ON A SUBSET OF THE UNITS: I. SIMPLE ESTIMATORS

Controlling Bias In Observational Studies: A Review

The Use of Covariate Adjustment in Randomized Controlled Trials: An Overview

Covariate adjustment and prediction of mean response in randomised trials

Covariate adjustment for two-sample treatment comparisons in randomized clinical trials: a principled yet flexible approach

Efficient Covariate Balancing for the Local Average Treatment Effect

Regression Analysis for Covariate-Adaptive Randomization: A Robust and Efficient Inference Perspective

The covariate-adjusted residual estimator and its use in both randomized trials and observational settings

Flexible Inference of Optimal Individualized Treatment Strategy in Covariate Adjusted Randomization with Multiple Covariates

The central role of the propensity score in observational studies for causal effects

Inference on Average Treatment Effect under Minimization and Other Covariate-Adaptive Randomization Methods

When does adjusting covariate under randomization help? A comparative study on current practices

Inference under Covariate-Adaptive Randomization with Multiple Treatments

Regression-based multiple treatment effect estimation under covariate-adaptive randomization

Covariate Balancing Sensitivity Analysis for Extrapolating Randomized Trials across Locations

Covariate Adjustment in Stratified Experiments

Accounting for Missing Covariates in Heterogeneous Treatment Estimation

Asymptotic properties of covariate-adaptive randomization

Inference for Two-stage Experiments under Covariate-Adaptive Randomization