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.

Bootstrapping Inference of Average Treatment Effect in Completely Randomized Experiments with High-Dimensional Covariates

Sharp variance estimator and causal bootstrap in stratified randomized experiments

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

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

Treatment bootstrapping: A new approach to quantify uncertainty of average treatment effect estimates

Regression-adjusted Average Treatment Effect Estimates in Stratified Randomized Experiments

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

Inference under Covariate-Adaptive Randomization with Multiple Treatments

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

Debiased regression adjustment in completely randomized experiments with moderately high-dimensional covariates

Treatment effect estimation under covariate-adaptive randomization with heavy-tailed outcomes

Leveraging baseline covariates to analyze small cluster-randomized trials with a rare binary outcome

Inference for Two-stage Experiments under Covariate-Adaptive Randomization

Approximate Bayesian Bootstrap procedures to estimate multilevel treatment effects in observational studies with application to type 2 diabetes treatment regimens

Randomization-based Joint Central Limit Theorem and Efficient Covariate Adjustment in Randomized Block 2K Factorial Experiments

Testing for Treatment Effect in Covariate-Adaptive Randomized Clinical Trials with Generalized Linear Models and Omitted Covariates

Blocking, rerandomization, and regression adjustment in randomized experiments with high-dimensional covariates

Randomization-based joint central limit theorem and efficient covariate adjustment in stratified $2^K$ factorial experiments

Asymptotic properties of covariate-adaptive randomization

Lasso-adjusted treatment effect estimation under covariate-adaptive randomization

A Unified Framework for Covariate Adjustment under Stratified Randomization