Unobserved Covariate Imbalance of Covariate-Adaptive Randomized Experiments

• Activity Number: 186 - Contributed Poster Presentations: International Chinese Statistical Association
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Yang Liu* and Feifang Hu
• Companies: George Washington University and George Washington University
• Keywords: unobserved covariates imbalance; covariate-adaptive randomization; minimization; stratified permuted block; clinical trails

Abstract

As balancing important covariates is always a critical issue in comparative studies such as causal inference and clinical trials. Stratified permuted block randomization and covariate-adaptive randomization are two popular methods implemented in practice. However, some important covariates could be unknown or unobserved due to certain difficulties. It is important to know the balance properties of these unobserved covariates under randomization procedures. In this paper, we propose a general theory to analyze the unobserved covariate imbalance by using the properties of the observed stratum imbalance under different randomization procedures and obtain their theoretical properties. The unobserved covariate imbalance under complete randomization, stratified permuted block, and minimization based covariate-adaptive randomization are compared. Numerical analysis of hypothetical clinical trials and mimic real clinical trials are used to justify our results.