COMPSTAT 2022: Start Registration
View Submission - COMPSTAT2022
Title: Pair-switching rerandomization Authors:  Ke Zhu - Tsinghua University (China) [presenting]
Hanzhong Liu - Tsinghua University (China)
Abstract: Rerandomization discards assignments with covariates unbalanced in the treatment and control groups to improve the estimation and inference efficiency. However, the acceptance-rejection sampling method used by rerandomization is computationally inefficient. As a result, it is time-consuming for rerandomization to draw numerous independent assignments, which are necessary for performing Fisher randomization tests. To address this problem, a pair-switching rerandomization method is proposed to draw balanced assignments more efficiently. Under pair-switching rerandomization, the unbiasedness and variance reduction of the difference-in-means estimator are obtained, and valid Fisher randomization tests are developed. The proposed method is applicable in both non-sequentially and sequentially randomized experiments. Moreover, an exact approach is proposed to invert Fisher randomization tests to confidence intervals, which is faster than the existing methods and applicable to any experimental design. Comprehensive simulation studies are conducted to compare the finite-sample performances of the proposed method and classical rerandomization. Simulation results indicate that pair-switching rerandomization leads to comparable power of Fisher randomization tests and is 3-23 times faster than classical rerandomization. Finally, the pair-switching rerandomization method is applied to analyze two clinical trial data sets, both demonstrating the advantages of the proposed method.