The Pairwise Matching Design Is Optimal Under Extreme Noise and Extreme Assignments

ABSTRACT We consider the general performance of the difference‐in‐means estimator in an equally‐allocated two‐arm randomized experiment under common experimental endpoints such as continuous (regression), incidence, proportion, count and uncensored survival. We consider two sources of randomness: the subject‐specific assignments and the contribution of unobserved subject‐specific measurements. We then examine mean squared error (MSE) performance under a new, more realistic simultaneous tail criterion . We prove that the pairwise matching design asymptotically performs best under the simultaneous tail criterion when compared with other blocking designs for all response types. We also prove that the optimal design must be less random than complete randomization and more random than any deterministic, optimized allocation. Theoretical results are supported by simulations.