Why randomize? Minimax optimality under permutation invarianceYuehao Bai
Why Randomize? Minimax Optimality under Permutation Invariancedoi:10.2139/ssrn.3475147Experimental designpermutation invarianceaverage treatment effectThis paper studies finite sample minimax optimal randomization schemes and estimation schemes in estimating parameters including the average treat- ment effectSocial ...
I assume the parameter of interest is τ′μ for τ=(τ0,τ1)′.As an example, if τ=(−1,1)′, then τ′μ is the ATE. Define An=∑i= Conclusion This paper shows the minimax optimality of the complete randomization scheme under permutation invariance. I also characterize the ...