Rebutal of anon's critique: In fact randomization is the limiting probability distribution for imbalance with minimization

摘要

In reply:The author takes issue [1] with the proposed new statistical analysis convention that should be used with rank minimization [2] and all other algorithms that fit the definition of minimization [3]. He thinks balancing on all characteristics regardless of their importance is illogical for three reasons. First, he says that randomization, not minimization, guarantees that unknown characteristics will be balanced "in probability." "In probability" means that if a study could be repeated an infinite number of times using randomization the effect of the unknown characteristic on the outcome will cancel out.