When can we consider a research sample as the ""best""?