We can say that when compared to a Merkle tree, inclusion proofs are constant size by membership witness. Can we say that also for exclusion proof bu nonmembership witness? Is there any advantage of nonmembership witness in an accumulator?
In accumulators based on groups of unknown order (such as RSA or class groups) or accumulators based on bilinear pairings, both membership and non-membership proofs are constant-sized (i.e. independent of the size of the set and the size of the subset you are proving membership/non-membership for).