Validating cluster size inference: random field and permutation methods