Empirical selection of matching factors in matched-pairs and matched-blocks small-sample research designs