Sample size and cut-off: Implications for intra-population matching probabilities