In fact, as shown in table 1, as the observed performance comes closer to the desired target we do require larger sample sizes to show significant differences. For example, you would need an audit ...