Fig. 3 shows the effects of misclassification of false positive and false negative comparisons on the apparent performance of a perfect test. In this simulation, there is no overlap between Ground Truth negative patients and Ground Truth positive patients. The test is adopted with 100% accuracy, so that the reduced performance values displayed under different comparative rates of misclassification are only due to uncertainty in the comparator. The variation in the comparative classification error rate between 0% and 20% results in a monotonous decline in CSA and other performance measures. Figure 3 also shows that the observed decrease in the apparent performance of the tests due to the noise of the comparator can be expressed in relation to the maximum possible performance without comparative noise.