Decisions Classifying and Test Utility
Whenever we use a test to make decisions, there will be some errors Increasing the correlation between test g and criterion will reduce errors Can manipulate the kinds of correct or incorrect decisions by altering our "cutoff" scores
Correct Decisions:
True Positives (TP) score well on both the True Negatives (TN) score poorly on both
test and the criterion
the test and the criterion
Incorrect Decisions:
Positives False (FP) score well on the test, False Negatives (FN) score poorly on the
but perform poorly on the criterion
test, but perform well on the criterion
1
Actual Performance Success Failure
Reject (predict failure) False Negatives True Negatives
A...