Specificity

Specificity is the proportion of truly negative people (by the reference standard) whom the test recognizes as negative: true negatives divided by all negatives. A test with high specificity rarely raises a false alarm; a positive result then makes the condition more likely - though for rare conditions often less than expected (see predictive values).

Example: Of 100 children without a diagnosis, 88 stay unremarkable in the screening: specificity = 88% (95% CI 80 to 93%).

Caution: The reference standard is taken as the truth; if it is itself uncertain, sensitivity and specificity are biased.