What does overall percent agreement measure in reliability studies?

• A. The proportion of observations where two tests/observers agree out of the total observations.
• B. The proportion of observations where both tests are positive divided by the total observations.
• C. The proportion of discordant results between observers.
• D. The average of positive rates between observers.

Answer: (A)

Absolute agreement between two raters or tests is what overall percent agreement measures. It captures how often the two assessments yield the same result across all observations. You count every case where both are positive or both are negative, then divide by the total number of observations. This gives a straightforward sense of consistency between raters or tests. It treats positive and negative agreements equally but does not adjust for the possibility that some agreement occurs by chance. That’s why, in deeper reliability work, chance-adjusted measures like kappa are also used. The other ideas don’t fit as well: counting only the cases where both are positive ignores agreements on negatives; looking at discordant results reflects disagreement rather than overall agreement; and averaging each observer’s positive rate doesn't measure how often they agree on the same observation.