Which statement best describes the kappa statistic?

• A. The observed agreement.
• B. The proportion of positive results among all observations.
• C. It evaluates how much better than random chance the observed agreement is, adjusting for expected agreement.
• D. It is the same as the expected agreement.

Answer: (C)

Kappa measures agreement between two raters that goes beyond what would be expected by chance, taking into account how often each rater tends to assign each category. It uses the observed agreement (how often the raters actually agree) and the expected agreement (how often they would agree just by luck given their individual rating patterns), and combines them in the formula kappa = (Po − Pe) / (1 − Pe). This means it tells you how much better than random chance the observed agreement is, relative to the maximum possible agreement beyond chance. If there is no agreement beyond chance, kappa is zero; if there is perfect agreement beyond chance, kappa is one; if agreement is worse than chance, kappa can be negative. So the statement that best describes the kappa statistic is that it evaluates how much better than random chance the observed agreement is, adjusting for expected agreement. The other ideas—just the observed agreement, the proportion of positives, or the expected agreement itself—don’t capture this chance-adjusted measure.