Cohen's kappa (two raters, yes/no)

Computes Cohen's kappa for two raters and a yes/no call, from the four counts of a 2x2 table.

Open the calculator → Runs in your browser. Data leaves only if you deliberately send a problem report from the interactive tool.

Example

Both raters said yes
95
First said yes, second said no
2
First said no, second said yes
3
Both raters said no
0

Result: agreement of 95% with a kappa near zero, the first kappa paradox

The tool opens with these values already filled in. Replace them with your own.

What the result means

The formula
Observed agreement minus the agreement chance gives from the raters own margins, divided by one minus that.
The labels
Below 0 worse than chance, 0 to 0.20 slight, 0.21 to 0.40 fair, 0.41 to 0.60 moderate, 0.61 to 0.80 substantial, 0.81 and above almost perfect.
A convention
Landis and Koch described their own divisions as arbitrary. The labels are not a standard.
Prevalence index
High when almost every case falls in one category, which holds kappa down however well the raters agree.
Bias index
High when the disagreements run in a direction rather than at random. Kappa alone cannot show that.
Undefined
When every case falls in one category for both raters, chance predicts complete agreement and kappa has no value.

What you enter

  • Both raters said yes
  • First said yes, second said no
  • First said no, second said yes
  • Both raters said no
How this is calculated

Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20(1):37-46. Byrt T, Bishop J, Carlin JB. Bias, prevalence and kappa. J Clin Epidemiol. 1993;46(5):423-429. Landis JR, Koch GG. Biometrics. 1977;33(1):159-174. Read the source ↗

A reference and educational tool. Not medical, legal, or financial advice, and not a substitute for clinician judgment.

Browse all tools

Built by Clay Good. Source on GitHub.