Cohen’s Kappa Calculator
Agreement data
Choose 2–6 mutually exclusive categories.
Use weighting only for ordered categories.
Enter labels separated by commas. The matrix rows are Rater A; columns are Rater B.
Data format
Contingency table counts
Each cell is the number of subjects receiving the row category from Rater A and the column category from Rater B. Counts must be non-negative whole numbers.
Live results
Cohen’s κ
—
—descriptive interpretation
Observed agreement
—
—
Expected agreement
—
from marginal distributions
Confidence interval
—
Approximate large-sample CI
Sample size
—
Wald test of κ = 0
Cohen’s kappa adjusts observed agreement for agreement expected from the raters’ marginal category frequencies.
How this is calculated
Enter valid counts to see the calculation.
Cohen’s κ is designed for two raters assigning the same set of categories. Weighted κ is appropriate when category order is meaningful. The confidence interval and p-value shown here use a large-sample delta-method standard error and normal approximation, so small or sparse samples should be interpreted cautiously.
Reference: Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20(1), 37–46. Common descriptive labels are only heuristics and should not replace subject-matter judgment.
This Cohen’s Kappa Calculator measures agreement between two raters while correcting for agreement expected by chance. Enter a square contingency table for binary or multicategory ratings to obtain Cohen’s κ, observed and expected agreement, an approximate confidence interval, a significance test, and a visual summary. For ordered rating scales, linear or quadratic weighted kappa can give partial credit when raters disagree by only one or a few category levels.
