TimelyScholarFree Tools › ICC Calculator (Intraclass Correlation)
Measure inter-rater or test-retest reliability. Paste ratings (subjects in rows, raters in columns) to get all six ICC forms with 95% confidence intervals, F tests and a plain-language interpretation.
FreeNo account neededYour data stays in your browser
| Type | Description | ICC | 95% CI | F | df | p |
|---|---|---|---|---|---|---|
| ICC1 | Single rater, absolute agreement, one-way random | 0.935 | 0.83 to 0.98 | 44.34 | 9, 20 | < 0.001 |
| ICC2 | Single rater, absolute agreement, two-way random | 0.936 | 0.82 to 0.98 | 54.02 | 9, 18 | < 0.001 |
| ICC3 | Single rater, consistency, two-way mixed | 0.946 | 0.85 to 0.99 | 54.02 | 9, 18 | < 0.001 |
| ICC1k | Mean of k raters, absolute agreement, one-way random | 0.977 | 0.94 to 0.99 | 44.34 | 9, 20 | < 0.001 |
| ICC2k | Mean of k raters, absolute agreement, two-way random | 0.978 | 0.93 to 0.99 | 54.02 | 9, 18 | < 0.001 |
| ICC3k | Mean of k raters, consistency, two-way mixed | 0.981 | 0.95 to 0.99 | 54.02 | 9, 18 | < 0.001 |
ICC2 = 0.94 (95% CI 0.82 to 0.98). Using Koo and Li's guideline, which is applied to the 95% CI, reliability is between good and excellent.
Choosing the form: use a two-way random-effects model (ICC2) when raters are a sample of a wider population of raters, two-way mixed (ICC3) when these are the only raters of interest, and one-way (ICC1) when each subject is rated by different raters. Use "single" if in practice one rater's score will be used, and "mean of k" if the average will be used.
Copy the columns from Excel or Google Sheets, with names in the first row.
The results update instantly in your browser.
Paste straight into Word, or download a CSV.
The intraclass correlation measures how strongly measurements of the same subject agree. Different forms suit different designs: one-way or two-way models, absolute agreement or consistency, and single or averaged measurements.
Report the model, type and form you used. Koo and Li suggest interpreting the 95% confidence interval: below 0.5 poor, 0.5 to 0.75 moderate, 0.75 to 0.9 good and above 0.9 excellent reliability.
For inter-rater reliability where raters are a sample of possible raters, ICC2 (two-way random, absolute agreement) is common; use the average-measures form only if the mean of several raters will be used in practice.
Use the ICC for continuous measurements and kappa for categorical ratings.