ICC coef correlação intraclasse
- Created by
- Renato Passos, Eng. de Software
- Reviewed by
- Renato Passos, Eng. de Software
Last updated: Apr 18, 2026
About this calculator
The ICC (Intraclass Correlation Coefficient) calculator measures the consistency or agreement among different observers or measurements. It is widely used in scientific, medical, and psychological research to assess the reproducibility of data. The formula compares the variance between groups (MSB) and within groups (MSW), adjusted by the number of observations (k).
The standard ICC formula is (MSB − MSW) / (MSB + (k − 1) × MSW). This calculates the proportion of total variability attributable to differences between groups, assuming observations within a group are more similar. Values close to 1 indicate high agreement, while low values suggest inconsistency.
Use this calculator when validating data reliability from multiple raters or methods, such as in clinical trials, motor skill tests, or satisfaction surveys. Ensure data follows a normal distribution and groups are independent for accurate results.
Common precautions include avoiding small or unbalanced samples and interpreting results within the study's context. Intermediate values may require complementary analysis, like skewness tests or residual variation checks, to confirm model fit.
Frequently asked questions
What does ICC measure?
ICC measures agreement or consistency among observers or measurements in quantitative data. It's widely used in clinical and tool validation studies.
How does it differ from Pearson's correlation coefficient?
ICC assesses agreement between groups or raters, while Pearson measures linear relationship between two variables.
What ICC value is considered good?
Values above 0.75 are high, but this varies by field. Medical studies often require ICC ≥ 0.90 for critical validity.
When should I use this calculator?
Use it when validating reproducibility in studies with multiple raters or repeated tests.
Are there prerequisites for using the results?
Yes, data must follow normal distribution and groups must be independent. Unbalanced samples may affect accuracy.