What is the ICC(2,1) model used for?

ICC(2,1) uses a two-way random effects model, treating both subjects and raters as random samples. It estimates the reliability of a single rater's measurement and allows generalisation to the population of raters. This is the most conservative and most appropriate ICC model for most clinical and educational research.

What minimum ICC is acceptable for clinical measurement tools?

Clinical measurement tools (physiotherapy outcome measures, inter-rater diagnostic agreement) require ICC ≥ 0.75 for good reliability; ≥ 0.90 for excellent. Educational assessments for individual decisions require ICC ≥ 0.80. ICC 0.60–0.75 is acceptable for research instruments used at the group level.

Intraclass Correlation Coefficient (ICC) for

How It Works

Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent

Note: Interpret the icc for research result against the clinical thresholds and context described above.

How to Use

Enter the values for the patient or scenario you are assessing. Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent. Use the icc for research result to inform your clinical assessment.

Glossary and Definitions

ICC: A key concept in ICC for Research: Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent.
Intraclass correlation: A key concept in ICC for Research: Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent.
Reliability: A key concept in ICC for Research: Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent.
Rater agreement: A key concept in ICC for Research: Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent.
Variance components: A key concept in ICC for Research: Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent.
ANOVA reliability: A key concept in ICC for Research: Calculate ICC from variance components for research reliability analysis. ICC(2,1) for two-way random effects (generalise to all raters); ICC(3,1) for two-way mixed (fixed raters). Koo and Mae (2016): ICC < 0.50 poor, 0.50–0.75 moderate, 0.75–0.90 good, > 0.90 excellent.
ICC for Research: The ICC for Research computes a result from the inputs you provide, based on the method described above.

Intraclass Correlation Coefficient (ICC) for Research

How It Works

How to Use

Glossary and Definitions

Frequently Asked Questions