Intraclass correlation coefficients describe reliability under a specified measurement model. Test–retest, intrarater, and interrater designs need different consideration.
Input definitions
| Item | Meaning |
|---|---|
| Targets | The participants or objects being measured. |
| Ratings | Measurements indexed by rater or occasion. |
| Design | How targets and raters were selected. |
Analysis choices
| Item | Meaning |
|---|---|
| Model | State the selected effects structure. |
| Type | Distinguish a single measurement from an average. |
| Definition | Distinguish consistency from absolute agreement. |
Interpretation
Report the complete ICC specification, software information, estimate, and interval. An unlabeled ICC value leaves the measurement question ambiguous. Choose the form to match the intended use of the measurements.
Scientific references
Related major models
From the principle to a scoped workflow
A reliability coefficient needs its measurement design and definition. Record the intended model, single-measure or average-measure target, and agreement or consistency interpretation before reading an ICC value.
| Learning path | What to inspect and what it establishes |
|---|---|
| Preserve raters, sessions and repeated observations | Use the preparation lesson to identify what is repeated, which observations belong together and what the measurement setting represents. Evidence scope: The lesson does not choose an ICC model for the study or provide native reliability-analysis acceptance. |
| Keep the estimate with its definition and evidence | Report the exact coefficient definition, available numeric output and limitations instead of relying on a favorable-looking summary value. Evidence scope: An authored method reference does not establish all ICC variants, interval procedures or a whole statistical product. |