Agreement analysis examines how far paired measurements differ. A difference-versus-mean plot makes systematic offset and changing spread visible.
Mathematical principle
Approximate limits of agreement: d̄ ± 1.96sd
The simple limits use the mean and sample standard deviation of paired differences.
Input definitions
| Item | Meaning |
|---|---|
| Paired measurements | Corresponding observations from both methods. |
| Difference direction | State which measurement is subtracted. |
| Measurement units | Keep the difference interpretable. |
Analysis choices
| Item | Meaning |
|---|---|
| Design | Repeated observations need appropriate treatment. |
| Scale | Inspect differences against magnitude. |
| Acceptance | Define scientifically acceptable differences separately. |
Worked mathematical figure
Interpretation
The simple normal-theory limits assume approximately normal differences with stable variability and an appropriate independence structure. Limits are not confidence intervals for the mean difference. High correlation alone does not establish agreement.
Scientific references
Related major models
From the principle to a scoped workflow
Before interpreting paired differences, retain the pairing identifiers, subtraction direction, units and measurement design. Keep an agreement question separate from an association or reliability question.
| Learning path | What to inspect and what it establishes |
|---|---|
| Pair and group repeated data | The preparation lesson keeps subjects, methods, raters, occasions and repeated observations aligned with the question being asked. Evidence scope: This is an illustrative study-preparation path, not a saved native Bland–Altman analysis or product-level statistical validation. |
| Report values and limitations with the figure | Keep the numeric output, calculation definition, units and stated limitations with any plot used in reporting. Evidence scope: A teaching figure does not establish limits of agreement or uncertainty for a new measurement study. |