Choose your reading route
eduKateSG · Why Mathematics? · Degrees of Equivalence, Reference Values and International Measurement Comparisons
Start with the question, audit the worked example, or jump to the assumptions and practice. Every chapter stays open for uninterrupted reading.
Understand the question
Define the quantity and follow the mechanism.Route 2 · Chapters 4–7
Reproduce the example
Audit the numbers one step at a time.Route 3 · Chapters 8–13
Check assumptions
See where design and calibration matter.Route 4 · Chapters 14–18
Explore applications
Connect the method to a concrete investigation.Route 5 · Chapters 19–23
Study and verify
Practise, inspect failure cases and read further.
Full chapter index · Practice and checking · Mathematics Learning Hub
In an international measurement comparison, laboratories do more than report numbers. Their results are related to a key comparison reference value and to one another through degrees of equivalence, each paired with an uncertainty. The mathematics preserves both difference and confidence in that difference.
This topic explains why mathematics is important in global science and trade: comparable standards depend on transparent differences, uncertainty propagation and agreed protocols. A degree of equivalence is a technical comparison result, not a league-table score.
Full chapter index
Chapters 1–3 · Understand the question
Chapters 4–7 · Reproduce the example
Chapters 8–13 · Check assumptions
Chapters 14–18 · Explore applications
Chapters 19–23 · Study and verify
CHAPTER 1 OF 23 · Understand the question
1. What a degree of equivalence represents
BIPM guidance describes degrees of equivalence as the degree to which national measurement standards are consistent with key comparison reference values and therefore with other national standards.
For participant i, a common form is D_i = x_i – x_R, paired with an uncertainty U_i. The sign shows whether the result is above or below the reference; the uncertainty shows how precisely that difference is known.
CHAPTER 2 OF 23 · Understand the question
2. The reference value is protocol-defined
The key comparison reference value, x_R, is determined by the method agreed for that comparison. It may be a weighted mean, a median, a regression-based value or another justified estimator.
Do not assume every comparison uses the same formula. The reference-value method, exclusions, correlations and linkage procedure belong in the final report.
Suppose x_A = 100.8 units with standard uncertainty u_A = 0.5 unit. Let x_R = 100.0 units with standard uncertainty u_R = 0.3 unit. In a simplified independent example, D_A = 100.8 – 100.0 = 0.8 unit.
The standard uncertainty of the difference is u(D_A) = square root of 0.5 squared plus 0.3 squared = square root of 0.34, about 0.5831 unit.
| Quantity | Value |
|---|---|
| x_A | 100.8 |
| u_A | 0.5 |
| x_R | 100.0 |
| u_R | 0.3 |
With coverage factor k = 2, U_A = 2 × 0.5831 = 1.1662 units. The reported pair is D_A = 0.8 unit with U_A = 1.17 units under the simplified assumptions.
An illustrative expanded interval is 0.8 ± 1.1662, from about -0.3662 to 1.9662 units. Because it includes zero, the observed difference is not resolved from zero at that stated coverage under this model.
| Quantity | Calculation | Result |
|---|---|---|
| D_A | 100.8−100.0 | 0.8 |
| u(D_A) | √(0.5²+0.3²) | 0.5831 |
| U_A, k=2 | 2×0.5831 | 1.1662 |
For two participants A and B, D_AB = x_A – x_B. If x_B = 99.6 units, then D_AB = 1.2 units.
With independent standard uncertainties u_A = 0.5 and u_B = 0.4, u(D_AB) = square root of 0.25 + 0.16 = square root of 0.41, about 0.6403. At k = 2, U_AB is about 1.2806 units.
Algebraically, D_A – D_B = (x_A – x_R) – (x_B – x_R) = x_A – x_B. The common reference term cancels.
Uncertainty does not always reduce to a simple independent root-sum-of-squares because comparison uncertainty and correlations may remain. Use the final report's covariance model.
CHAPTER 7 OF 23 · Reproduce the example
7. Difference plus uncertainty is one result
Reporting D without U can exaggerate a small observed gap. Reporting U without D hides direction and size. The pair belongs together, with unit and coverage factor.
A chart should show points and uncertainty bars. Ordering points without their bars creates a misleading ranking.
CHAPTER 8 OF 23 · Check assumptions
8. Degrees of equivalence are not performance grades
A participant's difference can reflect artefact drift, comparison transport, reference-value construction and shared uncertainties as well as its own measurement. The final report explains the model.
The purpose is mutual recognition and technical comparability, not public shaming.
Measurement comparisons support the CIPM Mutual Recognition Arrangement by establishing documented comparability among national standards. The BIPM Key Comparison Database records comparison results and degrees of equivalence.
This infrastructure helps calibration and measurement claims travel across borders with evidence.
Regional metrology organisation comparisons can be linked to a CIPM key comparison through participating institutes and a documented linkage procedure. BIPM guidance requires the linkage and associated uncertainty to be available in the final report.
Linkage is a mathematical bridge, not a copied reference number.
If x_i contributes to the reference value, x_i and x_R are correlated. The variance of their difference is var(x_i) + var(x_R) – 2cov(x_i,x_R).
Ignoring covariance can overstate or understate uncertainty. The comparison method should specify how self-inclusion and common standards are handled.
Transfer artefacts can drift. A BIPM gauge-block comparison determined some reference values as functions of time for long gauge blocks. A static reference would have mixed time change with laboratory difference.
Mathematics allows a time model, but the physical evidence must justify it.
Normalised error divides a difference by a combined uncertainty scale and supports a threshold rule in some comparisons. Degree of equivalence retains the difference and its uncertainty in the original unit.
Read the recent normalised-error article for that complementary ratio.
A Birge ratio evaluates overall statistical consistency across a set of results. Degrees of equivalence describe participant-to-reference and participant-to-participant differences.
One global diagnostic cannot replace the individual pairs, and individual pairs cannot replace the overall model check.
Several groups measure one stable safe object and agree on a classroom reference estimator. Calculate every group-to-reference difference and pairwise difference with a declared uncertainty model.
Then change the reference estimator and observe that pairwise raw differences stay fixed while group-to-reference differences may change. Label the activity as an analogy.
Use columns for participant, result, standard uncertainty, covariance with the reference, D_i, u(D_i), coverage factor and U_i. Keep units in metadata and formulas visible.
For pairwise comparisons, generate a symmetric matrix of x_i – x_j and a matching uncertainty matrix. Check that D_ij = -D_ji and D_ii = 0.
Represent results as a vector and uncertainties as a covariance matrix. Use linear contrasts to calculate participant-to-reference and pairwise differences. The variance of a contrast a is a-transpose Sigma a.
Test antisymmetry and unit consistency. Preserve the protocol version and reference-value calculation.
Students can connect subtraction, uncertainty, algebraic cancellation, matrices and graphs. Ask why pairwise differences form an antisymmetric table and why covariance enters with a minus sign.
Parents can encourage students to treat uncertainty as part of the answer. Comparability is stronger when limits are visible.
Degrees of equivalence support national metrology, calibration, standards, advanced manufacturing and scientific cooperation. The transferable mathematics includes uncertainty propagation, regression, covariance and transparent reporting.
Explore connected learning through the Mathematics Learning Hub and How Mathematics Works.
Change x_A to 100.4 and recompute D_A and its unchanged uncertainty. Double every value and uncertainty and verify how D and U scale. Add positive covariance between x_A and x_R and calculate how u(D_A) changes.
For three participants, build the full pairwise difference matrix and check every sign.
### Is zero difference required for equivalence? Observed differences rarely equal exactly zero. Interpretation uses the difference together with its uncertainty and the comparison protocol.
### Is the reference value the true value? It is the agreed comparison reference estimate with uncertainty, not an unknowable exact truth.
### Why can two reports use different formulas? Comparisons differ in measurand, artefact behaviour, correlations and reference-value method. The approved protocol governs each one.
### Can degrees of equivalence rank laboratories? They are technical comparison results with uncertainties, not simple performance rankings.
Use the current BIPM guidance Measurement comparisons in the CIPM MRA, the BIPM KCDB comparison display example and the recent CCQM-K82 comparison report. They document reference values, participant and pairwise degrees of equivalence, linkage and uncertainty. The worked numbers here are original teaching data.
Degrees of equivalence turn international comparability into visible mathematics. A difference preserves direction, uncertainty preserves humility, and the agreed protocol lets many laboratories speak a common measurement language.
Extended practice laboratory
Practice case 1: Move participant A
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 1. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 2: Move the reference value
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 2. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 3: Double all units
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 3. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 4: Add participant B
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 4. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 5: Build pairwise differences
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 5. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 6: Check antisymmetry
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 6. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 7: Check the diagonal
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 7. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 8: Add covariance with reference
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 8. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 9: Remove covariance
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 9. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 10: Change coverage factor
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 10. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 11: Use a time-dependent reference
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 11. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 12: Compare two reference estimators
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 12. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 13: Link a regional comparison
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 13. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 14: Preserve common reference cancellation
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 14. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 15: Include comparison uncertainty
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 15. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 16: Plot uncertainty bars
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 16. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 17: Avoid ranking
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 17. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 18: Interpret an interval containing zero
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 18. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 19: Interpret an interval excluding zero
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 19. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 20: Audit the KCRV method
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 20. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 21: Audit traceability
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 21. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 22: Create a covariance matrix
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 22. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 23: Check units
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 23. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.
Practice case 24: Write a protocol-limited conclusion
Setup: Use the article’s degree-of-equivalence example and change one declared input for case 24. Task: Recalculate from raw values, preserve units and state which assumption or design condition changed.
Check: Check the relevant invariant, separate the numerical result from the claim it supports, and name the next diagnostic or source of evidence. Rebuild the relevant counts before interpreting the coefficient, preserve the stated design, and write one sentence separating the numerical result from the claim it can support. Then vary one defensible assumption, calculate again, and explain whether the conclusion changes because of new data, a new reference model, or a new decision purpose. Finish by naming the unit, tolerance, and evidence needed for transfer. Record one graph or table that would expose failure hidden by the summary.