A peptide assay can produce nearly identical results each time and still give the wrong amount. It can also give a mean close to a reference while individual results vary widely. Accuracy and precision describe different aspects of measurement performance, so neither should be inferred from the other.
Keep accuracy, trueness and precision distinct
The International Vocabulary of Metrology describes accuracy as agreement between a measured value and the true value of the quantity being measured. It treats accuracy as a qualitative concept related to both trueness and precision, rather than a quantity with its own numerical unit.JCGM VIM — Measurement accuracy (opens in a new tab)
Precision concerns agreement among repeated measurements under specified conditions. Trueness concerns the relationship between the long-run average and a reference value. A finite experiment estimates aspects of those behaviours; it does not reveal an exact true value without uncertainty.JCGM VIM — Measurement precision (opens in a new tab)JCGM VIM — Measurement trueness (opens in a new tab)
| Question | Relevant evidence |
|---|---|
| Do repeats agree? | Spread under the stated measurement conditions |
| Is there a consistent offset? | Comparison with an appropriate reference |
| Is the result fit for its purpose? | The combined performance against defined requirements |
Validation reports sometimes use “accuracy” for a study of recovery or bias. Read the actual experiment and calculation rather than assuming that one heading covers every source of measurement error.
Compare a tight biased set with a variable centred set
Consider an original example with a hypothetical reference value of 10.00 mg. Method A gives 10.48, 10.50 and 10.52 mg. Its mean is 10.50 mg and its sample standard deviation is 0.02 mg. The repeats are close together, but the mean is 0.50 mg above the reference.
Method B gives 9.60, 10.00 and 10.40 mg. Its mean matches 10.00 mg, but its sample standard deviation is 0.40 mg. That mean alone hides much greater variation between individual measurements.
Calling A “more accurate” because its repeats agree ignores the offset. Calling B universally accurate because its mean matches the reference ignores the spread and the limited evidence.
Use a reference that matches the quantity being measured
The reference must describe the same quantity and material basis as the assay. An assigned peptide mass fraction is not interchangeable with a chromatographic peak-area percentage, and a dry-basis value differs from an as-received material value.
If a standard’s assigned content is wrong for the calculation, repeated sample measurements can inherit a consistent scale error. Excellent injection repeatability does not independently verify the standard’s amount.
A recovery experiment can investigate the effect of adding a known amount, but its scope depends on when that amount enters the procedure. Adding a standard after extraction cannot measure a loss that occurred before the addition.
For a real comparison, identify the reference, its uncertainty and the complete preparation. Otherwise, an apparent difference can combine analytical bias with a mismatch between the quantities being compared.
Report the evidence without collapsing it into one percentage
A useful performance statement names the tested range, the observed bias or recovery, the spread and the repeat conditions. It should also explain whether the results came from independent preparations or repeated measurements of one preparation.
An RSD describes relative spread; it is not a percentage of overall accuracy. A recovery value describes the outcome of its defined experiment; it is not a guarantee against every potential interference.
The required performance depends on the analytical purpose. A method used to compare large changes may face different demands from one used to resolve a small difference near a specification. The evidence should support the actual decision.
Sources and further detail
- JCGM VIM — Measurement accuracy (opens in a new tab)
Official definition and notes, including the relationship to trueness and precision and the absence of a numerical accuracy quantity.
- JCGM VIM — Measurement precision (opens in a new tab)
Official definition of agreement between replicate measurements under specified conditions.
- JCGM VIM — Measurement trueness (opens in a new tab)
Official long-run-average/reference distinction. Worked numbers are original illustrations, not reference-material results.
Sources checked 19 September 2026. Worked examples are illustrative unless a supplied report is explicitly identified. This article has not undergone independent scientific peer review.