Precision describes how closely repeated measurements agree under specified conditions. Accuracy concerns agreement with an accepted reference value and involves both systematic and random effects. Repeating nearly the same number does not establish accuracy by itself.
| Question | Concept | What to compare |
|---|---|---|
| Do repeated readings cluster together? | Precision | The spread of repeated measurements under stated conditions. |
| Do measurements agree with an accepted reference? | Accuracy | Reference agreement, including bias and random effects. |
| Are enough digits displayed? | Numerical presentation | Digits alone do not demonstrate either property. |
A fictional calibration example
Imagine a teaching material with an accepted reference value of 100 arbitrary units. It is not a patient sample, and these numbers are not medical ranges. One demonstration produces 109, 110 and 111. The readings cluster tightly, but their average is 110, ten units above the reference.
Another demonstration produces 98, 100 and 102. Its average is 100, but the individual readings have a wider spread. This illustrates why agreement among repeated readings and agreement with a reference are different questions. A matching average does not make every individual reading exact.
Check the arithmetic, then the meaning
For the first group, the sum is 330 and the mean is 330 divided by 3, or 110. Its largest-minus-smallest spread is 2. For the second, the sum is 300, mean 100 and spread 4. These simple summaries illustrate the distinction; three readings are not a complete method-validation study.
NIST treats accuracy as a qualitative concept and warns against using precision as its synonym. A laboratory describes performance with appropriate quantitative measures and stated conditions. The fictional example does not supply an accuracy percentage or allow comparison of real laboratories.
Why a patient cannot check this by averaging unrelated reports
Results from different collection dates are not repeated measurements of an unchanged reference material. The person, specimen conditions, method or clinical situation may have changed. Averaging them does not create a verified reference value or reveal the accuracy of either result.
Similarly, two devices displaying the same rounded number do not establish equivalence. Their intended use, specimen, units and performance information still matter. Quality assessment is a laboratory responsibility supported by suitable materials and procedures.
Useful questions about a result
- Am I comparing repeated analysis of one material or separate patient specimens?
- Are the method, units and conditions the same?
- Is the comment about repeatability, a method comparison or another investigation?
- Has the laboratory supplied an interpretation of the discrepancy?
For an unexpected clinical result, contact the responsible team rather than trying to grade a laboratory from a few portal numbers. The point of understanding these terms is to ask a more precise question, not to select a diagnosis, discard a result or calculate your own treatment adjustment.
Scope and authorship
This is general adult report-reading education, not personal medical advice, diagnosis or treatment. The clinician responsible for your care must interpret individual results. Follow direct urgent instructions rather than waiting for an online explanation.
Prepared with AI assistance for DoubtPay; checked against linked sources. Not reviewed by a clinician.
Start with the basic blood-test report guide or browse the Medical blog. View Report Samjho and its current terms.