Measurement uncertainty is a parameter that characterises the range of values within which the true value of a measurement is expected to lie, given a stated level of confidence. Every measurement (no matter how carefully made, no matter how precise the instrument), contains some degree of doubt. Understanding and quantifying that doubt is not a sign of weakness in a measurement system; it is a sign of technical rigour and honesty. ISO/IEC 17025:2017, the international standard for calibration laboratory competence, requires accredited laboratories to determine and report measurement uncertainty for all calibration results.
Why Measurement Uncertainty Exists
No physical measurement is perfectly exact. Even the most sophisticated reference standard used by a national metrology institute has some residual uncertainty in its realisation of the unit of measurement, which propagates downward through the calibration chain to every instrument calibrated against it. Multiple sources contribute to the total uncertainty in a calibration: resolution of the instrument being calibrated (a thermometer reading to 0.1°C cannot resolve smaller differences); repeatability, the standard deviation of repeated readings under identical conditions; reference standard uncertainty, carried from the reference standard's own calibration certificate; environmental conditions — temperature, humidity and vibration at the time of calibration; operator effects such as reading parallax and technique; and instrument drift between the start and end of a calibration sequence.
The Guide to the Expression of Uncertainty in Measurement (GUM)
The internationally accepted framework for calculating and expressing measurement uncertainty is the GUM, published jointly by BIPM, IEC, IFCC, ISO, IUPAC, IUPAP and OIML, and used by all ISO/IEC 17025 accredited laboratories worldwide, including those accredited under SAC-SINGLAS. The GUM classifies contributions into two categories: Type A, estimated by statistical analysis of repeated measurements (the standard deviation of ten repeated readings, for example); and Type B, estimated by other means — from reference standard certificates, manufacturer specifications, published data, or engineering judgement. Both are combined using the law of propagation of uncertainty to produce a combined standard uncertainty, multiplied by a coverage factor (typically k = 2 for approximately 95% confidence) to give the expanded uncertainty reported on the certificate.
How to Read Uncertainty on a Calibration Certificate
A well-prepared certificate states uncertainty alongside each result, typically as "Expanded uncertainty: ±0.3°C (k = 2, approximately 95% confidence)" — meaning the laboratory is approximately 95% confident the true value lies within ±0.3°C of the reported result. When interpreting a certificate, check that: uncertainty is reported for each measurement point (or the worst-case point is clearly stated); the coverage factor k and confidence level are stated; the result and uncertainty share the same units; and uncertainty is clearly distinguished from any stated tolerance. A certificate without measurement uncertainty does not comply with ISO/IEC 17025 and cannot have been issued by an accredited laboratory operating within scope. See our guide to how to read a calibration certificate.
Fitness for Purpose and Singapore Regulatory Context
For a measurement to be meaningful, its uncertainty must be sufficiently small relative to the tolerance being verified. A common guideline is the 4:1 ratio rule (test accuracy ratio): the calibration standard's uncertainty should be no more than 25% of the tolerance. If a temperature transmitter must be accurate to ±2°C, the calibration equipment should have an uncertainty no greater than ±0.5°C. When uncertainty is comparable to or larger than the tolerance, the result cannot reliably confirm whether the instrument is within specification, and the user must decide whether to accept the risk, use a more accurate reference standard, or reduce the tolerance requirement. Several Singapore frameworks reference this directly: HSA's GMP guidelines require pharmaceutical measuring equipment to be traceably calibrated with appropriate uncertainty; NEA requires accredited test reports for emissions monitoring; PUB references standards-based measurement for water quality licensees; ISO 9001:2015 clause 7.1.5 requires monitoring and measuring resources to be fit for purpose; and IATF 16949's measurement system analysis (MSA) requirements are closely related to uncertainty analysis.
Common Misunderstandings
A smaller uncertainty always means a better calibration. Uncertainty depends on the reference standard, method and the instrument being calibrated — poor repeatability in the instrument under test will report a large uncertainty even with a highly accurate reference standard. Measurement uncertainty is the same as instrument accuracy. Manufacturer-specified accuracy describes expected performance under defined conditions; uncertainty is an estimate of doubt in the specific result obtained under the actual conditions that prevailed — related but not identical. Only labs with the most advanced equipment can achieve small uncertainties. Reference standard quality matters, but good laboratory practice — controlled environment, experienced technicians, documented procedures — makes a significant difference too; SAC-SINGLAS assessment evaluates the complete uncertainty budget, not just the equipment on the shelf.
Evaluating Uncertainty Information
When selecting a laboratory or reviewing a certificate: is expanded uncertainty stated for each result (or at minimum the worst-case point)? Is the coverage factor and confidence level specified? Is the uncertainty small enough relative to the instrument's intended tolerance to make the result meaningful? Has the laboratory demonstrated its uncertainty budget through proficiency testing? Is it accredited by SAC-SINGLAS or an equivalent ILAC-MRA signatory body? Unitest Instruments provides full uncertainty statements on all certificates issued under SAC-SINGLAS accreditation LA-2023-0845-C. Contact our team or visit our calibration services page to discuss your requirements.