What Is Measurement Uncertainty?

Every measurement ever made is imperfect. No matter how advanced the instrument, how skilled the operator, or how controlled the environment, there will always be a residual doubt about the true value of what is being measured. This doubt — when quantified and expressed — is called Measurement Uncertainty (MU).

The International Vocabulary of Metrology (VIM, ISO/IEC Guide 99:2007) defines measurement uncertainty as a “non-negative parameter characterizing the dispersion of the quantity values being attributed to a measurand, based on the information used.” In simpler terms, it tells us the range within which the true value of a measurement most probably lies, along with a stated level of confidence.

As the UK’s National Physical Laboratory (NPL) explains in its Beginner’s Guide to Uncertainty of Measurement (GPG11): uncertainty of measurement is “the doubt that exists about the result of any measurement.” A measurement result is only considered complete when accompanied by a statement of this uncertainty — for example, a voltage of 250.22 V ± 0.85 V at a 95% confidence level. Without this, the result is technically incomplete and potentially misleading.

Why Measurement Uncertainty Matters in Laboratories

The stakes of ignoring or underestimating measurement uncertainty can be severe. Consider a pharmaceutical company testing the concentration of an active ingredient — if measurement uncertainty is not properly evaluated, the product may be released with insufficient or excessive dosage. A component manufacturer working to tight tolerances, or a food testing laboratory assessing contaminant levels against regulatory limits, requires proper uncertainty evaluation to avoid wrongly accepting or rejecting a product.

Measurement uncertainty matters because it:

• Enables valid conformity decisions — ISO/IEC 17025:2017 Clause 7.8.6 requires laboratories to report a statement of conformity along with the decision rule applied, which is only possible when measurement uncertainty is known.
• Protects end-users of measurement data — Customers making critical engineering, clinical, or regulatory decisions need to know the reliability of the data they receive.
• Demonstrates technical competence — NABL accreditation under ISO/IEC 17025:2017 requires laboratories to evaluate and report uncertainty for all calibrations and, where applicable, for testing.
• Supports international recognition — Uncertainty statements traceable to SI units through an unbroken calibration chain are the foundation of mutual recognition under the CIPM MRA and the ILAC Arrangement.
• Reduces risk of wrong decisions — Understanding measurement uncertainty relative to the tolerance band guards against both false acceptance (β-error) and false rejection (α-error) of products.

Error vs. Uncertainty: A Critical Distinction

These two terms are frequently confused, but they are fundamentally different. As NPL’s GPG11 explains:

• Error is the difference between the measured value and the true value of the measurand. Known systematic errors can be corrected for — for example, by applying corrections from a calibration certificate.
• Uncertainty is a quantification of the residual doubt that remains after all known corrections have been applied. It covers both the imperfect correction of systematic effects and the unpredictable spread caused by random effects.

NABL 141 (Issue 04, Feb 2020) reinforces this: applying correction to a measurement result to compensate for systematic effect does not make the result totally error-free. There remains a measure of uncertainty due to incomplete knowledge of the required correction. Even after calibration corrections are applied, uncertainty persists — and must be quantified.

Sources of Measurement Uncertainty

Uncertainty in any measurement arises from many sources simultaneously. NABL 141 and NPL’s GPG11 both provide comprehensive lists. The main sources include:

• The measuring instrument itself — Its resolution, repeatability, drift, and uncertainty as reported in its calibration certificate.
• Reference standards and reference materials — The uncertainty associated with the calibration of the reference standards used, propagated through the traceability chain.
• The measurement method and procedure — Approximations in the method, sampling effects, and procedure-related variability.
• Environmental conditions — Temperature, humidity, vibration, electromagnetic interference, and other ambient factors. NABL 141 cites the example of temperature affecting a micrometer used to measure length.
• The operator — Personal judgment, reading parallax, skill differences, and technique variability.
• The item being tested or calibrated — Its condition, stability, and preparation.
• Sampling uncertainty — Where sampling is part of the measurement process, the representativeness of the sample contributes to overall uncertainty.

The Two Types of Uncertainty Evaluation: Type A and Type B

The internationally accepted framework for evaluating measurement uncertainty is the Guide to the Expression of Uncertainty in Measurement (GUM), published by the Joint Committee for Guides in Metrology (JCGM). NABL 141 is directly based on the GUM. The GUM classifies uncertainty evaluation into two methods:

Type A Evaluation

Type A evaluation involves the statistical analysis of repeated measurements made under defined repeatability conditions. The standard uncertainty is calculated from the experimental standard deviation of the mean of a series of observations. NABL 141 recommends 4 to 10 readings under normal conditions; more readings improve reliability. The degree of freedom for Type A evaluation is (n − 1), where n is the number of observations.

Type B Evaluation

Type B evaluation covers all other means of estimating uncertainty — based on scientific judgment, prior experience, published data, calibration certificates, instrument specifications, or reference data. NABL 141 emphasises that a Type B evaluation can be as reliable as a Type A evaluation. Sources of Type B uncertainty include the uncertainty reported in calibration certificates (typically with k = 2 at ~95% confidence), manufacturer specifications, and standard reference data. The probability distribution assumed for Type B components (normal, rectangular, triangular) determines how the standard uncertainty is derived.

Key Probability Distributions in Uncertainty Estimation

• Normal (Gaussian) distribution — Applies to Type A evaluations and to Type B sources where a coverage factor is stated (e.g., calibration certificates). Standard uncertainty = U/k.
• Rectangular (uniform) distribution — Used when only the bounds of a range are known, with equal probability anywhere within. Examples: digital display resolution, manufacturer accuracy specifications. Standard uncertainty = a/√3, where a is the half-width.
• Triangular distribution — Applied when values near the centre of a range are more probable. Standard uncertainty = a/√6.

The Step-by-Step Uncertainty Estimation Process (GUM Approach)

NABL 141 prescribes a clear, systematic procedure for estimating measurement uncertainty using the GUM (bottom-up) approach:

1. Define the measurand — Clearly identify what quantity is being measured and write the measurement model.
2. Identify all sources of uncertainty — List every factor that could contribute to variability: instrument, method, environment, operator, reference standards, sample.
3. Quantify each component — Estimate the standard uncertainty of each source using Type A or Type B methods with appropriate probability distributions.
4. Calculate the combined standard uncertainty (u,) — Combine all components using root-sum-of-squares (RSS), applying sensitivity coefficients (cᵢ) where the measurement model is non-linear: u,(y) = √Σ[cᵢ · u(xᵢ)]²
5. Calculate the expanded uncertainty (U) — Multiply the combined standard uncertainty by the coverage factor k: U = k × u,. As required by NABL 141 and NABL 174, all laboratories must report uncertainty at a 95% confidence level. For a normal distribution, k = 2 gives approximately 95% confidence.
6. Report the result — Express as: Measured Value ± Expanded Uncertainty (unit), at 95% confidence with k = 2. Example: Voltage = 250.22 V ± 0.85 V (k = 2).

NABL 174: Practical Sample Calculations for Electrical Testing

NABL 174 (Issue 04, Feb 2020) — Sample Calculations for Measurement Uncertainty in Electrical Testing — provides worked examples of the full uncertainty budget process for electrical testing laboratories. It covers voltage, current, power loss in energy meters, MCB tripping characteristics, transformers, and three-phase induction motor power measurements.

For a Short Circuit Test of an MCB at 250.22 V, the uncertainty budget in NABL 174 shows:

• Type A (Repeatability from 5 readings): Standard uncertainty = 0.037 V
• Type B — Calibration certificate of digitizer: Standard uncertainty = 0.352 V (normal distribution, k = 2)
• Type B — Instrument accuracy: Standard uncertainty = 0.231 V (rectangular distribution)
• Type B — Display resolution: Negligible contribution
• Combined Standard Uncertainty (u,): 0.423 V
• Expanded Uncertainty (U) at k = 2, 95%: 0.85 V
• Reported result: Voltage = 250.22 V ± 0.85 V

NABL 174 mandates that all components of importance must be taken into account — including reference standards, method, equipment, environmental conditions, and the properties of the item under test — and that the degree of rigour depends on test method requirements, client requirements, and the narrowness of conformance limits.

NABL 142: Metrological Traceability — The Foundation of Valid Uncertainty

Measurement uncertainty is only meaningful when measurement results are metrologically traceable. NABL 142 (Issue 07, Jan 2021) — Policy on Metrological Traceability of Measurement Results — defines and mandates traceability requirements for all NABL-accredited Conformity Assessment Bodies (CABs), aligned with ILAC P10:07/2020.

Metrological traceability is defined as the “property of a measurement result whereby the result can be related to a reference through a documented unbroken chain of calibrations, each contributing to the measurement uncertainty.” This unbroken chain must ultimately connect to SI units or other internationally recognised references.

Under NABL 142, the acceptable hierarchy for achieving traceability is:

1. National Physical Laboratory (NPL), India — the NMI of India — or any other NMI covered by the CIPM MRA whose services are suitable for the intended use.
2. NABL-accredited or ILAC Arrangement-accredited calibration laboratories — where their accredited scope specifically covers the required calibration.
3. Certified Reference Materials (CRMs) — produced by NMIs listed in the BIPM KCDB, or by accredited Reference Material Producers (RMPs), or with values covered in the JCTLM database for laboratory medicine.

A critical point for laboratory managers: under NABL 142 Clause 4.2, if the calibration of an instrument contributes significantly to overall measurement uncertainty, traceability must be demonstrated. Where a calibration contributes insignificantly, the laboratory must have quantitative evidence to support this claim.

Alternative Approaches for Testing Laboratories: Beyond GUM

While calibration laboratories must follow the GUM approach rigorously, NABL 141 acknowledges that many testing laboratories face practical difficulties with the full bottom-up GUM approach for complex, multi-variable test methods. NABL 141 therefore recognises scientifically valid “top-down” approaches that derive uncertainty from experimental data:

• ISO 21748 — Use of repeatability, reproducibility, and trueness estimates from collaborative studies for uncertainty estimation. Particularly relevant for chemical and food testing laboratories.
• Control chart data — Internal quality control data from Shewhart charts or CUSUM charts can provide practical estimates of combined measurement uncertainty.
• ISO 11352 — Estimation of measurement uncertainty based on validation and quality control data, applicable for water quality testing laboratories.
• NORDTEST NT TR 537 — A widely used framework for environmental testing laboratories, combining within-laboratory reproducibility and bias uncertainty.

ISO/IEC 17025:2017 Requirements: What Laboratories Must Do

• Clause 7.6.1: Calibration laboratories shall evaluate measurement uncertainty for all calibrations. Testing laboratories shall also evaluate uncertainty; where the method prevents rigorous evaluation, an estimation based on understanding of the method is required.
• Clause 7.6.3: All significant uncertainty components shall be taken into account using appropriate methods of analysis.
• Clause 7.8.3: Calibration certificates shall include measurement results and associated measurement uncertainty.
• Clause 7.8.6: When a statement of conformity is made, the decision rule applied — considering measurement uncertainty — shall be documented and reported.
• Clause 6.5: Equipment shall be calibrated with traceability as required by NABL 142.

The Uncertainty Budget: A Practical Tool for Laboratories

Every NABL-accredited laboratory should maintain a documented uncertainty budget for each test or calibration parameter in its scope. An uncertainty budget is a tabulated breakdown of all uncertainty components — their estimates, probability distributions, standard uncertainties, sensitivity coefficients, and contributions to combined uncertainty.

A well-structured uncertainty budget serves several functions:

• Identifies dominant uncertainty sources, helping laboratories prioritise where to invest in equipment, environment, or training.
• Provides a transparent, auditable record for NABL assessors to verify during accreditation assessments.
• Enables the Calibration and Measurement Capability (CMC) — the best achievable uncertainty under best-case conditions — to be established and declared in the NABL scope of accreditation (per NABL 143).
• Supports traceability, because the uncertainty of the reference standard or CRM is always one of the inputs to the uncertainty budget.

How to Reduce Measurement Uncertainty

• Use better-quality, calibrated reference standards — Reference standards with smaller stated uncertainties reduce one of the most significant Type B inputs.
• Improve environmental controls — Temperature, humidity, and vibration controls reduce the contribution of environmental influence quantities.
• Increase the number of repeat readings — More measurements reduce Type A (repeatability) uncertainty. NPL recommends 4–10 readings as a practical minimum.
• Train and standardise operators — Operator-to-operator variability can be reduced through training, standard operating procedures, and proficiency testing.
• Apply calibration corrections — Applying known corrections from calibration certificates reduces systematic error, though the uncertainty of those corrections must still enter the uncertainty budget.
• Participate in Proficiency Testing (PT) — Comparison with other accredited laboratories validates the uncertainty estimate and confirms measurement capability.

What Goes on a Test or Calibration Certificate

Under ISO/IEC 17025:2017 and NABL accreditation requirements, calibration certificates must clearly report:

• The measured result with units
• The expanded uncertainty (U)
• The coverage factor (k) used
• The confidence level (typically 95%)
• The traceability of the reference standards used

For testing laboratories issuing a statement of conformity, the decision rule — describing how measurement uncertainty was taken into account in the conformity decision — must also be stated, per Clause 7.8.6 of ISO/IEC 17025:2017.

Conclusion: Measurement Uncertainty Is Not Optional

Measurement uncertainty is not a bureaucratic checkbox — it is the scientific foundation upon which confident measurement decisions are built. Whether you are a calibration laboratory issuing certificates with stated CMC values, a testing laboratory making conformity decisions against product specifications, or a manufacturer relying on measurement data for quality control, understanding and properly evaluating measurement uncertainty is essential.

NABL 141 provides the methodological framework — based on the internationally accepted GUM — for estimating and expressing uncertainty. NABL 142 ensures that the uncertainty chain is anchored to SI units through metrological traceability. NABL 174 shows, through practical worked examples, exactly how to construct an uncertainty budget for electrical testing. And NPL UK’s GPG11 reminds us that at its core, measurement uncertainty is simply the honest, scientific acknowledgement of the doubt that exists in every measurement we make.

NABL-accredited laboratories that invest in robust measurement uncertainty programmes are not just meeting accreditation requirements — they are building a reputation for technical excellence, earning the confidence of their customers, and contributing to the global network of trustworthy measurement that underpins international trade, safety, and scientific progress.