Combined standard uncertainty describes the standard-deviation-like uncertainty of a measurement result. In many applications, it is useful to report a wider interval that covers a substantial portion of the distribution of values reasonably attributable to the measurand. Expanded uncertainty is calculated by multiplying combined standard uncertainty by a coverage factor. The coverage factor depends on the desired coverage probability, the assumed probability distribution and the available degrees of freedom. A factor near 2 is frequently used when the distribution is approximately normal and the available information supports that approximation. However, k = 2 does not guarantee exactly 95% coverage in every situation. When the uncertainty of the standard uncertainty estimate is significant, a Student's t-distribution and effective degrees of freedom may be used to select an appropriate coverage factor. A confidence interval is a statistical interval constructed using a procedure that has a specified long-run coverage probability under its assumptions. In frequentist statistics, a 95% confidence procedure means that 95% of intervals produced by repeated use of the procedure would contain the fixed parameter. It does not mean that the particular calculated interval has a 95% probability of containing a fixed parameter. Expanded uncertainty and confidence intervals are related concepts but are not interchangeable in every context. The reporting method should identify the coverage factor, the interpretation of the interval and any relevant distributional assumptions.
Expanded Uncertainty, Coverage Factor and Confidence Intervals
Expanded uncertainty expresses measurement uncertainty as a wider interval obtained by multiplying combined standard uncertainty by a coverage factor. Coverage factors and confidence intervals help communicate the range and probability interpretation associated with statistical estimates, provided their underlying assumptions are clearly stated.
U = k u_c(y). An approximate confidence interval for a mean with unknown population standard deviation is x̄ ± t_(1−α/2,ν) s/√n, under the applicable statistical assumptions.
Explanation
Variables
| Symbol | Meaning | Si Unit |
|---|---|---|
U | Expanded uncertainty | Same unit as the measurement result |
k | Coverage factor | Dimensionless |
u_c(y) | Combined standard uncertainty | Same unit as y |
x̄ | Sample mean | Same unit as observations |
s | Sample standard deviation | Same unit as observations |
ν | Degrees of freedom | Dimensionless |
α | Significance level associated with the confidence procedure | Dimensionless |
Example or application
A worked example has not yet been added.
Definition
Expanded uncertainty is a quantity defining an interval about a measurement result that may be expected to encompass a large fraction of the distribution of values reasonably attributable to the measurand. It is obtained by multiplying combined standard uncertainty by a coverage factor.
Purpose
To communicate measurement uncertainty using an interval wider than the standard uncertainty and to support statistically justified interpretations of measurement results.
Importance
Expanded uncertainty is widely used in calibration certificates, laboratory reports, scientific publications and conformity assessments. Correct interpretation prevents misleading claims about the reliability of measurement results.
Standard Vs Expanded Uncertainty
Standard Uncertainty
Expresses uncertainty in the form of a standard deviation.
Expanded Uncertainty
Multiplies combined standard uncertainty by a selected coverage factor.
Relationship
Expanded uncertainty is not a separate physical error. It is a reporting form derived from combined standard uncertainty.
Coverage Factor
Description
A numerical multiplier used to obtain expanded uncertainty from combined standard uncertainty.
Selection Factors
Desired coverage probability.
Shape of the output probability distribution.
Effective degrees of freedom.
Reliability of the uncertainty evaluation.
Applicable standards or reporting requirements.
Common Illustrations
| Factor | Interpretation |
|---|---|
k ≈ 1 | For a normal distribution, an interval of approximately one standard deviation around the mean covers about 68% of the distribution. |
k ≈ 2 | For a normal distribution, an interval of approximately two standard deviations covers about 95.45% of the distribution. |
k ≈ 3 | For a normal distribution, an interval of approximately three standard deviations covers about 99.73% of the distribution. |
Important Note
These percentages are illustrations for a normal distribution. They must not be assumed for every distribution or uncertainty evaluation.
Confidence Intervals
Description
A confidence interval is an interval calculated using a statistical procedure designed to achieve a specified long-run coverage probability.
Confidence Level
A 95% confidence level describes the long-run performance of the interval-generating procedure under its assumptions.
Student TDistribution
The Student's t-distribution is commonly used for confidence intervals for a population mean when the population standard deviation is unknown and the sample is suitably modeled.
Normal Approximation
For sufficiently large samples or when the population standard deviation is known under appropriate assumptions, normal-based intervals may be suitable.
Distinction
A confidence interval for a population parameter is not automatically identical to an expanded uncertainty interval for a measurement result.
Worked Example
Problem
A measurement result is 50.00 mm, and its combined standard uncertainty is 0.20 mm. Use a coverage factor of 2.
Calculation
U = 2 × 0.20 = 0.40 mm
Result
50.00 mm ± 0.40 mm
Interpretation
The expanded uncertainty is 0.40 mm. The intended coverage interpretation must be supported by the probability model and uncertainty evaluation.
Degrees Of Freedom And Welch Satterthwaite
Description
When multiple uncertainty contributions have finite degrees of freedom, the effective degrees of freedom can be estimated to select a coverage factor.
Formula
ν_eff = u_c⁴(y) / Σ[uᵢ⁴(y)/νᵢ]
Interpretation
The approximation accounts for the relative sizes of uncertainty contributions and their degrees of freedom.
Limitation
The formula is applicable under its assumptions and should not be used without considering the statistical model and the nature of the contributions.
Reporting Requirements
State the measurement result and unit.
State the expanded uncertainty.
Provide the coverage factor.
Describe the intended coverage interpretation where appropriate.
Identify relevant distributional and degrees-of-freedom assumptions.
Avoid claiming a guaranteed maximum error unless that claim is separately justified.
Common Misconceptions
A coverage factor of 2 does not always guarantee exactly 95% coverage.
Expanded uncertainty is not the same as maximum permissible error.
A 95% confidence interval does not mean there is a 95% frequentist probability that a fixed parameter lies in the already calculated interval.
A wider uncertainty interval does not automatically indicate poor experimental technique; it may reflect a more complete evaluation.
Coverage factors should not be selected without considering the output distribution.
Reference
JCGM 100:2008, Evaluation of Measurement Data — Guide to the Expression of Uncertainty in Measurement (GUM).
NIST Technical Note 1297, Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results.
Related Topic
SCI-REF-003-081
SCI-REF-003-084
SCI-REF-003-086
SCI-REF-003-088
Sources and verification
Sources are shown as provided in the project data. Not every source has yet been independently verified. Please report errors or missing citations.
- Source verification pending; check an authoritative textbook, university, standards body, or peer-reviewed source.
Suggest a correction or source · More in Units & Measurement