The quality of a measurement cannot be described adequately by a single characteristic. A measurement system may produce closely grouped results that are consistently displaced from a reference value, or it may produce widely scattered results whose average is close to the reference. The terms accuracy, precision, trueness and repeatability describe different features of these situations.
Measurement accuracy refers to the closeness of agreement between a measured quantity value and a true quantity value. Since a true value is generally not known exactly, accuracy is usually assessed through suitable reference values, calibration evidence and measurement uncertainty. In metrology, accuracy is a qualitative concept and should not be treated as a quantity that always has one universal numerical formula.
Measurement trueness refers to the closeness of agreement between the average of an infinite number of replicate measured quantity values and a reference quantity value. In practice, a finite set of measurements is used to estimate this behavior. A measurement system with good trueness has little systematic bias relative to the reference under the stated conditions. Trueness is therefore closely associated with systematic measurement effects.
Measurement precision describes the closeness of agreement among measured quantity values obtained by replicate measurements under specified conditions. Precision is associated with the spread of results. It is often evaluated using standard deviation, variance, range or another appropriate measure of dispersion. High precision does not guarantee high accuracy because all readings can be closely grouped around a biased value.
Repeatability is measurement precision under a set of repeatability conditions. These generally include the same measurement procedure, operators, measuring system, operating conditions and location, with replicate measurements performed over a short period. Repeatability is therefore a specific type of precision rather than a synonym for all precision.
A measurement system can have good repeatability but poor reproducibility if results change substantially when operators, laboratories, equipment or other conditions change. Repeatability focuses on closely controlled conditions, while reproducibility concerns broader changes in specified conditions.
A useful conceptual model separates the central tendency of observations from their spread. Trueness is related to how the mean compares with a reference value, while precision is related to how widely observations are dispersed. Accuracy considers overall agreement with the reference, taking account of both systematic and random influences.
These concepts are often explained using a target analogy. A tightly grouped cluster at the target centre represents good precision and good trueness. A tight cluster away from the centre represents good precision but poor trueness. A broad distribution centred near the target represents relatively good trueness but poor precision. A broad distribution displaced from the centre indicates both poor precision and poor trueness. The analogy is useful, but actual measurement quality must be evaluated using a defined measurand, measurement model, reference and uncertainty assessment.
Accuracy, Precision, Trueness and Repeatability
Accuracy, precision, trueness and repeatability describe different aspects of measurement quality. Accuracy concerns agreement with a true or reference quantity value, trueness concerns agreement between the average of many repeated measurements and a reference value, precision concerns agreement among measured values, and repeatability describes precision under specified conditions that are kept substantially unchanged. Understanding these distinctions is essential for evaluating measurement systems and interpreting experimental results correctly.
Arithmetic mean: x̄ = (1/n)Σxᵢ
Sample standard deviation: s = √[Σ(xᵢ − x̄)²/(n − 1)]
Estimated bias: b̂ = x̄ − xᵣ
Absolute estimated bias: |b̂| = |x̄ − xᵣ|
Relative standard deviation: RSD = (s/|x̄|) × 100%, for x̄ ≠ 0.
A smaller standard deviation indicates greater precision under the conditions represented by the data. A smaller estimated bias indicates closer agreement of the mean with the reference, subject to reference uncertainty and the validity of the measurement model.
Explanation
Variables
| Symbol | Meaning | Unit |
|---|---|---|
xᵢ | The ith measured value | Unit of the measurand |
x̄ | Arithmetic mean of repeated measurements | Unit of the measurand |
xᵣ | Reference quantity value | Unit of the measurand |
n | Number of observations | Dimensionless count |
s | Sample standard deviation | Unit of the measurand |
b̂ | Estimated bias of the mean relative to the reference value | Unit of the measurand |
RSD | Relative standard deviation | % |
Example or application
A worked example has not yet been added.
Definition
Accuracy
Closeness of agreement between a measured quantity value and a true quantity value. It is a qualitative concept and is not itself assigned a universal numerical value.
Precision
Closeness of agreement among measured quantity values obtained by replicate measurements under specified conditions.
Trueness
Closeness of agreement between the average of an infinite number of replicate measured quantity values and a reference quantity value.
Repeatability
Measurement precision under a set of repeatability conditions, including substantially unchanged procedure, operator, measuring system, location and operating conditions over a short period.
Purpose
The purpose of distinguishing these measurement-quality characteristics is to determine whether a measurement system is closely grouped, correctly centred relative to a reference, stable under repeated conditions and suitable for its intended use.
Importance
These distinctions prevent misleading claims about instrument quality. An instrument that gives highly consistent readings may still be biased, while an instrument with an average near the reference may produce too much variation for a particular application. Separating the concepts supports calibration, validation, quality control and experimental interpretation.
Conceptual Comparison
| Concept | Main Question | Main Concern | Typical Evaluation |
|---|---|---|---|
Accuracy | How closely does the measurement agree with the true or reference value? | Overall agreement | Reference comparison and uncertainty assessment |
Precision | How close are repeated measurements to one another? | Spread or dispersion | Standard deviation, variance or another dispersion measure |
Trueness | How closely does the average of repeated measurements agree with the reference? | Systematic bias | Comparison of an estimated mean with a suitable reference |
Repeatability | How close are repeated results under substantially unchanged measurement conditions? | Precision under repeatability conditions | Replicate measurements using the same procedure and measurement system under specified conditions |
Relationship Between Characteristics
Good Precision Good Trueness
Results are closely grouped and their average is close to the reference value. This generally supports good measurement accuracy.
Good Precision Poor Trueness
Results are tightly grouped but consistently displaced from the reference. A systematic effect may be present.
Poor Precision Good Trueness
Results are widely dispersed, but their average may be close to the reference. Random variation may be substantial.
Poor Precision Poor Trueness
Results are widely dispersed and their average is displaced from the reference. Both random and systematic effects may be important.
Measurement Conditions
Repeatability Conditions
Same measurement procedure
Same operator where applicable
Same measuring system
Same operating conditions
Same location
Replicate measurements over a short period
Explanation
The conditions must be stated sufficiently clearly to make a repeatability claim meaningful. If important conditions change, the resulting precision may no longer represent repeatability.
Statistical Evaluation
Mean
The arithmetic mean summarizes the central value of a set of observations. It is useful for estimating the average result when the data and measurement conditions justify its use.
Standard Deviation
The sample standard deviation describes the spread of observations around their mean. It is a common measure of precision for suitable quantitative data.
Variance
The variance is the square of the standard deviation and describes dispersion in squared units.
Relative Standard Deviation
The relative standard deviation expresses the standard deviation relative to the magnitude of the mean. It is useful when comparing variability across measurements with different scales, provided the mean is meaningfully different from zero.
Reference Comparison
The mean can be compared with a reference value to estimate bias. The uncertainty of both the measured mean and the reference should be considered when interpreting the difference.
Worked Examples
| Title | Problem | Solution | Interpretation |
|---|---|---|---|
High precision with poor trueness | A reference mass is 100.00 g. A balance gives 100.40 g, 100.41 g, 100.39 g and 100.40 g. | The observations are tightly grouped around approximately 100.40 g, but their mean is about 0.40 g above the reference value. | The measurements show good repeatability and relatively high precision, but an apparent positive bias indicates poor trueness relative to the reference. |
Estimating precision | Repeated measurements of a length are 10.01 mm, 9.99 mm, 10.00 mm, 10.02 mm and 9.98 mm. | The mean is 10.00 mm. The sample standard deviation is approximately 0.0158 mm. | The small spread indicates relatively high precision for the conditions under which the measurements were obtained. |
Why a correct mean is not enough | Two methods produce the same mean value of 50.0 units. Method A has a standard deviation of 0.1 units, while Method B has a standard deviation of 2.0 units. | Both methods have the same reported mean, but Method A produces much less variation among repeated measurements. | The same average does not imply the same precision. The suitability of either method also depends on bias, uncertainty and intended measurement requirements. |
Common Misconceptions
Accuracy and precision are not interchangeable terms.
High precision does not guarantee high accuracy.
Trueness concerns agreement of the average with a reference, not merely the closeness of individual readings to one another.
Repeatability is precision under specified repeatability conditions, not every possible form of precision.
A small standard deviation does not prove that an instrument is correctly calibrated.
Accuracy is not universally calculated as 100% minus percentage error.
A measurement system should be evaluated against the requirements of its intended application.
Applications
Comparing laboratory measurement methods
Evaluating balances and dimensional instruments
Manufacturing process quality control
Calibration and verification
Medical diagnostic measurement systems
Sensor performance evaluation
Scientific experiments and research
Interlaboratory measurement comparisons
Related Topics
Introduction to Measurement Errors
Systematic, Random and Gross Errors
Reproducibility and Intermediate Precision
Absolute, Relative and Percentage Error
Measurement Uncertainty
Reference
| Title | Organization | Url |
|---|---|---|
International Vocabulary of Metrology (VIM), JCGM 200 | Joint Committee for Guides in Metrology | https://www.bipm.org/en/committees/jc/jcgm/publications |
NIST/SEMATECH e-Handbook of Statistical Methods | National Institute of Standards and Technology | https://www.itl.nist.gov/div898/handbook/ |
Related Topic
SCI-REF-003-073
SCI-REF-003-074
SCI-REF-003-077
SCI-REF-003-078
SCI-REF-003-080
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