Learning Objective
Distinguish accuracy, precision and reliability, interpret repeated measurements and identify factors that influence confidence in instrument results.
Definition
Accuracy describes the closeness of a measured value to a reference value or accepted quantity value. Precision describes the closeness of agreement among repeated measurements made under specified conditions. Reliability is the ability of an instrument or measurement system to perform consistently as required over time and under stated conditions.
Scientific Background
Measurement quality cannot be described by a single number or specification. Repeated readings may cluster closely together but remain offset from a reference, indicating good precision but poor accuracy. A system may initially perform well but become unreliable if it drifts, fails intermittently or is not maintained. Calibration, uncertainty evaluation, repeatability studies and performance monitoring provide complementary information.
Key Concepts
| Term | Explanation |
|---|---|
Accuracy | Closeness of agreement between a measured value and a reference value. It is a qualitative concept and should not be confused with precision. |
Precision | Closeness of agreement among repeated measurements under specified conditions. |
Repeatability | Precision under repeatability conditions, such as the same method, operator, instrument, location and a short time interval. |
Reproducibility | Precision under changed conditions, such as different laboratories, operators or instruments, as defined by the study. |
Reliability | The ability to maintain required performance over time, including consistency and dependable operation. |
Measurement uncertainty | A non-negative parameter characterizing the dispersion of quantity values attributed to a measurand, based on the information used. |
Comparison
| Criterion | Accuracy | Precision | Reliability |
|---|---|---|---|
Main question | How close is the result to a reference? | How closely do repeated results agree? | Does the system continue to perform as required? |
Typical evidence | Comparison with a suitable reference or standard | Repeated measurement statistics | Performance history, stability and failure records |
Possible problem | Systematic bias | Random variation | Drift, intermittent faults or degradation |
Worked Example
Problem
A reference mass is 100.00 g. Instrument A repeatedly reads 100.01 g, 100.00 g and 100.01 g. Instrument B reads 99.70 g, 100.20 g and 100.10 g. Compare their observed behavior.
Analysis
Instrument A readings are closely grouped and close to the reference, suggesting good precision and close agreement in this small example.
Instrument B readings are more widely spread, indicating lower precision; some readings are also farther from the reference.
A full evaluation would require a suitable measurement procedure, calibration information and uncertainty assessment.
Conclusion
Instrument A shows better observed agreement in this illustrative dataset, but a few readings alone do not establish long-term reliability.
Performance Factors
Calibration status and traceability
Sensor stability and drift
Environmental temperature, humidity and vibration
Resolution and noise
Operator technique and sample handling
Mechanical wear and electronic component aging
Maintenance and quality-control history
Improvement Methods
Calibrate with suitable reference standards
Use a validated measurement procedure
Control or record relevant environmental conditions
Perform repeat measurements when appropriate
Monitor drift and instrument performance over time
Maintain equipment and document repairs or adjustments
Common Misconceptions
Precision does not guarantee accuracy.
A single reading cannot establish repeatability.
Reliability is broader than short-term precision.
A calibration certificate does not guarantee correct use in every application.
More decimal places do not prove greater accuracy.
Practical Applications
Comparing balances and dimensional instruments
Evaluating laboratory measurement systems
Monitoring production and process-control instruments
Planning calibration intervals and maintenance
Assessing data quality in scientific research
Knowledge Check
| Question | Answer | Explanation |
|---|---|---|
Can an instrument be precise but inaccurate? | Yes. Repeated readings can agree closely while being offset from the reference value. | This pattern is consistent with a systematic bias. |
What information helps evaluate reliability over time? | Performance records such as drift checks, calibration history, maintenance records and failure history. | Long-term evidence is needed to judge whether an instrument continues to meet its requirements. |