What is the difference between precision and accuracy?
In measurement, accuracy concerns how closely a result agrees with the true or an appropriate reference value. Precision concerns agreement among repeated measurements under specified conditions. Everyday uses of “precision” and “sensitivity” can blur these distinctions. An instrument can repeatedly produce similar results that nevertheless remain far from the reference.
Repeated results and a reference value
1. Why is obtaining the same result insufficient? Imagine a balance displaying 102.0, 102.1 and 102.0 grams in three measurements of a known reference mass. If the reference value is 100.0 grams, the results agree closely with one another but share an offset. This illustrates why repeatability alone does not establish accuracy. The reference used for comparison also has uncertainty. These numbers explain a concept; they are not a conformity assessment for a particular instrument. 2. How do random variation and systematic effects differ? Slight differences between repeated readings can arise from random variation in the environment, sample or reading process. Incorrect zeroing, unsuitable calibration or a stable method effect can instead shift results in a particular direction. Separating these contributions depends on measurement design. A single reading cannot reliably establish the scatter. Repeated measurements provide information about random variation but do not automatically expose every systematic effect. 3. Does averaging correct every error? Independent random effects can be reduced through averaging under appropriate conditions. However, if every measurement is high by the same amount, the average remains high. More repetitions do not remove that common shift. Investigate the measurement method and its relationship to the reference first. Successive measurements can also be dependent; increasing their number may then contribute less new information than expected. Reports should state how independent the repetitions are.
Calibration, uncertainty and appropriate reporting
4. What do display digits mean? A device displaying increments of 0.001 unit does not guarantee accuracy at that level. Display resolution, repeatability and accuracy are different characteristics. Sensitivity is another distinct concept: it describes how much an output changes in response to a change in input. Everyday descriptions can mix these meanings. A technical document should identify the characteristic and conditions rather than merely describing an instrument as “very precise” or “very sensitive.” 5. What do calibration and uncertainty provide? Calibration establishes a relationship between an instrument's indication and reference values under stated conditions. Calibration does not necessarily mean the instrument is automatically adjusted; adjustment can be a separate operation. Measurement uncertainty helps express the dispersion of values that can be attributed to the measured quantity. Contributions from references, environmental conditions, the method and repeated readings are assessed. A result gains meaning through its unit and uncertainty, rather than a long string of digits alone. 6. When is a measurement fit for purpose? A kitchen balance and a measurement used in pharmaceutical production do not have the same requirements. Acceptable uncertainty depends on the comparison and the consequences of the decision. A small difference between results should not be interpreted as a real change without considering their uncertainties. Recording conditions, references, the number of repetitions and any corrections supports interpretation. The useful question is not simply how many digits a device displays, but which decision its result supports and within what limits. The calibration date and the range to which it applies should also be considered when evaluating the measurement result.