Accuracy in science describes the closeness of a measured quantity value to a reference value, accepted value or, in idealised terminology, the true quantity value. It is therefore concerned with systematic agreement rather than merely the consistency of repeated observations. The concept is important in measurement science because an instrument can produce highly consistent results that are nevertheless displaced from the reference value.
Accuracy is not normally treated as a quantity that can be assigned a universal numerical scale in the same way as length or mass. Instead, measurement results can be described as having greater or lesser accuracy relative to a defined reference. The reference must itself be appropriately established because an inaccurate reference can make an apparently accurate comparison misleading. Accuracy is consequently dependent on the measurement context, the measurand and the reference against which agreement is evaluated.
Accuracy is closely connected with measurement error and bias. If a measurement result differs from a reference value, the difference represents measurement error when the reference is sufficiently well established for that comparison. A systematic measurement error can shift results consistently in one direction, reducing accuracy even when repeated observations have little spread. Calibration is one of the principal methods for investigating and controlling such effects.
During calibration, instrument indications are compared with values provided by standards or reference systems with known characteristics. A correction can sometimes be derived and applied. Calibration does not necessarily remove all uncertainty because the reference, comparison procedure and correction itself have finite uncertainty. Accuracy therefore cannot be established simply by declaring that an instrument has been calibrated.
Accuracy should be distinguished carefully from precision. Precision concerns the closeness of agreement among repeated measured values, whereas accuracy concerns agreement with a reference. Four broad situations illustrate the distinction. Measurements can be both accurate and precise when they cluster closely around the reference. They can be precise but inaccurate when they cluster tightly around a biased value. They can be relatively accurate on average but imprecise when individual observations vary widely around the reference.
They can be both inaccurate and imprecise when measurements are widely scattered and displaced from the reference. These concepts are sometimes represented using target diagrams, but the underlying distinction is metrological rather than merely visual. Scientific reporting should avoid using precision as a synonym for accuracy because doing so can obscure systematic bias.
Accuracy depends partly on how the measurement system has been designed and calibrated. Instrument geometry, sensor characteristics, environmental sensitivity, reference standards, software algorithms and sample handling can all influence systematic deviations. Temperature can change instrument response or physical dimensions. Pressure can affect mechanical or fluid measurements. Electromagnetic interference can influence electronic sensors. Chemical matrix effects can bias analytical measurements. Biological assays can exhibit cross reactivity or systematic differences between sample types. Detecting such effects may require reference materials, interlaboratory comparisons, control samples, alternative measurement methods or calibration across the relevant measurement range. Accuracy is therefore not simply an intrinsic property of a physical instrument; it is a property of the measurement process under specified conditions.
Accuracy has practical importance whenever measurements are compared with thresholds, standards or physical laws. In manufacturing, dimensional accuracy determines whether components meet design specifications. In navigation, accurate position measurements affect location estimates. In medicine, inaccurate measurements can affect clinical classification or treatment decisions. In environmental monitoring, systematic sensor bias can distort estimates of pollutant concentration or temperature trends. In physics, inaccurate calibration can create apparent discrepancies between independent measurements. The significance of an accuracy problem depends on the application. A small bias may be irrelevant when only broad trends are needed but critical when measurements are used to distinguish quantities that differ by a small amount. Scientific measurement therefore requires accuracy appropriate to the intended purpose rather than maximal accuracy in every situation.
Accuracy can be evaluated using reference materials, certified reference values, calibration standards, comparison measurements and independent measurement methods. In analytical chemistry, a known reference material can test whether a method recovers a value close to the certified concentration. In dimensional metrology, calibrated artefacts can test instrument response. In temperature measurement, traceable reference standards can be used across relevant temperature ranges. Method comparison can reveal systematic differences between techniques.
Traceability means that a measurement result can be related to a reference through a documented, unbroken chain of calibrations, each contributing to measurement uncertainty. Traceability does not mean that a result is automatically accurate; it establishes a structured relationship with a reference. The quality of that reference chain and the associated uncertainty remain essential.
A common misconception is that increasing the number of measurements necessarily increases accuracy. Repetition primarily improves knowledge of random variation and can improve the precision of an estimated mean. It does not automatically remove a systematic bias. If a scale consistently reads 2 grams too high, weighing an object one hundred times may produce an extremely precise estimate that remains inaccurate. Detecting and correcting systematic effects requires calibration, controls or independent evidence. Another misconception is that a more expensive instrument is always more accurate. Instrument suitability depends on calibration, environmental conditions, operating range, sample characteristics and measurement procedure. A sophisticated instrument used incorrectly can produce worse results than a simpler instrument operated under controlled conditions.
Accuracy is therefore best understood as a contextual relationship between a measurement result and an appropriate reference rather than as a synonym for precision or reliability. Accurate measurement requires suitable standards, appropriate calibration, controlled procedures, a clearly defined measurand and consideration of uncertainty. Accuracy cannot normally be proven absolutely because the true value of a physical quantity is often not known with infinite certainty. Instead, scientists gather evidence that the measurement process agrees sufficiently with well established references.
High quality science combines accuracy with precision, uncertainty evaluation, reproducibility and appropriate experimental design. The objective is not merely to obtain numbers that look consistent but to obtain measurements that are appropriately close to the quantity being investigated and to understand the limits of that closeness.
