Opening the book…
Precision describes agreement among repeated measurements under stated conditions. Repeatability concerns a short interval under the same operating conditions. Intermediate precision considers variations within a laboratory, such as different analysts, days, or instruments. Reproducibility concerns differences between laboratories.
A tightly clustered set of results can still be biased. Conversely, a highly variable set can have a mean close to the reference by chance. A close mean alone does not establish that individual measurements are reliable.
The sample standard deviation, written s, describes the spread of observations around their mean. Its units match those of the original measurements. For a sample of n observations, the usual calculation uses n − 1 in the denominator. Population standard deviation is conventionally written σ and uses the population mean and population size; it should not be mislabeled on a sample formula.
Mean: x̄ = (x₁ + x₂ + … + xₙ) / n
Sample standard deviation: s = √[Σ(xᵢ − x̄)² / (n − 1)]
Here, xᵢ is one observed value, x̄ is the sample mean, n is the number of observations, and Σ means to add the indicated terms. At least two observations are needed to calculate this sample standard deviation.
The coefficient of variation (CV) is a measure of relative spread, not the difference between one observation and the mean. For positive ratio-scale measurements, CV is often reported as a percentage. Relative standard deviation (RSD) uses the absolute mean in the denominator.
CV (%) = 100 × s / x̄ for a meaningful positive mean
RSD (%) = 100 × s / |x̄|
CV can be misleading when the mean is near zero or when zero is arbitrary, as on the Celsius temperature scale. A low CV does not establish absence of bias.
For example, measurements of 98, 100, and 102 units have a mean of 100, a sample standard deviation of 2, and a CV of 2%. If the reference value is 100 units, the mean has 0% relative bias. Three results alone do not establish the full performance of a method.
Conceptual illustration of bias and precision. The center is the reference, blue points are repeated results, and the orange cross marks their mean. A mean close to the target does not make widely scattered individual results reliable. These simulated points illustrate concepts, not assay data.
This page titled 15.8: Precision was last modified on Mon, 09 Nov 2026 00:17:22 GMT and is shared under a CC BY-SA 4.0 license and was authored, remixed, and/or curated by Emalee MacKenzie.
The book could not be loaded just now.