Repeatability, Reproducibility and Coefficient of Variation

Repeatability and reproducibility can only be assessed when replicate measurements by each method are available. If replicate measurements by a method are available, it is simple to estimate the measurement error for a method, using a model with fixed effect of item, and then taking the residual standard deviation as the measurement error standard deviation. The precision of a measurement method is often given as the coefficient of variation (CV). The variation used to estimate the CV based on log-transformed data is the residual variation derived using some model for the mean value. Occasionally, the overall residual standard deviation is just divided by the overall mean and used as estimate of the CV. Controlled Vocabulary Terms coefficient of variation; measurement error; residual variation