A theoretical evaluation of linearity

Abstract

The measure of linearity is an important part of the evaluation of a method. According to the NCCLS guidelines (Document EP6-P), results of a linearity experiment are fit to a straight line and judged linear either by visual evaluation, which is subjective, or by the lack-of-fit test. This approach depends on the precision of the method, is not necessarily conclusive, and fails to be quantitative. We define linearity as a measure of how well a first-order (linear) polynomial fits the data compared with a higher-order (nonlinear) polynomial. The major property of a linear polynomial is that the first derivative is a constant. The nonlinearity of a method can be measured by the difference between these two polynomials (first-order and higher-order) at specific values or, as an average, the root-mean difference. This approach is independent of the precision of the assay and is conclusive, quantitative, and objective.