In eCerto: Statistical Tests for the Production of Reference Materials

Linearity

The linearity of an analytical method is tested using the mean normalized analyte values at different concentrations $\overline{x}_j$.

Analyte values are computed for each replicate $i$ at calibration level $j$ as the peak area ratios of analyte and internal standard $x_{i,j}=\frac{A_{Analyte}}{A_{IS}}$. Here, $\overline{x}_j$ is the mean of the $i=1..n$ replicates at level $j$. The total number of calibration levels is denoted as $N$.

For each analyte, a linear model $y=b_0+b_1 \times x$ over all $\overline{x}_j$ is computed and the following parameters are reported:

the number of calibration levels used in the linear model $N$
the smallest number of replicates within all calibration levels $n$
the probability of error ($alpha$) selected by the user
the result uncertainty $k$ selected by the user and depicted as $1/k$
the coefficients $b_0$ (intercept) and $b_1$ (slope) of the linear model
the limit of detection $\text{LOD}$
the limit of quantification $\text{LOQ}$
the standard error of the estimate $s_{y,x}$
the standard error of the procedure $s_{x0}$ (absolute precision)
the coefficient of variation of the procedure $V_{x0}$ (relative precision)

Additionally, the residuals $e$ of the linear model are tested:

to follow a normal distribution (Kolmogorof-Smirnov Test, $P_{KS,Res}$)
to show a trend, i.e. a drift associated with measurement order (Neumann-Test, $P_{Neu,Res}$)
for the level showing the highest absolute residual to be an outlier (F-Test, $Out_F$)

Note The F-Test is applied sequentially until no further outlier is detected.

For comparison the data is fitted using a quadratic model $y=b_0+b_1 \times x+b_2 \times x^2$. The residuals from both models, the linear and the quadratic one, are compared using a Mandel-Test calculating $P_{Mandel}$.