Combinatorial Regression for Analytical Chemistry

A current problem in analytical chemistry is to match the uncertainty of several independent sample replicates with the measurement uncertainty associated with first-order models. According to contemporary guidelines, calibration lines for method validations should be prepared with few standards, frequently without any blanks. These suggestions weaken the validity of the statistics, which are meant to address truth and accuracy rather than precision. As was previously shown when looking at the pooled calibration (PoPC) paradigm for method validation, excellent precision does not imply good accuracy. A novel technique known as combinatorial regression (CR) was created to estimate slopes, intercepts, and standard deviations to reduce disagreements regarding the computation of measurement uncertainty in accordance with the calibration line. The related standard deviations of several replicates were discovered to be too large to be applied to uncertainty. Concurrently, the idea of employing the numerical values of concentration residuals as promising estimates of measurement uncertainty arose from the fact that the standard deviations of the IUPAC equations are too small to be used in the computation of uncertainties. High-resolution continuous-source flame atomic absorption spectrometry (HR-CR FAAS), which has been demonstrated to produce inconsistent findings for the analysis of elements Na, Mg, and Ni, was used to evaluate the CR method on the determination of copper. The CR provided coefficients of variations (CVs) that rose with concentrations and number of replicates, favoring analyzing low concentrations in contrast to the IUPAC equations.