Lack-of-fit testing of ion chromatographic calibration curves with inexact replicates

Calibration studies involve the preparation and analysis of replicates for multiple concentrations of standards. Curves that are fitted through the data are evaluated for their adequacy of fit. A helpful test is a lack-of-fit procedure, which is performed easily by most statistical software. When coupled with R adj 2, the procedure differentiates between data that are not linear and those that are simply noisy. The test requires data from exact replicates of the various standard levels involved. However, in ppt-level ion chromatography, the above condition may be impossible to meet. With the common anions (e.g., chloride, nitrite), the working standards must be prepared by mass and all liquids must be poured; transfer pipets contaminate at these concentrations. Since it is virtually impossible to pour out the desired mass exactly, final concentrations will vary slightly. Consequently, a different approach is needed for lack-of-fit testing. This paper discusses reasonable alternatives and applies them to actual data.