Calibration Experiment (Quantitative) • The calibration of the candidate method relates the response of an instrument to the concentration of an analyte. The relationship between concentration and response can be linear or non‐linear (curvilinear). • Determine the dose response curve (instrument response versus concentration) using at least six concentrations (including a blank) over the concentration range of interest. Test at least four replicates at each concentration. • Fit calibration curve with an appropriate regression model. Plot residuals vs. concentration and examine for non‐random patterns. Residuals should be at <15% for each concentration but can be at 20% for the lowest nonzero concentration. If a non‐random pattern is observed, use a different calibration model.
Made with FlippingBook Learn more on our blog