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