A stochastic approximation method for assigning values to calibrators.

A new procedure is provided for transferring analyte concentration values from a reference material to production calibrators. This method is robust to calibration curve-fitting errors and can be accomplished using only one instrument and one set of reagents. An easily implemented stochastic approximation algorithm iteratively finds the appropriate analyte level of a standard prepared from a reference material that will yield the same average signal response as the new production calibrator. Alternatively, a production bulk calibrator material can be iteratively adjusted to give the same average signal response as some prespecified, fixed reference standard. In either case, the outputted value assignment of the production calibrator is the analyte concentration of the reference standard in the final iteration of the algorithm. Sample sizes are statistically determined as functions of known within-run signal response precisions and user-specified accuracy tolerances.