Measuring and maximizing precision in analyses based on use of calibration graphs

Area of Science:

• Analytical Chemistry
• Statistical Methods in Chemistry

Background:

• Traditional calibration procedures in chemical analysis can lead to a loss of precision.
• Existing measures of precision (e.g., correlation coefficient, relative standard deviation, detection limit) may not be sufficiently informative.

Purpose of the Study:

• To introduce and advocate for the use of confidence-band statistics in chemical analysis.
• To replace current calibration methods with more precise alternatives.
• To establish improved parameters for assessing analytical precision.

Main Methods:

• Application of confidence-band statistics for precision measurement.
• Implementation of weighted least-squares and multiple-curve calibration techniques.
• Development of a general scheme for routine analytical method development.

Main Results:

• Confidence-band statistics provide a robust basis for measuring and enhancing analytical precision.
• Weighted least-squares and multiple-curve calibration techniques offer superior precision compared to current methods.
• New parameters derived from confidence-band statistics offer more meaningful insights into analytical performance.

Conclusions:

• Confidence-band statistics represent a significant advancement for improving precision in chemical analysis.
• The proposed techniques and parameters are suitable for the development of robust routine analytical methods.