OPTIMIZATION OF GAMMA-RAY SPECTROMETRIC AND OTHER MEASUREMENTS YIELDING BIASED RESULTS BY MEANS OF INFORMATION-THEORY

Analytical methods and procedures yielding biased results can be effectively optimized with Information quantities like information gain and profitability of the results. This possibility has been studied on a simple γ-countlng experiment. We have supposed that the differences in the height of the samples counted are the main source of a measurement bias. Both the bias and precision of such simple measurements can be calculated In a theoretical way by the program geo developed for this purpose. The program also calculates the values of information quantities like gain and profitability of the results to find the optimum conditions for various experimental arrangements such as constant counting time or counting error. The dependence of the Information gain and profitability on the measurement parameters, especially the bias and precision, shows a critical influence of the measurement bias on the Information content of the results. Many conclusions derived for this simple case can be adopted for analytical measurements in general. © 1990, American Chemical Society. All rights reserved.