How to measure information obtained from biased quantitative analyses

Abstract

It is shown that the use of the divergence measure for expressing the information content of the results of a quantitative analysis is justified only then if we need not to expect the rise of a systematic, i.e., statistically significant mean error. However, unless we can exclude the rise of a systematic error we have to adopt the information measure I(r, p, p0) proposed earlier; then of course every even statistically insignificant mean error has effect upon the information content. It is studied the relationship of this measure to another information measure used in a specific case for expressing the information content of biased results of a quantitative analysis.