Uncertainty of qualitative chemical analysis: General methodology and binary test methods

Some general problems of qualitative chemical analysis are considered. In particular, the context of its development is outlined, the concept of hypothesis testing as a basis for the detection and identification of chemical substances is presented, a classification of the types of analysis is given, and the uncertainty (reliability, accuracy) of qualitative determinations is discussed. Various errors arising from the use of binary test methods are reviewed, and methods for estimating these errors are surveyed. The estimates are based on the false (or true) rates of positive and negative results. The concentration dependence of these rates is discussed. It is shown that the testing results become more reliable if the prior probability of the presence of the analyte in the sample is known. Another method for improving the validity of analysis is to use two test methods or to confirm the testing data by other analytical means. In these cases, the positive and negative predictive values are estimated using Bayesian statistic. The approaches to the estimation of the uncertainty of test methods widely used in medicine and toxicology are extended to qualitative chemical analysis as a whole.