Bayesian methods for ion-selective electrodes

With the increasing use of ion-selective electrodes in environmental and health applications, it is important to know the precision of estimated concentrations. A Bayesian model for non-linear calibration is introduced which provides estimates of measurement precision by incorporating uncertainty in calibration parameters and inherent random noise in emf response. The analysis of lead in 17 soil samples demonstrates that large gains in precision are possible when calibrations are extended to include multiple electrodes and standard addition data. The results highlight the need for improved calibration and routine use of standard addition as ion selective electrodes become increasingly popular for demanding, real world applications.