Abstract

This statistical note describes a practical approach to handling the problem of correlated errors when calibrating linear and nonlinear models. The reader will find a detailed description of the parameter estimation method appropriate when there is a need to consider the effect of correlation among residuals (repeated measurements, longitudinal observations) . The presented approach is not model-specific. The procedure and conclusions, however, are based on practical experience with calibrating the stem-profile model derived by Zakrzewski (1995)