ABSTRACT
In this chapter, the authors focus their attention on the analysis of measurement data using linear geostatistical models. They provide answers to the following questions. How can they formulate an appropriate geostatistical model for the data? What methods of inference should they use? The authors describe likelihood-based and Bayesian methods of inference for the linear model. The fit of the linear geostatistical model can often be improved by applying a transformation to the response variable. Common choices are a power transformation or a log-transformation. The log-transformation is especially common for applications in the life sciences because many biological processes operate multiplicatively rather than additively. Validation of a statistical model consists of applying diagnostic procedures to check whether the data are compatible with the assumptions incorporated into the model. The authors show how model-based geostatistical prediction relates to the classical geostatistical approach known as kriging.
