Factorial effects, random blocks, and longitudinal data: Two simple analysis methods

In this paper, we introduce two simple methods for parametric linear (mixed) modeling of a class of industrial research problems. These problems concern industrial experimenting where treatments are applied in a blocked factorial design with subjects as blocks and where the data are longitudinal response data. We discuss model estimation and model testing with both methods. The models are fitted to the data from an experiment that was performed in a personal-care research environment and we compare the estimation and testing results. One of the methods is extended to be able to model personal-care experiments more adequately. Finally, suggestions for further research are given.