Ch. 5. Mixed models and repeated measures: Some illustrative industrial examples

Abstract

Publisher Summary The analysis of data collected from most experiments in the industrial setting requires models that involve more than one variance component. Such models need a mixed models approach to provide an adequate analysis. Data from randomized complete block designs, incomplete block designs, split-plot designs, strip-plot designs, and repeated measures designs require models with more than one variance component and having fixed effects parameters for factorial effects and regression coefficients. This chapter uses a set of examples to demonstrate the use of mixed models analysis to address the analysis of data. The major features of these examples are the identification of the various sizes of experimental units and corresponding error terms and then being able to describe the ways in which the respective error sums of squares are computed. With the development of software such as Proc Mixed, such models can be fit to the data and are thus used to provide more appropriate analyses of data sets.