ABSTRACT

Summation notation becomes very laborious and sometimes nearly impossible to use when one works with unbalanced xed effects models, random effects models, or mixed effects models. This problem can be solved by using a matrix form representation of the model. This chapter contains a discussion of the construction of the matrix form of the model and a description of how to use the matrices to obtain least squares estimators, to test hypotheses, to compute least squares or population marginal means, and to construct con dence intervals. The concept of estimability is discussed in Section 6.3.