Optimal unbiased estimation of variance components (Q1083160)

This manuscript deals with the problems arising in obtaining optimal unbiased estimates of variance components in the general mixed model. The author provides the reader with the necessary background material and the historical development of the subject matter. The material is divided into ten chapters. Chapter 1 is set aside for the matrix formulation of the general mixed model and for a discussion of selecting appropriate criteria along with their properties for obtaining estimates of the variance components. The next chapter deals with linear and matrix methods which prove valuable in simplifying work in later chapters. After discussing linearization techniques for the basic model in chapter 3, the author introduces the concept of 'estimable function' in chapter 4, along with a discussion of the ordinary least squares estimates and their relationship to other variance component estimation methods. Optimal unbiased estimates in certain general models are given in chapter 5. The next chapter is concerned with the problem of obtaining estimates of the variance components which are unbiased translation invariant, and optimal in the sense of having the least variance. Chapters 7 and 8 include the basic material on rings and algebras for later use for the structure theory of variance component estimation. The algebraic techniques developed in chapters 7 and 8 are applied in chapter 9 to the variance component problems. In the last chapter, structure theory is applied to non-negative estimation, and to the Jordan decomposition of any optimal estimate into independent quadratic estimates. In the concluding remarks some directions for new work in this area are provided. The manuscript is written with great clarity, and includes in it a good amount of useful and valuable information on optimal unbiased estimation of variance components. It is going to be valuable to researchers in this area. It includes in it an adequate list of references, but contains no index set.