Linear Estimation and Design of Experiments : Joshi, D.D

Linear estimation and design of experiments planned as a text-book for students, and not as a manual for practising statisticians, the book covers the General theory of linear estimation and its application to the analysis of data in fixed-effects models. It offers an integrated approach placing proportionate emphasis on mathematical theory, its application to standard designs, and computational techniques. What distinguishes this book from other texts on experimental designs is the use throughout of R.C. Boses approach to the theory of least squares, based on the concept of estimable functions. Among the many advantages of this approach are the following: the use of side conditions in the solution of normal equations is dispensed with, sums of squares and degrees of freedom are defined unambiguously, the distributions of sums of squares are obtained directly without appealing to cochrans theorem, hypotheses which can be tested are separated from those which cannot, the role of interactions to test for factor additivity in factorial experiments is clearly brought out. The book will be of use to Honours and postgraduate students of statistics.