ABSTRACT
The problems considered in this book (described in Chapter 1) have their origin in practice, as in reality the usual assumptions related to the distribution of the response variable and the random errors have to be relaxed and generalized to include some specific pattern in the correlation structure of errors. It is assumed that the response function can be represented by a polynomial of degree one or two. All the described results related to robust rotatability, slope-rotatability, optimality etc., are only for first-and second-order response models. Similar study can be made for higher order polynomials also. Some results on third order models have been communicated which are not included in this book.
