This Wiley Classics Library Edition published seventeen years after the first version has the original material of the six chapters together with two unpublished articles written by the author in 1977 and 1983 on Multiple Imputation (MI). Several new contributions to this area, rapid computational algorithms, acceptance and use of Bayesian techniques, software specially designed for MI, a variety of applications of MI in quite diverse fields and certain theoretical developments on MI are the main reasons for bringing out this edition.
After an introductory chapter of 26 pages, the necessary Bayesian and frequentist statistical theory is presented in chapter 2 (48 p.). In chapter 3 (38 p.) the underlying Bayesian theory is discussed. Chapter 4 (41 p.) evaluates the operating characteristics of the procedures for analyzing multiply-imputed data sets using randomization theory. Chapter 5 (48 p.) and chapter 6 (42 p.), respectively, discuss the cases with ignorable and non-ignorable non-responses. Each chapter has interesting problems. There are about 120 references at the end of the book. There is an author index and a subject index.
In view of the reasons mentioned above for bringing out this edition, the release of this edition is indeed a great help to the survey statisticians. The font used for reprinting the appendices makes it quite hard to read them and forces the reader to consult the originals. Most of the earlier typographical errors seem to have been corrected except perhaps for a couple or so, such as Pathak, D.K. (read P.K) and Smith, T.W.F. (read T.M.F) in the references and author index. I would have preferred an updated list of References at the end.