Theory Of Point Estimation

This second, much enlarged edition by Lehmann and Casella of Lehmann's classic text on point estimation maintains the outlook and general style of the first edition. All of the topics are updated, while an entirely new chapter on Bayesian and hierarchical Bayesian approaches is provided, and there is much new material on simultaneous estimation. Each chapter concludes with a Notes section which contains suggestions for further study. This is a companion volume to the second edition of Lehmann's "Testing Statistical Hypotheses".

This book is concerned with point estimation in Euclidean sample spaces. The first four chapters deal with exact (small-sample) theory, and their approach and organization parallel those of the companion volume, Testing Statistical Hypotheses (TSH). Optimal estimators are derived accord ing to criteria such as unbiasedness, equivariance, and minimaxity, and the material is organized around these criteria. The principal applications are to exponential and group families, and the systematic discussion of the rich body of (relatively simple) statistical problems that fall under these headings constitutes a second major theme of the book. A theory of much wider applicability is obtained by adopting a large sample approach. The last two chapters are therefore devoted to large-sam ple theory, with Chapter 5 providing a fairly elementary introduction to asymptotic concepts and tools. Chapter 6 establishes the asymptotic ef ficiency, in sufficiently regular cases, of maximum likelihood and related estimators, and of Bayes estimators, and presents a brief introduction to the local asymptotic optimality theory of Hajek and LeCam. Even in these two chapters, however, attention is restricted to Euclidean sample spaces, so that estimation in sequential analysis, stochastic processes, and function spaces, in particular, is not covered.

This book focuses on the meaning of statistical inference and estimation. Statistical inference is concerned with the problems of estimation of population parameters and testing hypotheses. Primarily aimed at undergraduate and postgraduate students of statistics, the book is also useful to professionals and researchers in statistical, medical, social and other disciplines. It discusses current methodological techniques used in statistics and related interdisciplinary areas. Every concept is supported with relevant research examples to help readers to find the most suitable application. Statistical tools have been presented by using real-life examples, removing the “fear factor” usually associated with this complex subject. The book will help readers to discover diverse perspectives of statistical theory followed by relevant worked-out examples. Keeping in mind the needs of readers, as well as constantly changing scenarios, the material is presented in an easy-to-understand form.

Written by one of the main figures in twentieth century statistics, this book provides a unified treatment of first-order large-sample theory. It discusses a broad range of applications including introductions to density estimation, the bootstrap, and the asymptotics of survey methodology. The book is written at an elementary level making it accessible to most readers.

The purpose of this book is to explore several specific areas of research in two distinct but related fields: digital signal processing and modern control and estimation theory. There are enough similarities "and" differences in the philosophies, goals, and analytical techniques of the two fields to indicate that a concerted effort to understand these better might lead to some useful interaction and collaboration among researchers.The author writes that his examination "will in general not be result-oriented. Instead, I have been most interested in understanding the goals of the research and the methods and approach used. Understanding the goals may help us to see why the techniques used in the two disciplines differ. Inspecting the methods and approaches may allow one to see areas in which concepts in one field may be usefully applied in the other. The book undoubtedly has a control-oriented flavor, since it reflects the author's background and also since the original purpose of this study was to present a control theorist's point of view at the 1976 Arden House Workshop on Digital Signal Processing. However, an effort has been made to explore avenues in both disciplines in order to encourage researchers in the two fields to continue along these lines."Indeed, the book contains numerous suggestions for new research directions and speculations on possible new results, all of them a direct result of the purposeful mixing of the ideas of the two disciplines. For the benefit of researchers who may wish to follow up some of these suggestions and speculations, the author has assembled a comprehensive bibliography, consisting of more than 600 references.In order to achieve his unique perspective of viewing each field in the context of the other, the author examines such topics as stability analysis of feedback control systems and digital filters subject to the effects of finite wordlength arithmetic; linear prediction, parameter identification, and relationships involving Kalman filtering and "fast" algorithms; system synthesis, realization, and implementation; two-dimensional filtering, decentralized control and estimation, and some of their connections with image processing; and aspects of nonlinear system theory, including homomorphic and bilinear systems.