I've seen a one-star review on this book, and it totally threw me off. I AM a professional statistician, and what's more I am a survey statistician. And in my niche, this book is the one of the most information-intensive and important monographs out there.



The book represents a set of invited talks from a conference at Southampton (UK), the location of the world's leading group in survey statistics. The conference was held in honor of T. M. F. Smith, a prominent survey statistician, in 1999. Those talks were delivered by highly recognized contributors to the survey statistics field: Ray Chambers and Chris Skinner provided editorial overviews of book parts as well as contributed some chapters; Roderick Little -- missing data and Bayesian methods, Jon Rao and Roland Thomas -- chi-square tests, Jerry Lawless -- history event analysis, Wayne Fuller -- multiple phase samples, and others... all at the top of their respective fields.



This is certainly NOT the first reading in survey statistics (for which I'd send people off to Lohr's Sampling: Design and Analysis or may be Korn and Graubard's Analysis of Health Surveys books), and not even the second reading (for which I'd recommend the previous collection coming from Southampton -- the 1989 book edited by Skinner, Holt and Smith, or Mary Thompson's 1997 book Theory of Sample Surveys or may be even Sarndal's book Model Assisted Survey Sampling, although some would argue the latter two are too technical/mathematical -- well that's my personal preference, I believe one needs to understand the basics deeply before going into more advanced topics). This is a monograph for people deeply in the field of survey statistics. I assigned several chapters from this book as readings for my advanced graduate class, as those are the most concise and clear treatments of topics such as design- and model-based inference (a chapter by Binder and Roberts) or population based case-control studies (a chapter by Alastair Scott and Chirs Wild).



I should note that it is likely to be difficult to use this book as a reference (unlike the classic works of Kish Survey Sampling or Cochran Sampling Techniques, 3rd Edition that do contain the foundations such as stratified and clustered samples), as it does not list all the typical designs and situations. It is a research work; some of the topics are outlined but briefly, other topics or techniques might sooner or later become outdated and replaced by newer ones, and the interest of all the authors is in complex situations. Inevitably to make any progress in those, some simplifications had to be made, so the reader may not find particular answers for his particular data and design situation, but there should be enough ideas scattered throughout the book to indicate the typical directions of biases or variance inflations that might be occurring in practice.



The list of topics highlighted in the book has probably been comprehensive to represent the frontier of research in late 1990s, and includes the foundations of survey inference, specific features of the models with categorical and continuous responses, missing data, and longitudinal data. Since then, at least two new area appeared: confidentiality and disclosure risks, and small area estimation (see Rao's Small Area Estimation).



Finally, I must applaud the editors for producing a book out of a conference that makes a coherent sense. A typical conference proceedings book is usually a nightmare, with a bunch of unrelated stuff bound under the same cover for the reasons of being delivered at a particular place at a particular time. Not with this one! The notation has been unified reasonably well across different authors; the papers have been combined (and some split) into more natural groupings, etc. The index of topics and authors is not a five minute work, either.