Sampling

Rating:
Summary: great second edition
Review: I am still personally attached to Cochran's classic "Sampling Techniques" which was marvelous for its insight and clarity. it was also helpful to me in my first research work in survey sampling that I coauthored with Tommy Wright. But unfortunately Bill Cochran is no longer with us. Several good survey sampling books both theoretical and applied have come out since then. Thompson's first edition was a big success. Unfortunately I never read his first edition. But I did get the second edition.

It is well written and covers both the model based and design based approaches. There is an adequate treatment of the theory. As an applied statistician I particularly like the fact that he presents the results and formulae for sample size determination problems.

All the key topics are addressed and there is ample reference to current and past literature. The book could be useful for a graduate or undergraduate course in survey sampling. Bootstrap ideas are presented and the book is really very modern in its approach. I was also surprised to see such topics as Line and Point Transects, Capture-Recapture Sampling and Kriging.

There are even some solutions to chapter exercises My only criticism would be that for a course text it would be more useful to have a somewhat larger collection of exercises at the end of each chapter and a correspondingly larger set of selected solutions.

Rating:
Summary: Thompson treats the details of sampling theory best
Review: In the last 5 years or so, as a practioner, I have examined and used, or tried to use, 5 or 6 current texts on modern Sampling theory and practice. Steve Thompson's has been the most consistently useful. He offers more than the conventional wisdom, more than the standard technique. He describes alternative and special methods. when they are best used and how, complete with pros and cons. His generally clear explanations go beyond the usual mathematical formalism to the practical aspects.

For applied folks such as myself, this is really critical. You do not need to be a professional mathematician to make good use of this volume.

In particular, I have made good use of the Thompson's approach to double sampling technique in my own work sampling from the Net, a field loaded with methodological pitfalls.

If you are seriously concerned with generating valid samples in practical applications, Thompson will become a trusted friend.