The late Leslie Kish published this classic text on survey sampling in 1965 with Wiley. In 1995 they reprinted it in their statistics classic series. Professor Kish had a long career at the University of Michigan where he taught survey sampling to social scientists. There he also established a summer program in survey sampling design and one of the top consulting services for designing and implementing surveys.

I had the pleasure of meeting Dr. Kish in the late 1970s when he was consulting on sampling issues for the newly formed US Department of Energy. At Oak Ridge National Laboratory I was working on survey sampling research for generating and validating energy data. Dr. Kish was charming and very humble. He treated young researchers like me the same way as he would treat his famous colleagues.

This book covers all the practical aspects of surveys and emphasizes the human population studies that Dr. Kish was involved in. But he is quick to point out in the introduction, the generality of the methods and their applications to other populations and many other disciplines (his experience was mainly in the social sciences at that time). The mathematics is presented as simply as possible. There is a great deal of verbal description throughout the text. He wrote it with practitioners in mind and tried to keep the mathematics simple.

It differs in style and coverage (but not quality) from the more technical book by Cochran. He puts more effort into the discussion of philosophical and practical issues than Cochran does and he illustrates his ideas through many practical and real examples.

For a book written in the 1960s many researchers might be surprise to find the detailed coverage of issues of bias and nonsampling errors in surveys, including problems of missing data and non-response and a whole chapter on sampling from imperfect frames.

All the important classical topics are covered including stratification, cluster sampling, sampling proportional to size, area sampling, and multi-stage sampling.

The topics are all there. Today the issues of nonsampling errors and adjustment for underreporting (e.g. in decennial Censuses)are as important as ever.

This book is a great reference source. It only lacks some of the modern advances that may play roles in important applications such as the implimentation of hierarchical Bayesian models through the use of Markov Chain Monte Carlo methods and the bootstrap and jackknife to better assess survey uncertainty.

Originally designed as an undergraduate classroom text for a third course in statistics, Kish provides an adequate number of homework problems at the end of each chapter and he provides a brief selection of solutions to some of the problems in the first seven chapters.

I have used this book for a number of years & found that in most cases it has been THE book that I should have in my library. It is a must for the work place to buy.

I studied this book in my university here in Ecuador, and i believe that is a good book; in this moment i am working in "Census Agricultura Project" and think to recommend to buy