摘自Amazon

摘自Amazon

This book (which is becoming somewhat of a classic) is simply outstanding. The author's philosophical and mathematical reasoning is impeccable. This book is very deep and will not just teach you theory and techniques but will teach you how to think and how to approach and formulate problems, as well as how to interpret the results obtained by various methods and use them in a practical setting. This book is very practically oriented. The best thing about this book, however, is that it is so clearly written. Berger is an outstanding writer with the ability to communicate difficult concepts without watering them down.

This book covers decision theory and Bayesian statistics in much depth. While it is a high-level text oriented towards researchers and people with strong backgrounds, it is clear enough that someone learning this material for the first time would have little trouble with it. It provides ample review and clear exposition of key mathematical and statistical concepts such as sufficiency, convexity. Its exposition of invariance (with respect to groups of transformations) is both the clearest and most rigorous I have found in any statistics text. In my opinion, there are no weak or unclear sections in the book, and the difficulty level does not rise disproportionately in later chapters the way it does in many books on similar subjects.

This book is rich with examples, and the examples are mostly of a practical nature, in contrast to the "toy" mathematical examples that dominate many books written at this advanced level. The exercises are diverse and extensive, and have a good gradient of difficult level for building both technical skill and depth of understanding. The exercises are more carefully worded and constructed than is typical for books at this level. Most of the typos have been caught and corrected in the revised edition.

This is an old book. The author, in his philosophy, was arguably well ahead of his time. The ideas contained in this book are highly modern. However, the use of computers in statistics has changed since this book was written. This book is a book on theory and will teach you how to do things by hand. I do not see this as a weakness at all, but one should be aware of it when considering this book. But, as the other author noted, it will not teach you algorithms, numerical techniques, or how to use a statistical computing package.

I think this book would make an outstanding textbook for a course in statistical decision theory or Bayesian statistics. It would also be useful as a supplement for a course in statistical inference. Perhaps more importantly, it is very useful for self-study. I think this book would make an excellent addition to any statistician's collection--and it would certainly be useful to people working in more practical settings, such as business, science, or social science. If you are going to buy any one advanced, theoretical book on statistics, this would be the one to buy.