Matteo's Wasps' Nest

The only class I am taking this semester (and the second-to-last before I am done with classes) is APMA 2670, Mathematical Statistics 1, an advanced graduate course on statistical inference. Among the suggested references for the course there is P.J. Beckel, K. A. Doksum, “Mathematical Statistics, Vol. I, 2^nd Ed.” If you read the Amazon users reviews for it, you will think this is a bad book, much worse than the (30-or-something years old) first edition. It looks like the biggest weaknesses of this revised edition are the huge amount of typos, which hinders clarity, and the scarcity of the index, which endagers the usefulness of the book for the researcher.

The fact is that those reviews refer to the first printing of the second edition of the book, which was indeed flawed with tons of typos, as shown by the 15-pages-long errata. If you ever think of getting the book, be sure to get the “Updated Printing” version. There are still some typos, but not more than in any scientific book (I retain the right to revise this statement as I go on reading the book). I still have no opinion about the index, whether it was improved or not, since I am using it as a textbook and not as a research reference.

As far as the contents of the book are concerned, I am not ready yet to do a full review and I doubt I’ll ever be, but I have to say that the Authors make lots of references to the many examples they present, which is, in my opinion, not such a good things to do, because you can find sentences like:”The problems like the one presented in Example X.Y.Z. are called foobar“. Now, extrapolating a class of problems from a single example is, at least for me, a very difficult inference task (:)), so I would have preferred more formal definitions and a less heavy use of examples to explain fundamental concepts. Examples should help in understanding previously formally defined concept, not be used as implicit definitions. I am also quite puzzled by some notational choices: why on Earth did the Authors choose to denote the set of reals number as $R$ and not as the more classic $\mathbb{R}$ (yes, I should install a LaTeX WordPress plugin). The style is pretty dry, but I did not expect anything different, so I don’t consider this an issue.
Something else bothers me instead: the fact that foot notes are not printed at the foot of the page they appear in, as the name should suggest. They are instead collected at the end of each chapter, which makes checking them quite annoying. Luckily there aren’t many footnotes.

I have to note that the Authors claim, in the Preface to the Second Edition, that Volume II of the book is expected in 2010, with topics like nonparametric methods, bootstrap and Markov Chain Monte Carlo. I don’t see any sign of the existence of this book anywhere, but I hope the Authors will do a better job in having typos fixed before printing the book.

This review may and should be expanded. =)

I have another bad tale involving this book but the main “villain” of the novel is actually Amazon and not the book nor its Authors. I bought a first copy of the book from Amazon (I mean, directly from Amazon, not from a 3rd party seller listed on Amazon) and it was one of the last two left in stock (the book is currently out of stock on Amazon, but should be available shortly). The description clearly said that the volume is hardcover, but the copy I got was softcover. What is much worse, the pages inside were not printed: they were non-page-aligned scans/xeroxs of a genuine copy of the book. Clearly I was not satisfied given that I paid around $60 for my copy, therefore I filed for return and full refund. Meanwhile, I ordered another copy of the book, this time from a 3rd part seller listed on Amazon. This time I got a genuine print. I still have to send my bad copy back, but I’ll update this post when I get the refund (or if I do not get it).

Posted from Providence, Rhode Island, United States.