Customer Reviews

Nonparametric Statistical Methods, 2nd Edition

5 star:		(5)
4 star:		(1)
3 star:		(1)
2 star:		(0)
1 star:		(1)

 

 

In the 1970s this text became a classic on the subject of nonparametric methods. It was written for practitioners and students. It is introductory and comprehensive. It describes the methods accurately but does not cover the theory. Later Randles and Wolfe wrote a companion book covering the theory. This revision is much larger and covers the many advances over the past 20 years. It covers bootstrap methods as well. Also computational advances are discussed.

Conover's "Practical Nonparametric Statistics" is another fine book for practitioners. I also recommend Lehmann's book on nonparametrics. It was published in 1975 and is not easy to find these days.

This is an excellent book on a somewhat underutilized group of statistical techniques. It could be used for a course in nonparametric statistics at the graduate level in Psychology or the social sciences, although I don't think the whole book could be covered in a semester.

It is perhaps more valuable as a reference for the practicing data analyst. Because of the format, it is relatively easy to find a procedure that does what you want. There are 11 chapters, the first of which is an introduction, and the others each cover one type of problem (e.g. the one-sample location problem). Within each chapter are a variety of procedures, each of which is discussed in the same format: Procedure, large-sample approximation, ties, example, comments, properties and problems.

In addition, there are close to 200 pages of tables, many of which I haven't seen elsewhere.

Overall, highly recommended for anyone who needs to use or teach these techniques.

It will take about 3 weeks to get used to the convoluted organization of this book. The explanations are actually fairly good, as are the examples, but the set up of text, comments, examples and problems is often difficult to follow. Tables are not where they are expected, references are circular, and so on. I will say that at least the system, as it is, is 'applied' consistently, so once you get used to it, it only slows understanding slightly. Of course, the editors were statiticians, not editors...

I fully recommend this book as part of one's nonparametric statistics library. It has wonderful detailing of the assumptions for the methods, and the book is tied together by a strong practical applications focus.

This book is great for what it is - an encyclopedia for non-parametric methods. This isn't the text to read for an exposition about the development and all theory and proof behind the method (Lehmann is great for that) this is the place to go where you say my data has such and such a structure, and I have such and such hypothesis about it. This doesn't mean that this is a cookie cutter book - it provides the assumptions about each tests, the formulas for hand calculation (great for checking if the R function was coded correctly!), the relative efficiencies, info about ties and the asymptotics. as well as the notes at the end. In this sense it has something for must levels. I agree though the tables are a bit silly and the price is a bit high

A couple of notes:

1) Some people have critiqued the structure - I actually like it but can see how it can be a bit confusing

2) The book is definitely about classic non-parametrics based on rank tests. There is much more to non-parametrics but this book fits a good niche

If you are a practitioner who realizes ANOVAs are silly you will get something out of this book. And if you have a good understanding of theory it won't feel too dumbed down.

If classic means outdated, this is the one.

1) Although published in 1999 when we already have 1G CPU speed CPUs, the authors seem unaware of the world of computing. This book still includes 180 pages of tables for thumb searching. It is true that the book recommends and uses statXact and Minitab in many examples. So what's the point to include so many irrelevant and outdated tables in the book? It is much easier to just include the compute codes and outputs. Most modern applied stat books do this.

2) This book is written in a cookbook style with a strong focus on hand calculation. I understand that the authors are quite senior, but they just should not update a 1970s classic with more old fashioned hand calculation. In a sense, the authors are still living in the 1970s when computers were not accessible to almost all regular folks.

3)I don't know what the level the book is supposed to be. Sometimes I feel stupid when I was reading it. The authors detailed simple things again and again, wasting lots of precious space on irrelevant things, while leaving important justification of methods out of the book. Respect readers!

4) the most annoying feature is the way the authors present the ideas. As a cookbook, the authors postpone almost all pseudo-theoretical development at the end of section under the name of comments, and these comments seem disorganized sometimes. The point is this kind of structure disrupts the flow of logic. Readers have to flip back and forth to figure out how and why the formula form. I am not opposed to postpone complicate theories at the end of section, but for simple things such as how you get the means and variances or what the motivation behind the formula is, a couple enlightening sentences are good enough to make readers stop wondering. But the authors did not think that way.

There are more annoying things in the book, but since I have just finished the first three chapters (which are simple stuff made complicate by the authors), I would not say more bad things. It is ironic that the instructor who chose this book as the textbook for nonparametric data analysis talks computing all the time in the class! the unjustified fame of the authors kills innocent people.

By the way, In the 21th century, I don't know how many people these days are still doing analysis on data with sample size less 20, and still use sign rank or similar stuff as the main analytic tools. It is nice to have an overview of these classical ones, but a heavy book devoted to this overkill. And an updated one still focusing on hand calculation, please pardon my French, that sucks.

Readers, be warned by this outdated book.

I found this book to be very helpful and it required minimal interpretation from academia to understand. More so for the practicioner than the theoretician.

I found this book to be very helpful and it required minimal interpretation from academia to understand. More so for the practicioner than the theoretician.