These comments are about the new third edition of Bryan
Manly’s “Multivariate Statistical Methods: A Primer”.   I should make it
clear right away that I am confident that the fault lies not with Bryan
himself but with some [Junior? Ex?] copy writer at CRC Press. Whilst I
scarcely know Bryan myself (ten minutes chat at a large conference a while
ago) I know we have many mutual friends and acquaintances, apologies in
advance if my complaint is taken personally.

I am sure I am one of many who think highly of this book and I have
included it on my reading lists for my own multivariate courses for many
years. I recommend it particularly to numerate but non-mathematical
scientists (especially graduate archaeologists) as being full of good
common sense, useful intuitive explanations and lots of interesting
examples including the raw data, many of them with an archaeological
background.  My only reservation has been that the second edition (1994)
is beginning to seem a little dated.

Thus I was very glad to see the announcement of a third edition with the
clear statement “Topics new to this edition include confirmatory factor
analysis, the use of mixture models for cluster analysis, and the emerging
techniques of data mining and neural networks” (taken verbatim from the
CRC website and repeated on Amazon).

Thinking of Chapman & Hall as one of the top three or four statistical
publishers of high repute and knowing the second edition well I ordered
this using Amazon’s seductive 1-Click rather than the lengthy process of
requesting an inspection copy from the publisher.

It has arrived and on the back cover it states “Features: Presenting new
topics including confirmatory factor analysis, the use of mixture models
for cluster analysis, neural networks, and data mining.  Provides all the
data used in the book on a companion website.”

Looking in the index under Data Mining it takes you to P201 (Chapter 13,
Epilogue): “Recent developments in multivariate analysis have been made in
the closely related field of data mining…………. . This topic has not been
considered in this book,…” (and then the reader is referred to Hand et al
(2001) – a good reference.

Under Cluster Analysis, Mixture of several populations, you are taken to
P135 , “An approach to clustering which has not been considered in this
chapter assumes that the available data come from a mixture of several
distributions…”

Confirmatory factor analysis is not included in the index, nor the
contents list but in the factor analysis chapter, P102, it is mentioned in
the final paragraph of ‘further reading’ and the reader referred to two
other texts for details.  Neither neural networks nor a website for data
sets are mentioned in the index or contents list or preface and as yet I
have not found them in the text anywhere.  However, the data sets are
easily found in the standard CRC site at
http://www.crcpress.com/e_products/downloads/.

I suppose the lesson is that the buyer should beware, but I think it sad
that we have to be aware of publishers such as this.

Nick Fieller




--
Dr Nick Fieller
Department of Probability & Statistics
University of Sheffield
Sheffield, S3 7RH
U.K.

tel: +44 (0) 114 222 3831 (direct & office/voicemail)
fax: +44 (0) 114 222 3759

http://www.shef.ac.uk/nickfieller