Ken Kwong-Kay Wong
BOOK REVIEW
Business and management researchers would probably agree that some applied research projects
have limited participants because of the project nature. Surveying multinational CEOs, female senior
executives, or physically disabled workers can be challenging because the sample size can be small and
the data distribution is often skewed. These problems may cause researchers to draw incorrect
inference and prevent them from carrying out structural equation modeling (SEM) where strict data
assumptions are required.
Partial Least Squares (PLS) is a soft-modeling approach developed by Herman Wold in the mid
60s. Since PLS is insensitive to data non-normality, with no parameter identification problem, and has
relatively small sample size requirement, it is often considered by researchers to be a good alternative
to traditional covariance-based approach in SEM.
The first edition of the Handbook of Partial Least Squares was just published by Springer in
February 2010, after a 3-year publication delay. This 800-page book is the second volume in the
Springer Handbooks of Computational Statistics series. It is positioned to be a comprehensive reference
guide that explores the concepts, methods, and applications of the PLS statistical procedure. The book
is written for professors, PhD students, and research professionals who want to gain a better
understanding of this emerging multivariate analysis approach. Since it assumes some knowledge of
intermediate statistical knowledge, this book is not a beginner text for college students. The global
research community would benefit from this handbook due to a general lack of PLS publication in the
market. This book is timely as PLS has gained increasing attention of the research community in the
past decade.
This handbook is edited by V. Esposito Vinzi et al. who are renowned experts in the PLS field. A
whopping 33 articles are contributed by 74 authors from the global academic and research community.
There are three main parts to the handbook, which one should read in sequence from the book's
beginning to end.
The first part explores the PLS methodology in general. The articles are grouped into five sections
to help readers understand PLS in a step-by-step manner. The first section (Chapter 1-3) consists of
three articles, discussing the basic concept and model assessment of PLS. I enjoy reading Dijkstra's
article the most because he reveals the history of PLS from a first hand perspective. As Wold's PhD
student at the Wharton School, he brings readers back in time to view the early development of PLS
vividly. Dijkstra clearly explains the "basic design" and how PLS can be used to construct proxies for
the latent variables. The second section (Chapter 4-7) extends the PLS Path Modeling discussion to
help readers design advanced, multi-block models. I find Chin and Dibbern's article interesting as they
present a new permutation-based procedure for carrying out multi-group PLS analysis. The issue of
classification is being covered in the third section (Chapter 8-10), while the fourth section (Chapter 11-
14) illustrates how PLS path modeling can be used in customer satisfaction studies. Advanced PLS
regression modeling is explored in the fifth section (Chapter 15-17). Readers may find the paper by
Wold, Eriksson and Kettaneh useful as it explores the use of PLS in data mining and data integration.
Part two (Chapter 18 to 27) is my favourite as it shows how PLS can be applied to solve
marketing problems. It covers case studies ranging from employee satisfaction, brand preference,
customer loyalty, customer value, web strategy, to total quality management. I believe business and
management researchers will appreciate the various examples presented in the book. PLS is presented
as a good alternative to traditional AMOS or LISREL approach in conducting SEM analysis if the
model is designed properly. Kristensen and Eskildsen's article "Design of PLS-Based Satisfaction
Studies" is a must read for marketers who want to make use of PLS in their customer satisfaction
projects.
The final part of this book (Chapter 28 to 33) is designed to be a tutorial for PLS learners. It is a
collection of "how to" articles, guiding readers to properly design and build PLS models. I enjoy
reading Chin's article "How to Write Up and Report PLS Analyses" as it helps readers to report their
research findings in a professional manner.
While this book serves as a great resource to those who are new to PLS, researchers who have
been actively using this statistical procedure and following its development since the late 90s may find
it a slight disappointment. This is because the book mainly consists of extended version of existing PLS
literature that has been published in the past decade. Another imperfection can be found in the tutorial
section. While the articles in part three help readers to choose the right software and develop advanced
model for PLS analysis, the book lacks sufficient guidelines to alert readers what "not to do" when
using PLS in their research projects. As MIS Quarterly (MISQ) has twice pointed out in its editorial
(Marcoulides & Saunders, 2006; Marcoulides, Chin & Saunders, 2009), PLS is not a magical silver
