Partial Least Squares is a family of regression based methods designed for the an- ysis of high dimensional data in a low-structure environment. Its origin lies in the sixties, seventies and eighties of the previous century, when Herman O. A. Wold vigorously pursued the creation and construction of models and methods for the social sciences, where “soft models and soft data” were the rule rather than the exception, and where approaches strongly oriented at prediction would be of great value. Theauthorwasfortunatetowitnessthedevelopment rsthandforafewyears. Herman Wold suggested (in 1977) to write a PhD-thesis on LISREL versus PLS in the context of latent variable models, more speci cally of “the basic design”. I was invited to his research team at the Wharton School, Philadelphia, in the fall of 1977. Herman Wold also honoured me by serving on my PhD-committee as a distinguished and decisive member. The thesis was nished in 1981. While I moved into another direction (speci cation, estimation and statistical inference in the c- text of model uncertainty) PLS sprouted very fruitfully in many directions, not only as regards theoretical extensions and innovations (multilevel, nonlinear extensions et cetera) but also as regards applications, notably in chemometrics, marketing, and political sciences. The PLS regression oriented methodology became part of main stream statistical analysis, as can be gathered from references and discussions in important books and journals. See e. g. Hastie et al. (2001), or Stone and Brooks (1990),Frank and Friedman (1993),Tenenhauset al. (2005),there are manyothers.
The "Handbook of Partial Least Squares (PLS) and Marketing: Concepts, Methods and Applications" is the second volume in the series of the Handbooks of Computational Statistics. This Handbook represents a comprehensive overview of PLS methods with specific reference to their use in Marketing and with a discussion of the directions of current research and perspectives. The Handbook covers the broad area of PLS Methods from Regression to Structural Equation Modeling, from methods to applications, from software to interpretation of results. The Handbook features papers on the use and the analysis of latent variables and indicators by means of the PLS Path Modeling approach from the design of the causal network to the model assessment and improvement. Moreover, within the PLS framework, the Handbook addresses, among others, special and advanced topics such as the analysis of multi-block, multi-group and multi-structured data, the use of categorical indicators, the study of interaction effects, the integration of classification issues, the validation aspects and the comparison between the component-based PLS approach and the covariance-based Structural Equation Modeling. Most chapters comprise a thorough discussion of applications to problems from Marketing and related areas. Furthermore, a few tutorials focus on some key aspects of PLS analysis with a didactic approach. This Handbook serves as both an introduction for those without prior knowledge of PLS as well as a comprehensive reference for researchers and practitioners interested in the most recent advances in PLS methodology.
