"This approach to data analysis represents an important and original school of thought." Short Book Reviews, Publication of the International Statistical Institute "The volume begins with an excellent discussion of data treatment, with an emphasis on the fact that the statistical and modeling treatment depends on choices and decisions made by the analyst. I believe that this is an important and often overlooked point, and I appreciated John P. Van de Geer's emphasis on the responsibility of the analyst. . . . Van de Geer is quite conscious of addressing a relatively broad audience, based on the care taken to provide interpretations of results in the context of various functional areas. . . . The volume is at its best (which is really good!) when Van de Geer is describing implications of the results, providing wonderfully intuitive interpretations of relations and of byproducts of the analytical approach. An example of this strength of the book is the discussion of the implication of optimal quantification with respect to eigenvalues on page 56. (I must admit that I have never felt as comfortable with eigenvalues and their interpretation as I did while reading this book!)." --Elizabeth L. Rose in Structural Equation Modeling Quote for both books "The set would be appropriate for use in a graduate course, with guidance from an instructor who has expertise in this approach to multivariate analysis. The interested researcher will find the set to be very helpful, particularly in terms of developing a coherent and accurate interpretation of the results." --Elizabeth L. Rose in Structural Equation Modeling Aimed at researchers and students interested in non-linear analysis of categorical variables, Multivariate Analysis of Categorical Data explains multivariate analysis geometrically rather than through the use of formal algebraic formulae. Asterisked sections, however, are included that present an algebraic explanation for readers who are more familiar with classical multivariate analysis. The author first covers what the desirable properties of a geometric display should look like on the basis of classical analysis. Next, he discusses how these desirable properties can be enhanced by gradually eliminating the restrictions imposed by a priori technique. He then explores such topics as the relation of principal components analysis; canonical analysis and generalized canonical analysis to one another; optimal quantification for one, two, and three variables; and how many dimensions are needed for optimal quantification. This book is appropriate for readers who have some familiarity with multivariate analysis.
Classical Methods of Multivariate Analysis
Principal Components Analysis
Properties and Risks of Optimal Quantification