Excerpt from the January 2014 review in Science Books and Films
by Donald. E. Myers
University of Arizona, Tucson, AZ.
Other than saying that the objective is to explain concepts in sensory and consumer science, the authors do not indicate an intended readership. Although it has some aspects of a text book it does not include any exercises. It is organized around seven "themes" which are Difference Tests, Rating and Ranking Methods, Claims Support, Drivers of Liking® and LSA, Portfolio Optimization, Probabilistic Multidimensional Scaling and Combinatorial Tools. In turn each "theme" is treated by a number of two page scenarios, each of which is reprinted from IFPress. Each scenario introduces and describes an application pertaining to foods, beverages, pharmaceuticals, air care, packaging, fragrances and energy bars among others. Each scenario includes a list of references. There are also two appendices listing published papers; one is chronological and the other is by subject. There are also some 27 statistical tables that are referenced in various scenarios. The Differences theme is exactly what the title implies and introduces Thurstonian scaling. It comprises thirteen scenarios. Examples of difference problems include "which of these two items is the sweetest" or "choose the most different from among three items". Since there are many different statistical tests that might be used it is desirable to know which one is the most powerful or which one can detect differences with the smallest sample size. To compare different tests it is necessary to have a common scale, i.e. Thurstonian scaling. This is important for another reason, unlike the physical sciences, data in sensory or consumer sciences is often discrete. In particular the data may only be nominal or ordinal. While one could do the arithmetic to compute means and/or variances the results may not make any sense, e.g. if the data are ranks then the "average rank" does not have any real meaning. Thurstonian scaling is discussed and used in most of the scenarios and most of the subsequent themes. The Ratings and Rankings theme has only five scenarios. Ratings considered include intensity ratings, just-about-right and relative to reference scales. Rankings might be full, first and last. Using Thurstonian scaling this theme is shown to be related to the Differences theme. The third theme is (advertising) Claims Support which has seven scenarios. The problem is how to substantiate a ratio or multiplicative claim, equivalence testing as well as how to handle no count differences in preference and difference testing. The fourth theme is Drivers of Liking® and Landscape Segmentation Analysis. This theme has twelve scenarios. LSA is a method for generating maps to place individual ideals, concepts or products. There are examples to find motivations for product consumption, latent ideals or groups of consumers who agree on what they like. LSA has been found useful to separate product and branding effects. The fifth theme is Optimizing Product Portfolios. Again the title is very suggestive of the topics covered in the theme. There are five scenarios which pertain to product mixes.--Donald E. Myers, University of Arizona, Tucson, AZ