The Nature and Uses of Expectancy-Value Models in Consumer Attitude Research

In tests of two expectancy-value models of attitude reported recently in JMR, Sheth and Talarzyk [18] tested a model attributed to Rosenberg [14] while Bass and Talarzyk [3] tested a model attributed to Fishbein [7, 8]. The two models were tested on a common data base using essentially the same measures.^ Unfortunately, neither article provided a valid test of either of these two models and served only to confuse rather than to clarify some of the issues regarding the use of expectancy-value models as a basis for understanding some aspects of consumer behavior. Because of the growing interest of consumer researchers in an expectancy-value approach to attitude structure and dynamics and behavioral prediction, it may be useful to specify the shortcomings of these two studies and clarify some essential characteristics of the expectancy-value approach. We intend to show that the authors: (1) did not provide tests of any expectancy-value model let alone those of Rosenberg or Fishbein; (2) actually proposed a new model which might be termed an adequacy-importance model; and (3) demonstrated that their model actually has less predictive validity than an expectancy-va lue model.