Reply to Comments on the Nature and Uses of Expectancy-Value Models in Consumer Attitude Research JAGDISH N. SHETH* My reply to the comments of Cohen, Fishbein, and Ahtola is lengthy because I substantially disagree with a number of points and feel their comments are extraneous to most findings in JMR. Here is a summary of my views: 1. In their enthusiasm to relate other theories to Fish- bein's, the authors have misunderstood and misin- terpreted Rosenberg and the antecedent functional (means-end) school of thought based on the concept of valued states. 2. Contrary to assertions made by Fishbein, I don't think the Fishbein and Rosenberg models are equivalent or even similar in structure. The illusion of an equivalent structure comes from the weighted- sum formula used by both to obtain an index of cognitive structure. 3. There is no Bass-Talarzyk-Sheth theory of attitude. However, a comprehensive theory of attitude-be- havior relationship developed by Sheth some time ago is known to all three authors. 4. It is indeed naive to assert superiority of one model over other models when the studies (1) reflected different settings (e.g., laboratory experiment vs. field survey), (2) involved different issues (e.g., Negroes should be allowed in white neighborhood, story about Mrs. Williams, brands of grocery prod- ucts), and (3) utilized different statistical and meth- odological procedures (e.g., chi-squared, simple cor- relations, multiple correlations). YOUR VIEW OF ROSENBERG VS. MY VIEW OF ROSENBERG The basic contention of Cohen, Fishbein, and Ahtola is that the operational measures of perceived instrumen- tality and value importances in the Purdue study are not equivalent or even comparable to Rosenberg's theory. It * Jagdish N. Sheth is a Professor of Business and Research Professor at the University of Illinois. 462 is unfortunate that the authors did not carry out a study to prove their point; mere assertion based on a different interpretation is not enough. Let me describe why I still believe measures utilized in Purdue data are adequate measures of Rosenberg's perceived instrumentality and value importance. Fortunately, Rosenberg has reported his theory and findings in several places albeit with varying degrees of clarity [7, 8, 9], making it easier to assess what he means by value importance and perceived instrumentality. I find the following quotes a clear statement of his theory: "When a person has a relatively stable tendency to respond to a given object with either positive or nega- tive affect, such a tendency is accompanied by a cogni- tive structure made up of beliefs about potentialities of that object for attaining or blocking the realization of valued states .... Thus strong and stable positive affect toward a given object should be associated with beliefs that it leads to the attainment of a number of important values, while strong negative affect should be associated with beliefs that the object tends to block the attainment of important values. Similarly, moderate positive or negative affect should be associated with beliefs that relate the attitude object either to less important values or, if to important values, then with less confidence about the relationship between these values and the attitude object" [9, p. 17-8]. "A similar study was conducted by the author in which the attitudinal affects of a large number of sub- jects were measured with regard to two separate social issues. One month later these same subjects took a "test of cognitive structure" requiring them to rate a group of "Values" both for the degree of positive or negative reward each represented and also for the extent to which each value was believed to be fostered or de- feated through the influence of each of the two attitude objects, respectively" [8, p. 321]. It is hard for me to see from the above descriptions that a belief about the object attaining or blocking a valued state means the same thing as the probability or Journal of Marketing Research, Vol. IX (November 1972), 462-5