Fishbein Theory and the Bass-Talarzk Problem

In a recent note in JMR, Cohen, Fishbein, and Ahtola 6 argue that two other articles in JMR 3, 15, presented as using expectancy-value models of attitude, do not in fact use measures consonant with those models. The Sheth-Talarzyk article criticized 15 was concerned with applications of Rosenberg's theory. Cohen, Fishbein, and Ahtola 6 claim that Rosenberg had been misinterpreted; Sheth in his response 14 argues that Rosenberg was correctly interpreted. I do not wish to enter into this rather fruitless argument here. More interesting, it seems to me, is the discussion between Cohen, Fishbein, and Ahtola 6 and Bass 2. Bass 2 accepts the arguments of Cohen, Fishbein, and Ahtola 6 that the measures in the original Bass and Talarzyk article 3 had not accurately represented Fishbein's theory of attitude. However, he points out that the direction of interest in the original article 3 was not in testing Fishbein's theories in any way, but rather in exploring the question of whether it was possible to show that brand preference within individuals could be predicted from belief measures about brands. Bass and Talarzyk are particularly interested in this point because they mistrust cross-sectional regressions, that is, the prediction of brand preference to a single brand across a set of individuals from beliefs measured across that same set of individuals.