Attitude-based models for binary choices: A test for choices involving an innovation

Several studies either implicitly or explicitly converted the well-known theory of reasoned action into a theory for choice. In this paper we elaborate upon such an attitude-based choice theory by proposing different model variants for binary choices. The models vary in two respects: (1) the level at which alternatives are compared (level of comparison), and (2) the way the comparison takes place at each of these levels in reaching a choice (comparison mechanism). Based on these dimensions 45 models were formulated that were examined empirically by logistic regression on choice probabilities. The data set consisted of measurements on 467 entrepreneurs for different financial services, one of which was an innovation. Of the different comparison mechanisms, subtraction, without explicitly incorporating the similarity of the choice alternatives, turned out to be the most adequate. For these models. equal weighting of the attitudinal components could not be rejected, which made it impossible to test at which level the alternatives were compared. Structural equation modeling of the same data shows that the equality of the weights for the attitudinal components can be explained by the affective component of attitude being a major predictor of choice.