Choice, deferral and consistency

We report an experiment that tests whether forcing subjects to make active choices makes their decisions less consistent. We find that forced choice significantly increases the incidence of choice reversals. We also propose and apply a new combinatorial-optimization method that allows for analyzing decision datasets that include a "choice deferral" default option, and to categorize each of these datasets depending on whether its best explanation comes from models of unconstrained or unattractiveness-constrained utility maximization, a Bayesian model of rational information acquisition or a model of rational indecisiveness. Applying this method to our data we find that about three quarters of all relevant subjects are best explained by either unconstrained or unattractiveness-constrained utility maximization, and that most of the remaining ones are best explained by rational indecisiveness or the Bayesian model. Two types of questionnaire data finally suggest that indecisiveness was the reason why the majority of subjects in the last category deferred choice.