Thoughts on Analyses of Intransitive Intertemporal Preference

A contender in the descriptive analysis of intertemporal choice is the tradeoff model (Scholten & Read, 2010), which provides a parsimonious and psychologically plausible account of many preference patterns in intertemporal choice, among which occurrences of intransitivity. In past applications of the tradeoff model to intransitive choices, individuals made all choices in a collection only once, and choices were often aggregated across participants. Dai (2017) examined whether intertemporal preferences are transitive in a Bayesian analysis of repeated individual intertemporal choices. Weak, moderate, and strong stochastic intransitivity rarely occurred. In this reply, I first provide a brief sketch (and in no way an exhaustive exposition) of the breadth of evidence that the tradeoff model addresses, and how it does that. I then point out several things that must be kept in mind while appreciating Dai’s contribution: (a) the paucity of intransitive choices does not argue against the tradeoff model, or any attribute-based model of its generality; (b) the interpretation that Dai offers of relative nonadditivity, a complex preference pattern previously observed in aggregate choice, including intransitive choice, is ad hoc, and lacks the generality that the tradeoff model provides in accounting for a host of preference patterns, among which relative nonadditivity; (c) substantive psychological models that maximize scope and minimize lack of parsimony, and that survive in empirical contests between them, are ultimately the route toward good judgment about the psychology of intertemporal tradeoffs.