Evidence of intertemporal bias in a stated preferences route choice experiment

Intertemporal biases can occur in both retrospective assessments (e.g., recalling travel satisfaction) and prospective choices (e.g., choosing a route). In the former, memories of experiences are influenced by specific portions of the event, causing the recalled value to diverge from the normative or averaged value. In the latter, when choices involve alternatives that yield time-delayed outcomes, the utility of each alternative may be affected by the temporal order of those outcomes. Although such effects have been widely documented in various domains, recent studies of retrospective bias in transport argue that travel satisfaction evaluations are not subject to these distortions, and prospective biases have been scarcely explored. This article revisits intertemporal biases in transport using a stated preference route choice experiment in which two randomly assigned groups of participants choose between two car routes differing in the timing of congestion. The first group (N=205) was instructed to imagine taking each route for the first time (prospective evaluation), while the second group (N=198) was asked to imagine having taken both routes previously and to choose which one they would prefer to repeat (retrospective evaluation). Results show that, in both conditions, participants exhibited a bias toward routes with better endings, consistent with various findings from behavioral research and contradicting some findings reported for transportation. A mixed logit model with random parameters confirms that the effect is consistent across individuals. Further analysis indicates that the bias is not moderated by automatic behavioral tendencies, which are often assumed to influence cognitive biases. These findings point to a systematic distortion in the assessment of travel alternatives with temporally ordered attributes. We conclude by discussing implications for transport models, which, by the nature of the problem, deal with choices or evaluations of outcomes occurring at different time points and are therefore prone to intertemporal biases.