Behavioral aspects of travel information: models and experiments.

It is hoped by many that travel information provision may serve as a means to change traveler behavior in ways that are beneficial to the transport system, and to society as a whole. Such hoped for behavioral changes mainly concern shifts in mode-, route- and departure time-choices amongst car-drivers. A prerequisite for any effect of travel information on traveler behavior, is that travelers acquire information that is available to them. This paper provides two complementary attempts to gain insight into the acquisition of information among travelers, based on the notion of regret-avoidance: a traveler may be aware of the fact that, due to uncertainty in the transport network, he may end up having chosen an alternative that turned out to be less attractive than a non-chosen mode. This induces regret on the side of the traveler. This traveler anticipates this possibility of regret and aims to minimize expected regret when choosing from available alternatives. In order to grasp the full behavioral complexity of the decision to acquire information, addressing all relevant behavioral mechanisms, a theoretical model of response to information is constructed first. It hypothesizes that information is acquired only when the difference in expected regret between the choice situation before and after having acquired the information outweighs the cost of acquiring the information. After having derived this theoretical model, we proceed by proposing an second, data-oriented formulation of information acquisition. This second model is based on the notion that travelers will seek information only when the expected regret that is associated with a given choice situation exceeds some threshold, which may vary from person to person. This econometric model provides a less elaborate, and therefore econometrically more tractable, perspective on travel information acquisition. We estimate this second model using data collected by a multimodal travel simulator with information provision, and discuss estimation results. For example, it indeed appears that regret-thresholds are on average substantial and vary significantly across individuals. For the covering abstract see ITRD E137145.