Recent work on risky choice modelling has sought to address the theoretical shortcomings of expected utility theory (EUT) by using non-expected utility theoretic (non-EUT) approaches. To date, however, there is little evidence to show whether the complexity of non-EUT actually leads to better model performance. Moreover, almost all the relevant research has adopted stated choice data which, although flexible and cheap, has limited validity. This thesis empirically investigates the feasibility and validity of non-EUT approaches in revealed preference (RP) contexts, in which travel time distribution is extracted from historical travel time data to subsequently present systematic comparisons between EUT and non-EUT approaches. Additionally, this thesis also discusses implementations based on these empirical results and, in particular, highlights the influence of non-EUT on the valuation of travel time savings.
A risky choice framework is proposed so as to incorporate non-EUT into a Random Utility Maximization structure. The non-EUT approaches modelled in the thesis consist of Subjective Expected Value Theory, Subjective Expected Utility Theory, Weighted Utility theory, Rank Dependent Expected Value, Rank Dependent Expected Utility, Prospect Theory, and Cumulative Prospect Theory. The first dataset is collected from the SR91 corridor in California and involves a choice between a free flowing and reliable tolled facility and a congested and unreliable un-tolled facility. The second case study is based on the London Underground (LU) system and involves the choice between alternative competitive underground services linking pairs of stations.
This thesis provides insights into how EUT and non-EUT models perform in the real world. The RP methodology and risky choice framework offers an avenue for future research to identify a wider range of alternative choice theories using realistic data. The empirical results suggest that there are merits in applying non-EUT to the modelling of travellers’ risky choice behaviours.