Travel behaviour modelling at the interface between econometrics and mathematical psychology

Whilst the field of choice modelling has been dominated by approaches based on random utility maximisation (RUM), there has recently been a considerable rise in work considering alternative model structures that attempt to incorporate behavioural insights from psychology, behavioural science and other disciplines. Thus far, most alternative models have only involved small steps from the firm foundations of RUM models. However, a key issue with models not based on RUM is the loss of its micro-economic foundations and hence the ability to conduct welfare analysis. This thesis puts forward the key argument that if we are to lose this benefit, a newly proposed model needs to allow for rich account of behaviour and steps should be taken to move further from the tried and tested. Given that choice models developed within mathematical psychology have been specifically designed to explain contextual effects where alternatives impact each other, it is surprising that there has not yet been more of a bridge between the disciplines. This is hence the key aim of this thesis: to build bridges between the disciplines by bringing ideas and models from mathematical psychology into choice modelling in the context of travel behaviour research. We provide a large number of underlying methodological improvements and adaptations for two accumulator models, decision field theory (DFT) and the multi-attribute linear ballistic accumulator (MLBA). The work in this thesis provides thorough and detailed applications of both models to a wide range of travel behaviour choice contexts, such that the precise nature of the models can be established and the models can be contrasted to standard choice modelling methods. Furthermore, we establish best practices for the implementations of both models. Crucially, tests across a wide variety of choice scenarios demonstrate that these models regularly outperform standard choice models in terms of model fit, as well as providing useful behavioural insights. We also develop new frameworks for choice models implementing quantum logic, which has made a successful transition into cognitive psychology but has not yet been discussed in depth in the context of travel behaviour research. Results from applications of quantum models suggest that they provide an accurate account of changes in choice context. Of course, adding yet more possible models to the mix creates further issues for analysts in deciding which structure to adopt. With this in mind, we discuss in depth the concept of model averaging, demonstrating its potential within the context of travel behaviour research and showing how it can be applied across a wide variety of models to generate interesting insights by combining evidence across models. Overall, we demonstrate that if we are to step away from the firm foundations of RUM, the move towards models developed in mathematical psychology will provide considerable benefits for those who make the leap.