Most modal split models have been based on the assumption of rational behaviour in an individual's choice evaluation of the generalised costs of modal alternatives. This paper integrates conceptual and empirical information from a wide range of sources and points towards an alternative way of looking at modal choice. The main conclusion is that the car is usually perceived as the superior mode for vehicular travel and that the potential user is committed to its use largely through the act of purchasing it. The conceptual structure of a sequential modal split model is outlined as one that is based on a four-stage decision-making framework which considers the role of learning and habit-formation. In the conclusion, the implications of this approach are considered in terms of the conventional modal split and trip generation submodels, and certain policy measures are assessed.
Utility Theory and Learning Theory
The nature of the choice decision is fundamental to modal split modelling. For journeys over very short distances, walking is a mode available to almost all the population; but, over longer distances (> 1 kin), the availability of public transport and vehicular private transport is essential to movement. The decision to use public transport immediately imposes severe restraint on movement except in high density urban areas where there may be frequent services. Such a decision can be conceptualised in the conventional theory of rational choice behaviour where a number of modes are available to a given individual, each characterised by certain attributes. The individual weights accorded to each of these alternatives are based upon his own system of preferences, reflecting his own tastes and any external constraints placed on his freedom of choice -he then selects the alternative that maximises his own utility in terms of expected outcomes from his actions (e.g. . In recent years utility theory has been one of the main methods by which the decision-making process has been modelled to represent modal choice.
There are two main problems with this approach. First, expected utility measures are based upon how trip makers should make decisions rather than