ABSTRACT
The purpose of this paper is to unite two current approaches to the modelling of dispersed spatial interaction behaviour — namely the entropy-smoothing approach (Erlander, 1977, 1980, Boyce et al, 1981, 1983) and the cost-efficiency approach (Smith, 1978-A, 1983-A). At first glance, these approaches appear to be quite different. For entropy-smoothing is deterministic in nature and treats spatial interactions as continuous flows, while cost-efficiency is probabilistic in nature and treats spatial interactions as discrete events. But further reflection suggests that both approaches are quite similar in spirit. For in each approach, spatial actors are essentially regarded as cost-minimizers who, in the presence of other behavioural factors, tend to exhibit some deviation from pure cost-minimizing travel behaviour. Hence, in spite of their formal differences, both approaches stem from a common view of spatial interaction behaviour.
