Abstract:
In many published studies involving discrete class, a large class of models has been utilised without apparent consideration of either the statistical or the behavioural relationships that exist between different model structures. In this paper the authors consider a number of alternative ways of modelling systems of discrete choice. Joint labour force participation and the fertility decision problem are used as an example. The focus is on the outcome of two decisions: the decision to have children, and whether to work -- full-time, part-time, or not at all. Unlike the majority of work in this area, a class of discrete choice models is subjected to a series of non-nested tests.