Alternative specifications of the random disturbances for trip frequency models

The purpose of this paper is to present alternative functional specifications for models of shopping trip frequency and to illustrate the influence of the modelling assumptions on the interpretation of the determinants of trip frequency. The data used for this analysis is a 23-day diary of shopping travel by able bodied elderly individuals in Lawrence, Massachusetts. The alternative models are, in addition to ordinary least squares, an integer dependent variable model, and an error component model of a time-series of cross-sections.The findings suggest that, when models are developed that consider explicitly the discrete nature of the daily trip generation variable (i.e. the number of trips taken by an individual on a given day), forecasts which are not significantly different from the ordinary least squares forecasts are obtained.