Occupational stratification of urban work-trips in metropolitan Adelaide, 1971

Abstract

This study examined the suitability of the gravity-based formula for trip distribution which was proposed by the U.S. Bureau of Public Roads (1965) for use in modelling the journey to work in Adelaide in 1971. The BPR formula was developed into a computer programme which can be calibrated for data on trips between the Local Government Areas of Adelaide. The model was assessed, as a predictor of all work-trips then the trip data was separated into male/female and several occupational categories each of which was modelled separately. The accuracy of prediction for each category was assessed then the predicted numbers of trips were aggregated to assess the benefit of separately modelling several categories of worker. The major findings of the study were: (I) Actual patterns of trip distribution can be modelled to a very high level of correlation by the Gravity Model; provided reliable data is available on travel-times; and numbers of trips originating and terminating per zone. (2) Numbers of trips to the CBD can be reliably estimated by the Gravity Model; (3) outer suburbs cannot be modelled as accurately as the more stable inner suburbs; (4) Small increases in accuracy of modelling can be created by separation of data into occupational categories; and (5) Commonly used goodness-of-fit statistics such as chi-squared values are not reliable indicators of the accuracy of a model. Correlation analysis proved to be a most useful measure of goodness-of-fit and high levels of correlation were found between square roots of observed and estimated numbers of trips. It was also noted that: Workers in different occupations and different sexes can have noticeably different work-trip distributions, but the gravity model can be calibrated to replicate a wide range of distributions.