It is well known that growth-factor methods of predicting the distribution of traffic from one zone to another have the disadvantage that predictions based on one definition of the zoning system will not be consistent with those based on another definition. The authors show that the gravity model does not necessarily have this disadvantage. By suitable averaging of the interzonal costs, the predictions made after aggregating zones into larger units will be consistent with the predictions made with the original zones. The gravity model is therefore said to have the property of compressibility. The authors also point out that if data for some zones or even just some interzonal transfers is missing, the predictions made by both the gravity model and the Furness growth-factor model can nonetheless be consistent with the predictions that would have been obtained had that information been available. As data can be excluded without affecting the results, the models are said to have the property of excludability. The paper also discusses a number of practical ramifications that a knowledge of these properties allows.
Samenvatting