This article presents a development and calibration of a cross classification mode choice model using socioeconomic variables and impedance values grouped according to thresholds. A detailed analysis of the 1974 and 1981 origin-destination survey data for the Madrid Area indicated that the socioeconomic characteristics of the user are of greater importance in mode choice than the relative impedance values of the available modes. A category analysis of the data led to the design of a cross classification model which disaggregated users by their social, demographic and economic characteristics, and relates these to their choice of mode. An additional analysis indicated that the marginal impedance between competing modes influences mode choice only at particular threshold values. Thus it was possible to include categories of marginal disutility as another dimension in the model. The model adopts a nested form where initially the total trips are split between the walk and mechanized modes as a function of distance. Then the mechanized trips are split into public transport and car trips as a function of generalized cost difference between the two modes. And finally car trips are divided between drivers and passengers as a function of car disutility. The model was calibrated using disaggregate trip data but with impedance values derived from a zonal network. The calibration was principally manual with a low level of statistical sophistication. The principal object of the study was to investigate the factors that influence mode choice without assuming beforehand any particular mathematical relationships. Nevertheless, the model was compared with an aggregate logit model calibrated using the same survey data, and both models were used in a study which compared alternative fare systems for Madrid.
Abstract