METHODOLOGY FOR SHORT-RANGE TRAVEL DEMAND PREDICTIONS. ANALYSIS OF CARPOOLING INCENTIVES

This paper presents a methodology for predicting changes in travel patterns for short-range transport options, including carpooling incentive policies. The methodology is based on the application of disaggregate travel demand models. These models are based on the multinomial logit probabilistic choice model. The data used to estimate the coefficients of these models are taken from home interview surveys and represent a cross-section of households in Washington, DC, USA. The dependent variables of the models are the reported travel choices made; the independent variables are socio-economic characteristics measures of travel times and costs, and survey estimate of employment and land use characteristics in the urban area. The models represent the direct and indirect effects on travel behaviour: (1) shifts in mode of work trips from driving alone and transit to carpooling; (2) alternative use of cars left at home for non-work travel; and (3) changes in car ownership level. Results results are presented of predicting from a case study application of this methodology to various carpooling related policies. The paper concludes with a summary of major findings and recommendations for further improvements in carpooling and short-range demand prediction methods. /TRRL/