| This dissertation examines whether the positive experience with High Speed Rail (HSR) in Europe and Japan is likely to be transferable to North America without significant modifications. It studies how important the access advantage was to the success of the Japanese system.; A utility impact analysis revealed that access and egress related variables account for 26---43% of the total absolute utility of typical air and rail travel. Terminal fixed effects were found to be about four times more important for HSR than for air. They capture not only accessibility attributes, but also the attractiveness of the terminal's location.; If the utility of HSR travel in the United States would depend on the magnetism of large downtown stations as much as it appears to in Japan, innovative solutions must be found to replicate this effect. Two solutions are outlined in the first and last chapters. They require a paradigm shift to convergence, seeing urban, regional, and high speed ground transportation as one coherent system.; Almost 50 000 air and rail observations of the 1995 Intercity Travel Survey could be used to estimate separate airport pair and HSR station pair choice models. The HSR choice set consisted of 1 260 station pairs and had to be randomized. Weights compensate for heteroskedasticity and extend the sample to a typical Fall day with daily expansion factors.; The small air and large HSR coefficients estimated with nested logit suggested that a new airport would mostly draw passengers away from other airport pairs, while adding HSR stations would have a high impact on other modes.; Methodologically, the research found that a log plus linear specification for access and egress distance is often called for. Linear, quadratic, and cubic functions underestimated the disutility of feeder distance at the median and mean by up to two thirds.; When randomizing choice sets the researcher should only draw conclusions from a set of models, each using a different number of alternatives, to detect converging values and standard errors likely to be incorrect. Using this approach, strong evidence of a threshold effect for access and egress started to emerge. |
