Stated Choice for Transportation Demand Management Models: Using a Disaggregate Truth Set To Study Predictive Validity

Abstract

Discrete choice models have expanded the ability of transportation planners to forecast future trends. Where new services or policies are proposed, the stated-choice approach can provide an objective basis for forecasts. Stated-choice models are subject to a range of experimental error not found in revealed-preference designs. Primary among the concerns facing researchers is the ability of respondents to understand and operate on hypothetical choice scenarios in a manner that will reproduce choices made under actual situations. These concerns are specified in the scaling factor. Estimation of the scaling factor has proceeded through various ways to link actual decisions to comparable decisions made under hypothetical conditions. However, where the alternative is new, real decision data are not available. The level of error incorporated in a study where no real-world information on the scaling factor is available is examined. The test of predictive validity focuses attention on the switching behavior of commuters at a single employment site. The results indicate that switching behavior between single-occupant vehicle and high-occupancy vehicle modes is forecast within 1 percent by stated-choice techniques and within 10 percent by backcasting techniques with revealed-preference data.