The design of high occupancy vehicle (HOV) lanes and other HOV facilities depends on volumes of HOVs expected to use the facilities. Currently, there is no good method for predicting vehicle occupancy on specific highway facilities. The basic objective of this research was to incorporate the best of previous transportation models with new information on psychological and demographic determinants of mode choice into a model that forecasts vehicle occupancy for specific highway facilities. In order to do this, four different data sets were investigated. Each one allowed the investigation of one or more aspects of a comprehensive model to forecast vehicle occupancy. The basic approach tested in this research was the "integrated model of consumer choice," first proposed by Tybout and Hauser in 1981. It incorporates a wide variety of factors involved in transportation decision-making. Most of the findings from the four data sets analyzed for this study were consistent with the literature and with each other. The research concluded that, in order for a transportation model to adequately forecast vehicle occupancy, it should take into account that 1) mode choice changes over time, 2) attitudes and perceptions are important in mode choice, 3) attitudes and perceptions can be influenced by experience, 4) household composition is important, 5) commute length varies by type of job and location, 6) constraints influence the process in complicated ways, and 7) two person carpools are different from larger carpools. The dynamic aspects of mode choice are critical to understand. Until we have more good time series data and the ability to adequately understand it, our models of mode choice and vehicle occupancy will be deficient.
Washington State Transportation Center (TRAC)
Demographics, Forecasting, High occupancy vehicle lanes, High occupancy vehicles, Mathematical models, Mode choice, Models, Psychological aspects, Transportation, Travel demand management, Vehicle occupancy.