Date of Award

Fall 1994

Document Type

Thesis

Degree Name

Master of Science in Transportation - (M.S.)

Department

Executive Committee for the Interdisciplinary Program in Transportation

First Advisor

W. Patrick Beaton

Second Advisor

Kyriacos Mouskos

Third Advisor

Naomi G. Rotter

Abstract

This research quantifies the potential biases resulted from two different data generation methods used in transportation modeling: Revealed Preference (RP) and Stated Preference (SP) techniques. Revealed Preference technique is the conventional approach to generate data. It relies on observed or reported data of actual behavior. Stated preference technique is a new data generation method. It creates transportation scenarios using hypothetical data. Conventional studies favor the use of revealed reference. However, full description of advantages and weaknesses of using revealed preference technique is not available in literature, neither point-to-point comparison between stated preference and revealed preference techniques. This research contributes to the literature by demonstrating the relative magnitude of biases inherent to both approaches.

The method to explore approach-specific to generate data is simulation. The simulation work concentrates on biases found in RP and SP in the statistical estimation component, although biases also exist in the forecasting component. Simulation in RP case focuses on errors-in-variables; while, simulation in SP case concentrates on the internal design of the data matrix.

Based on the results from the simulations, the research points out the potential biases in two models used to forecast model shift behavior in New Jersey. The work concludes with a list of future research needs.

Share

COinS