Driver Response to Guidance: a Comparison of Data from a Route-choice Simulator and a Self-completion SP Exercise



Driver Response to Guidance: a Comparison of Data from a Route-choice Simulator and a Self-completion SP Exercise

Authors

PALMER I A, BONSALL P, University of Leeds and SHIRES J, University of Oxford, UK

Description

Transport researchers have, in recent years, paid increasing attention to the prediction of drivers' response to new forms of information and guidance such as Variable Message Signs (VMS) and in-car navigation/route guidance aids.

Abstract

Transport researchers have, in recent years, paid increasing attention to the prediction of drivers' response to new forms of information and guidance such as Variable Message Signs (VMS) and in-car navigation/route guidance aids.

Prominent among the tools used in this research are route choice simulators designed primarily to explore the impact of information and guidance systems on driver route choice.. They overcome, to a large extent, the difficulty, or in some cases impossibility, of gathering data on real-world response to systems that are not yet in widespread production or use. Also, compared to more conventional questionnaire techniques, they can give the subjects a more realistic impression of the consequences of their decisions: However, route-choice simulators are more expensive to develop and use than conventional Stated Preference (SP) techniques and so the question arises: is the extra expense involved in using a route-choice simulator worthwhile?

This paper compares data from a matched pair of studies of drivers' response to VMS messages and other forms of guidance: the first used the VLADIMIR route choice simulator; the second used a self-completion SP questionnaire designed to replicate the VLADIMIR exercise as closely as possible. The analysis, involving the specification of mixed data set discrete choice models, addresses the following questions:

* do the two techniques produce comparable data and can the two data sets be combined to produce hybrid models?

* is there any significant difference between models constructed from data from the two sources?

* is the extra effort in developing a simulator justified?

Publisher

Association for European Transport