Stated Preference Analysis of Driver Route Choice Reaction To Variable Message Sign Information

Highway Authorities in many parts of the world have, for some years, been using variable message panels mounted above or beside the camageway to communicate short messages to motorists. Most such applications have been concerned with hazard warning and speed advice. However, their use to deliberately affect route choice is an area of great current interest. It is recognised that they have a potential role in managing demand to match the capacity available, not only to alleviate acute problems caused by roadworks and accidents, but also to contribute to satisfactory performance of networks operating close to capacity over extended periods of high, but variable, demand. The installation and operation of the panels is not cheap and there is a widespread belief that overuse, or inappropriate use, of the messages may lead to them losing their credibility with the motorists and thus ceasing to be effective. It is therefore very important to understand the likely response of motorists to various messages before displaying them and even before selecting sites for the installation of panels.

A number of researchers have explored drivers' responses to traffic information and route advice offered via variable message signs (VMS). Evidence from traffic counts suggests that messages can persuade somewhere between 5% and 80% of drivers to divert. Clearly this range of estimates is far too wide to support the use of VMS for fine tuning the pattern of demand. A major contributor to the uncertainty, however, is the varying, and often unknown, proportion of drivers whose destination makes the message relevant to them. More detailed studies involving driver interviews downstream of the VM!3 site to determine the relevance of the message, as well as the response to it, include those by Kawashima (1991) and Durand-Raucher et al. (1993). These studies have produced more precise estimates of compliance but the results are obviously limited to those messages which were on display at the time the interviews were being conducted.

A number of researchers have sought to overcome this restriction by examining response to a range of messages presented via a stated preference exercise (see for example Hato et al., 1995; Shao et al., 1995 and Bonsall and Whelm, 1995), via a route-choice-simulator (see for example Firmin, 1996; Bonsall and Merrall, 1995 ; Bonsall and Palmer, 1997) or via a full scale driving simulator or system mock-up (see for example Mast and Ballas, 1976 and Brocken and Van der Vlist, 1991). This research has suggested that response is highly dependent on message content, subjects' network knowledge, and on the extent of any implied diversion.

We see particular value in extending this earlier work to consider a wider range of messages and to determine whether the route-choice-simulator results can be repeated and extended using a somewhat cheaper methodology - namely stated preference analysis. The objectives of the work reported in this paper were thus:

to extend to our existing database on drivers' response to traffic information and route advice provided in variable message signs, to include a wider range of messages.

to construct explanatory models of drivers' route choice behaviour in response to a variety of messages

to explore the factors influencing this response

to compare these results with previous results obtained using a variety of data collection methods

to draw policy conclusions, where appropriate, on the use of variable message signs to influence drivers' route choice

to draw conclusions, where appropriate, on our data collection and modelling methodology.