DRIVER REACTION TO UNRELIABLE
TRAFFIC INFORMATION
While transportation engineers would like to provide reliable traffic information
to motorists, the highway system, at times, has operating situations that make it
difficult to achieve this goal. Congestion, delays, and accidents can sometimes make
information provided to motorists unreliable when it is received. Such unreliability
may cause drivers to discount, or even ignore, traffic messages displayed on programmable
road signs or other information-delivery systems. Currently, there are few results
based on empirical studies to guide the highway engineer concerning what level of
accuracy is needed to gain driver acceptance and trust. In some domains, a single
bad experience is enough to prevent people from using a service or machine again.
For example, few people continue to put coins into a defective vending machine. Traffic
information systems must be sufficiently reliable so that motorists continue to accept
and use the systems. The goal of this research is to acquire data that the highway
engineer can use to select a level of information reliability that will maintain the
driver's acceptance and use of route guidance information.Method
The Battelle Route Guidance Simulator was used to collect data in two experiments.
This simulator consists of two linked 486 computers driving two displays (figure 1).
One display provided a real-time video of Seattle traffic that had been digitized
and stored on hard disks. The other display featured a map of Seattle with a touch
screen. The driver used this screen to obtain traffic information and to select traffic
links. A moving dot represented the driver's vehicle. When the dot entered a traffic
link, the appropriate video was rapidly retrieved from the hard disk and displayed
in real time on the other monitor.
In the first experiment, 26 links were displayed, giving the driver 29 possible
routes between the origin, Westlake Center in downtown Seattle, and the destination,
Bellevue Square Mall. The second experiment used 31 links and 33 possible routes.
These routes traversed a variety of roads, including congested city streets, four-lane
State roads, and an interstate highway in an urban setting. Links had either light
traffic (level of service A) or heavy traffic (level of service E or F). Due to the
topography of Seattle, the driver must cross Lake Washington to reach the destination.
Since there are only two bridges across the lake, the experimenters retained some
control over traffic congestion encountered by the driver regardless of the route
selected. Simulated trips took about 22 minutes in light traffic and 37 minutes in
heavy traffic.
In the
first experiment, 48 drivers were tested; and in the second experiment, 24 were tested
(table 1). All drivers were familiar with the Seattle area and drove on Seattle highways
at least twice per week.The driver's goal was to reach the destination as quickly
as possible by choosing links that he/she thought had the least amount of traffic
and the shortest travel time. A penalty was assessed for selecting links that were
not optimal. The maximum penalty per trip was $13.59. In Experiment 1, drivers were
charged $0.10 to query a link. Link information was free in Experiment 2. Furthermore,
an additional penalty was assessed whenever a driver encountered heavy traffic. These
penalties simulated the effects of traffic delays on the road.
Results
Figure 2 shows how
the penalty changed with the reliability of traffic information. In the first experiment,
traffic information whether traffic flow on the selected link was light or heavy was
either 100 percent or 77 percent accurate. In the second experiment, traffic information
was either 100 percent, 71 percent, or 43 percent accurate. In both experiments, drivers
reached their destination faster and had lower penalties when traffic information
was accurate. Thus, drivers were able to benefit from using the traffic information
system. One possible explanation for the results is that drivers are only willing
to purchase or to use traffic information when it is 100 percent accurate. However,
in the first experiment, drivers in both reliability conditions spent the same amount
of money for information. In the second experiment, information was free. Therefore,
this possible explanation is not satisfactory, and the benefit of using the traffic
information system is real.
Figure 3 shows how driver trust in the simulated traffic information system,
rated on a scale from zero to 100, changed over the four trials of the first experiment.
For the first two trials, when information was 100 percent accurate, trust increased.
But for the last two trials, when information was 77 percent accurate, trust decreased
after participants drove over inaccurate links. However, trust was restored by subsequent
accurate links.Figure 4 shows driver trust for the second experiment. Trust was
high for the first two trials when information was 100 percent accurate. Trust decreased
on the last two trials, especially for the less accurate (43 percent) condition. Again,
trust was restored for links on which accurate information was presented. Furthermore,
the trust decrement relative to 100 percent accurate information was small for the
71 percent condition.
Study Implications
The following points may be useful for highway engineers who provide traffic information
to drivers:- Drivers use and benefit from accurate traffic information.
- Information that is less than 100 percent accurate can be useful.
- Driver trust in an unreliable system recovers when accurate information is presented,
but the recovery is not always complete.
- Traffic information reliability above 70 percent is recommended.
For More Information
This research was conducted by Battelle.
Form more information, contact:
M.
Joseph Moyer
Engineering Research Psychologist, HSR-30
703 285-2008.
Back to Human Factors