«BASIC HUMAN DECISION MAKING: An Analysis of Route Choice Decisions by Long-Haul Truckers John Holland Knorring Advisor: Professor Alain L. Kornhauser ...»
In addition to income, the level of education that the driver has obtained is strongly correlated with the number of alternate routes considered and selected. AbdelAty et al. were able to show through their research that people’s usage of alternate routes increased as the level of education increased.18 This result is not too surprising for two reasons. First and most obvious, greater levels of income have traditionally been associated with higher levels of education. Second, higher levels of education are strongly correlated with more intelligent individuals. Intelligent individuals have a higher propensity to think through the route that they will take and as a result will consider multiple routes.
17 Khattack, A. et al. Effect of Traffic Reports on Commuters’ Route Choice and Departure Time Changes. Proc., Vehicle Navigation and Information Systems Conference, Part 2, Society of Automotive Engineers, Warrendale, PA., Oct. 1991, pp. 669-679.
18 Abdel-Aty, Mohamed et. Al. Models of Commuters’ Information Use and Route Choice: Initial Results Based on Southern California Commuter Route Choice Survey. In Transportations Research Record 1453, TRB, National Research Council, Washington, D.C., 1994, pp. 46-55.
2.5.2 Number of Alternate Routes Considered Stephanedes et al. were able to show that in general, drivers rarely consider more than two alternate routes, and when they do consider a third route it is usually only under extreme circumstance such as weather. In their stated preference study of commuters in Minneapolis and St. Paul Minnesota, fewer than 3% of the commuters considered a third alternative route.19 These results would strongly suggest that passenger vehicle drivers essentially pick between one of two routes considered. Assuming that long haul truck drivers’ cognition is the same as commuter traffic, we will use this principal of limited alternate route considerations as a basis for this thesis by only considering the downtown route and one bypass route.
When considering alternate routes there are basically only two ways that this is possible for drivers. The first method is by looking at a road map. This is the primary method used for picking routes for individuals that are not familiar with the roads around them. If an individual has a road map in front of them, they have every single alternative route possible; however, most of the routes are discarded in order to reduce cognitive overload. In general they are left with a few practical routes to choose from. From there, the individual uses his own utility function and optimizes his route by maximizing his utility. The second way in which people decide on alternate routes is by use of detailed cognitive maps that an individual may have built up over time. The problem with this method is that individuals will only have detailed cognitive maps of their surroundings if and only if they are quite familiar with the roads and have driven on them for an extended period of time. When drivers go into unfamiliar areas, they are unlikely to have the 19 Stephanedes, Yorgos et al. Demand Diversion for Vehicle Guidance, Simulation, and Control in Greeway Corridors. In Transportation Research Record 1220, TRB, National Research Council, Washington, D.C., 1989, pp. 12-20.
One of the most important factors in choosing routes as it relates to this thesis is how the individual decision makers respond to risk and their willingness to take risks.
When a driver becomes delayed on his selected route, he will, in general, attempt to determine how long he will be delayed based on his own observations of the traffic flow around his vehicle.20 The driver basically has two options. First, he can stay on the route and just build the delay into his expected time of arrival. His second option is that he can
disadvantage of switching to another route is that the driver most likely has no information about the traffic flow on the alternate route. In fact, the only way that he would have information about the alternate route is if he was told by an outside agent i.e.
radio or another driver as to the conditions on other roadways. As a result, the variance in the amount of time that it will take to travel the whole route will be greater.
When talking about individuals risk preference and risk behavior, the concept of variance comes into account. In this case with the driver stuck in traffic, if he switches routes, there is a chance that the new route is congested as well and he will spend an equal or greater amount of time stuck on the alternate route. However, the driver does not know the answer to this question before he makes his decision to switch or to stay.
One can only determine the likelihood or propensity that a driver will switch to alternate
routes given his willingness to take risks. By gambling on an alternate route when the driver already has a good idea of the time that it will take him on his current route given the level of congestion, the driver is showing that he has at least a fairly healthy appetite for risk. Khattak et al modeled this concept of risk behavior in drivers. They were able to show that a driver’s “inclination to adventure and discovery” does influence route choice decisions.21 As a result, for any study of this nature, it will be important to have an understanding of the risk preferences of the individuals involved in the study.
Obtaining individuals preference for risk is very difficult to do. It is almost impossible to determine from a revealed preference data set as revealed preference data sets only show the decisions people actually make and not why they make them or what factors they took into account to formulate a decision. Risk preferences are determined through a series of very specific questions similar to the question posed earlier about taking a gamble on one outcome that varies in quantity or choosing to take another option that has a reduced outcome, but has greater certainty in the outcome. The data set for this thesis does not include such questions, so it will be very difficult to model the behavior of drivers on an individual basis. However, it is possible in some cases to model aggregate levels of risk preference for groups of drivers based on their choice to use toll type routes.
2.5.4 Traffic Information Traffic information can come from a number of different sources: personal observations, radio, television, roadside displays, CB radio, cellular phone and even the 21 Khattak, Asad et al. Factors Influencing Commuters’ En Route Diversion Behavior in Response to Delay. In Transportation Research Record 1318. TRB, National Research Council, Washington, D.C., 1991 pg 143.
Internet.22 Khattak et al. looked at the effects of traffic information in the use of alternative routes for drivers. They suggest that drivers formulate route choice decisions in real time. In other words, drivers start off with a route in mind, but then when they see what the conditions are like on the chosen road, or find out from another previously listed source, they are then likely to make an in-vehicle decision to switch to an alternate route.
The result of this is that as new information comes in, from local observations or from another agent, the driver will make new decisions on the route to take. The problem with this method though is that it insists that traffic information be correct. For the most part, traffic reports available from news services are reasonably accurate and can be trusted, but this topic is beyond the scope of this project.
Time of day considerations would seem to clearly be a factor in determining route choice decisions. For commuter drivers who drive to work and must be there at some specific time, choosing a route to follow would be very important depending on the time of day. If a driver is familiar with the traffic patterns at different times of the day, he may be able to avoid certain traffic tangles that at other times of the day do not exist. In a study done by Mahmassani and Stephan on commuters, they simulated real life decision making by asking actual commuter drivers route choice and departure time questions.
The results of the study were that commuters would first pick a route, and then they would determine what time they had to leave their origin to arrive at their destination by
the appropriate time. It was shown that the drivers would modify their departure time by a significant amount before they would modify their route.23 In considering time of day as a decision factor for long haul truckers, the analysis is most likely slightly different. There has been very little research done in this area, however, one could speculate. Long haul truckers are similar to commuter drivers in many aspects of route choice decision-making, and they are also quite different in other areas. In the area of time of day considerations, it is fairly easy to argue that long haul truckers are much more flexible than other drivers. In general, a truck driver will pick up a load and then have a scheduled time of delivery for the load. For long haul truckers this may be a few days in the future. Because the law limits truck drivers in the number of hours per day that they can drive, the truckers will want to pick the optimal times of the day for them to drive so that they can cover the most distance in the shortest period of time. Getting caught in a morning or evening rush hour is not a productive use of a trucker’s time, not to mention being in a traffic jam uses extra fuel and causes additional wear on the machinery.
2.5.6 Congestion In the area of route choice decision-making, there have been numerous studies that have looked at congestion as a factor of route choice decision-making. Todd Burner ’99, as an example, looked at congestion on alternate routes via perceived speed curves as influences on the route choice decision process. Burner raised an important question in 23 Mahmassani, Hani and Douglas Stephan. Experimental Investigation of Route and Departure Choice Dynamics of Urban Commuters. In Transportation Research Record 1203, TRB, National Research Council, Washington, D.C. 1988, p 69.
his analysis of perceived speed curves. Studies had shown that drivers, when either they observed congestion or were given actual congestion information, they modified their routes.24 Burner suggested that not only observed congestion or information about congestion ahead were the only factors in route choice decision making, but almost more importantly, the perception of congestion either ahead on the chosen route or on alternative routes was a very important decision factor in the utility functions. If one assumes this to be true, then the components of the Utility Function are not limited to observable factors and people’s perceptions and feelings also contribute to the utility function.25 There has been some work done to try to explain this decision making process.
Chee-Chung Tong attempted to explain this by using behavior decision theory. This theory stipulates, “An individual’s judgment and choice are two integral parts of the same process.”26 It is really quite simple. First, people perceive what is going on around them and formulate some sort of judgment. Essentially, people do not react or make decisions based on what is actually happening, rather they make decisions based on what they perceive is happening. The result as it pertains to this study is that there are other factors involved in the decision making process that contribute to the decisions that are being made that are going to be difficult to quantify.
24 Vaughn, Kenneth et al. Route Choice and Informatiokn Use: Initial Results from Simulation Experiments. In Transportation Research Record 1516, TRB, National Research Council, Washington, D.C., 1995, pg 61.
26 Tong, Chee-Chung et al. Travel Time Prediction and Information Availability in Commuter Behavior Dynamics. In Transportation Research Record 1138, TRB, National Research Council, Washington, D.C., 1987, pg 1.
2.5.7 Additional Factors It is safe to assume that the seven previously stated factors are not the only factors that contribute to the utility function. However it is also likely safe to assume that these seven factors contribute a majority of the characteristic shape of the utility function as well as explain a large portion of its variance. In other words, these factors provide a very good first approximation. Other factors that could be considered in future studies that might influence decision-making and further quantify the characteristic shape of the utility function are: the scenery or type of neighborhood that the route goes through, the quality of the road, the number of turns along the route, or the number of rest areas along the route.
2.6 Work done by Todd Burner ‘99 The work done by Todd Burner has proven to be very helpful in the writing of this thesis. He did excellent work in setting a framework for congestion analysis. His thesis primarily focused on route choice decision making with anticipated congestion on alternate routes. He accomplished this by looking at a number of different case studies across the U.S. highway network. He chose the cases based on a few different criteria.
He wanted to look at cities that had a major heavily traveled highway that passed through the central business district (CBD) of the city. In addition, each case city must have also had a bypass route around the downtown area that for the most part has a more limited access than the downtown route.
After selecting the case cities, Todd then developed perceived speed curves for the downtown route in relation to the bypass route. The method that he used to determine these perceived speed curves will be covered in greater detail later, as the same method will be used in this thesis. Using the results of the Logit model, he forecasted the expected percentage of traffic that would use the bypass route given the perceived speeds on the downtown route. If the perceived speeds on the downtown route were lower than the posted speed limits, the reduction in speed was considered a result of congestion.