Passenger Wait Time Perceptions at Bus Stops: Empirical Results and Impact on Evaluating Real- Time Bus Arrival Information

This study quantifies the relationship between perceived and actual waiting times experienced by passengers awaiting the arrival of a bus at a bus stop. Understanding such a relationship would be useful in quantifying the value of providing realtime information to passengers on the time until the next bus is expected to arrive at a bus stop. Data on perceived and actual passenger waiting times, along with socioeconomic characteristics, were collected at bus stops where no real-time bus arrival information is provided, and relationships between perceived and actual waiting times are estimated. The results indicate that passengers do perceive time to be greater than the actual amount of time waited. However, the hypothesis that the rate of change of perceived time does not vary with respect to the actual waiting time could not be rejected (over a range of 3 to 15 minutes). Assuming that a passenger’s perceived waiting time is equal to the actual time when presented with accurate real-time bus arrival information, the value of the eliminated additional time is assessed in the form of reduced vehicle hours per day resulting from a longer headway that produces the same mean passenger waiting time. The eliminated additional time is also assessed in the form of uncertainty in the headway resulting in the same extra waiting time. Naturally, such benefits of passenger information can Journal of Public Transportation, Vol. 9, No. 2, 2006 90 only be confirmed when the actual effect of information on the perception of waiting time is quantified. Motivation and Hypothesis The background motivating this study is first discussed, and the objective is then presented. Real-time bus arrival information—for example, delivered to prospective passengers waiting at bus stops via variable messages signs (VMSs)—can be useful to transit passengers for a multitude of reasons. Passengers can use their waiting time more productively, select which route they would want to take, or choose to select an alternative mode of transportation. Whatever the prospective passengers’ choices are, providing them with real-time information reduces the uncertainty inherent to transit systems. Empirical evidence shows that the time travelers spend outside the transportation vehicle of choice (e.g., waiting at a stop) is more onerous than the time they spend inside the vehicle in motion to their destination (Ben-Akiva and Lerman 1985). This is partly due to the higher degree of uncertainty associated with waiting for a transit vehicle. This phenomenon is well characterized by Duffy (2002): “People don’t mind waiting for a bus if they know how long it’s going to be. Even if they have to waste the time, at least they know it’s going to be 15 minutes. Otherwise they’re sitting there thinking the bus will be along in about two minutes, and when it doesn’t show, then they start getting frustrated.” In general, reducing waiting time uncertainty is expected to improve passenger satisfaction, and ultimately increase bus ridership. Mishalani et al. (2000) studied the value of information to passengers in terms of using the waiting time more effectively, while Hickman and Wilson (1995) studied the value in terms of improved route choice. This research focuses on passengers’ perceptions of their waiting time at stops (outside the vehicle) and, as a result, the possible reduction in such times when real-time passenger information is provided. To study the perceptions of waiting time, a survey of prospective passengers at bus stops was conducted. The collected data were then analyzed whereby the relationship between perceived and actual waiting times was investigated. The main hypothesis of this study is that without real-time bus arrival information, passengers are likely to perceive waiting time to be greater than it actually is. When accurate bus arrival information is provided, it is assumed that passengers will perceive their waiting time to be equal to the actual waiting time. In this case, a passenger will arrive at a stop and look at the VMS, which will display the minPassenger Wait Time Perceptions at Bus Stops 91 utes until the next bus is expected to arrive on the route of interest. The VMS will continue to update the expected time while the passenger waits. Without bus arrival information provided to passengers, the relationship between perceived and actual waiting times is expected to follow a function where the perceived time is greater than the actual time. The form of the function might depend on the magnitude of the waiting time. For example, the additional time due to exaggerated perceptions may be further magnified under long waits in comparison to short waits. One can also imagine the opposite situation, where longer waits may be perceived more accurately due to the more conscious recognition of time under such conditions. The main objective of this study is to model and quantify the difference between perceived and actual passenger waiting times at bus stops in the absence of accurate real-time bus arrival information and to investigate the effects of duration of the actual waiting time and socioeconomic variables on this difference. This objective is achieved in the context of a pilot study by estimating models that describe waiting time as perceived by passengers waiting for buses at stops. To do so, appropriate field data were collected. Once the difference between perceived and actual waiting times were quantified, possible factors that might affect the magnitude of the difference were explored. Moreover, an analysis of the potential benefits of providing accurate bus arrival time information at bus stops was carried out. Data Collection Data were collected by surveying passengers waiting at bus stops for Campus Area Bus Service (CABS) buses, which are operated by the Transportation and Parking Office of Ohio State University in Columbus. CABS serves the campus community, which includes close to 50,000 students, resulting in an annual ridership of approximately 4 million. The operation consists of 15 to 20 40-foot buses running simultaneously on several routes of lengths ranging from 2 km to 8 km on and in the areas surrounding the campus. The transit service used in this pilot study is small enough to be manageable, yet large enough to reflect situations pertinent to more extensive transit services in urban areas. Nevertheless, it would be important to build upon this pilot study in future research by examining larger transit systems. Extensions to larger systems would render the findings more applicable to a wider set of conditions, most notably, longer routes and more heterogeneous traveling populations. Journal of Public Transportation, Vol. 9, No. 2, 2006 92 Three students surveyed 83 passengers over a period of approximately one year, from spring 2001 to spring 2002. A surveyor went to a bus stop, noted the arrival time of a passenger, and later asked him or her a series of questions. A response rate close to 100 percent was achieved. Bus Stops A set of appropriate bus stops was first determined for the purpose of conducting the surveys on the basis of four criteria. The first criterion was to choose a stop that does not serve many routes. In fact, a stop that serves only one bus route is ideal. Fewer routes will help the interviewer know, or possibly guess, which route a passenger is going to choose before he or she gets on the bus. A surveyor sitting at a stop has a general idea of when the next bus will arrive based on both the published headway and observations of the buses over a period of time. If a bus stop serves many different routes, it becomes more difficult, if not impossible, to know the route a random passenger plans to use. The passenger might then board a bus before the surveyor has a chance to conduct the interview, thus missing a data collection opportunity. The second criterion for selecting a good bus stop for surveying purposes is to use a stop that serves routes with longer headways. Longer headways are attractive because it is desirable to have the option to survey a passenger after a notable wait. While data with short waiting times are needed for a complete data set, not all the data should be collected after a 3to 4-minute wait. A longer headway allows the interviewer more options on when to survey passengers and observe longer waits. One issue with longer headways, though, is whether a schedule is published. This issue leads to the third criterion. It is helpful to select bus stops where passengers arrive totally randomly. Such arrivals typically occur when only the headway on a route is published, rather than the scheduled time of bus arrivals. When schedules are published and headways are long, most passengers will likely arrive shortly before published arrival times, thus reducing the opportunity of observing relatively longer waiting times. Finally, it is productive to collect data at stops with relatively high demand. Passengers must arrive frequently at a bus stop to ensure that interviewers do not experience a large amount of idle time. Otherwise, longer survey times will be needed to produce the same number of observations. Based on the above criteria, a total of five stops on three routes were selected. Two of the routes have published headPassenger Wait Time Perceptions at Bus Stops 93 ways of 10 minutes and one a published headway of 15 minutes. None, however, has a published schedule. Interviews and Observations During the survey process, when a passenger arrived at the bus stop, the arrival time was noted. The first one or two passengers that arrived after the previous bus departure were selected. This selection strategy ensured that passengers with a wider range of wait times were interviewed and that the interviewer was able to survey them without the bus coming before the interview was complete. The interviewer decided when to survey the passenger. This is selected largely on when the next bus was thought to be arriving. The interviewer typically began the interview with a passenger at least a minute before the expected bus arrival to ensure enough time to complete the interview without the passenger feeling anxious about catching the bus. As already discussed, the interviewers had a good idea as to when the next bus would arrive because of their knowledge of the previous bus arrival time and service operations. In addition to the surveyor’s name, date, and weather conditions, the following is a list of variables observed for each interviewed waiting passenger: 1. origin of the passenger (bus stop where the survey is conducted); 2. passenger’s arrival time to the stop; 3. time the passenger was surveyed; 4. passenger’s gender; 5. passenger’s race; 6. passenger’s perceived waiting time; 7. whether the passenger was wearing a watch; 8. destination of the passenger; 9. maximum time the passenger would be willing to wait were real-time bus arrival information provided; 10. approximate walking time to the destination from the bus stop at which the passenger was waiting; 11. whether the passenger had a time constraint (such as making it to a class at a certain time); 12. familiarity of the passenger with the transit service, as measured by frequency of use in number of trips per day; Journal of Public Transportation, Vol. 9, No. 2, 2006 94 13. passenger’s car ownership status; and 14. passenger’s affiliation with the university (undergraduate student, graduate student, staff, faculty, or visitor). Items 1 through 7 were always recorded by the surveyor, while items 8 through 14 were generally collected, time permitting. The first question, asked after a brief introduction, was “In minutes, how long do you think you have been waiting?” It was important to specify the desired level of accuracy to the passengers to avoid the possibility of their rounding off to the nearest 5 minutes. Perceived Waiting Time Models The model development and estimation aimed at quantifying the difference between perceived and actual waiting times, along with the exploration of factors influencing that difference, are discussed in this section. The modeling consists of two parts. First, a simple ordinary least squares (OLS) regression model of perceived versus actual waiting times is estimated. Hypothesis tests are then applied to determine whether a significant difference between the two variables exists and to assess the nature of the relationship. Second, the impact of socioeconomic variables on the relationship between perceived and actual waiting times is investigated. Perceived versus Actual Waiting Times A simple linear regression model of the following form is estimated: p = β0 + β1a + ε (1)