A Literature Review of the Passenger Impacts of Real-Time Transit Information

A Literature Review of the Passenger Impacts of Real-Time Transit Information

Abstract

Recently, it has become common practice for transit operators to provide real-time information (RTI) to passengers about the location or predicted arrival times of transit vehicles.  Accompanying this is a growing body of literature that aims to assess the impacts of RTI on transit passenger behavior and perceptions.  The objective of this research is to compile a comprehensive literature review of studies that assess the passenger impacts of RTI provision.  The results suggest that the primary behavioral changes associated with providing RTI to passengers pertain to decreased wait times, reductions in overall travel time due to changes in path choice, and increased use of transit.  RTI may also be associated with increased satisfaction with transit service and increases in the perception of safety when riding transit.  Important areas for future research were also identified, including assessing actual behavioral changes of path choice of transit riders and conducting cost-benefit analyses post implementation of RTI systems.  The primary contribution of this study is a comprehensive review of the passenger benefits of RTI, which has immediate implications for public transit operators considering implementation or expansion of RTI systems.

Keywords

Public transit; real-time information; wait times; ridership; path choice

1. Introduction and Motivation

The transit industry has benefitted from numerous technology changes over the past two decades. One example is the widespread availability of real-time information, which provides the precise position or predicted arrival time of a transit vehicle at a stop or station.  Real-time information is typically used for operations and control purposes by the transit provider, but it is increasingly distributed directly to passenger via signage at stops or stations and through web-enabled or mobile devices (Schweiger, 2011).  As this industry practice has increased, the body of literature evaluating the passenger impacts of this new information source has also grown, which presents an opportunity to synthesize and compare findings.  Synthesizing initial trends is particularly important for transit providers who want to understand the impacts of real-time information to appropriately plan service and to weigh trade-offs between investments.  Subsequently, this study aims to conduct a comprehensive literature review of studies pertaining to passenger impacts of real-time information.

2. Terminology and Theory

Real-time information refers to up-to-the-minute tracking of transit vehicles by automatic vehicle location systems or track circuit systems. Vehicle location information is sent to a central server, typically located at the transit provider, and then it is disseminated to riders, either directly or through application programming interfaces (API) used by third party software developers.

2.1 Terminology

Although real-time information (RTI), real-time transit information (RTTI), real-time passenger information (RTPI), and advanced passenger information systems (APIS) are all commonly used terms in the literature, for this paper, the term real-time information (RTI) will be used.

A few distinctions between RTI and other forms of transit information should be made. Schedules refer to the predefined location and time of vehicles published by the transit operator. When transit vehicles are running on-time, schedule information and RTI are the same, but when transit vehicles are not, RTI is a more accurate method of tracking the actual location of transit vehicles.  Furthermore, transit service alerts are notifications that report major delays, and while these are often provided to passengers in real-time as incidents occur, they can include varying levels of information. There is prior literature that focuses on the passenger impacts of alerts about major delays (e.g., Bai and Kattan, 2014); however, this is not reviewed here.

2.2 Theoretical Impacts of RTI on Passengers

When considered from the passenger perspective, RTI can impact numerous transportation decisions made by the traveler.  Theoretically, RTI could impact a person’s decision to make a trip (travel choice), the decision to take transit for that trip versus another mode (mode choice), the specific path that the person takes on transit (route choice), the stop at which the person boards a transit vehicle (boarding stop choice), and the time at which a person leaves his/her origin to arrive at that stop (departure choice).  This decision-making process provides a passenger-centric framework that is shown on the left side of Figure 1.  Although the diagram in Figure 1 is shown with a nested structure, this decision-making process may be sequential or simultaneous.

[Figure 1 near here]

Because an individual’s decision-making process can be difficult to capture, the behavioral outcomes associated with RTI are often easier to quantify.  The right side of Figure 1 represents some of the most important behavioral outcomes that could be associated with using RTI.  RTI could impact levels of transit use by impacting a passenger’s decision to make a trip or not make a trip (travel choice) or by impacting a passenger’s choice to take transit versus another mode (mode choice).  Both of these impacts can be quantified as increased (or decreased) transit use.  RTI could also influence which route a passenger chooses (route choice) and therefore impact their overall travel time, as different paths typically have different travel times.  Likewise, RTI could play a role in the decision of which stop at which a passenger boards a transit vehicle (stop boarding choice) or what time they choose to leave their point of origin (departure time choice), and both of these could impact the traveler’s total travel time.  Travel times are also believed to influence transit ridership, creating a cyclical relationship between the behavioral impacts.  Because the prior literature often divides transit travel time into wait time and in-vehicle travel time components, the wait time component is presented separately in the literature review that follows.

The framework shown in Figure 1 only displays the transit passenger decision-making process and subsequent behavioral implications of RTI, but it does not include changes to passenger feelings and perceptions.  Impacts on passenger feelings, such as satisfaction and perception of safety, will also be considered in this literature review.

3. Methodology

Studies published in peer-reviewed journals or conference proceedings over the last 20 years (since 1995) are categorized based on the impact on passengers.  Technical reports (such as Mehndiratta et al., 2000) for which peer review status was undetermined are not included in this review.  Each study is summarized in a table using five key dimensions.

The first dimension is the mediathrough which RTI is provided to the transit rider.  Initially, RTI was frequently provided via stationary signage located at bus stops or in train stations, which are sometimes referred to as countdown clocks.  Variable message signs, such as the one shown in Figure 2, display the location of the transit vehicle or a predicted arrival time (Schweiger, 2003).  RTI is increasingly provided to passengers’ personal devices, including websites accessed on computers or mobile phones, text messages to cell phones, and smartphone applications, such as the iPhone and Android applications shown in Figure 3.  There is a small literature that aims to understand which media of RTI transit riders prefer (e.g., Caulfield & O’Mahoney, 2009); however, this is not reviewed here.

[Figure 2 near here]

[Figure 3 near here]

The second and third dimensions are the modeof transit and the locationwhere the study was conducted.  Because different modes and geographic regions have varying levels of transit service and potentially varying preexisting perceptions of the transit system, the specific mode and location of the study is noted if an actual transit system was evaluated.

The fourth and fifth dimensions summarized for each study are the methodologyused to evaluate the passenger impacts and the key findings. Study methodologies are classified into three categories (survey, simulation, or econometric analysis) and further analysis details are given. The vast majority of prior studies utilized survey-based methods, and for these, the sample size of the survey and statistical method utilized were noted.  Regarding the findings, numerical values or ranges to quantify changes in passenger behavior or perception are presented whenever possible.

Finally, it is important to note that only studies pertaining to passenger impacts are included in this literature review. There are numerous other areas of related research pertaining to RTI, such as development and evaluation of arrival prediction algorithms (e.g., Lin and Zeng, 1999; Chien, Ding & Wei, 2002; Cats & Loutos, 2016a; Cats & Loutos, 2016b), industry strategies for disseminating RTI (e.g., Schweiger, 2011; Barbeau, Borning and Watkins, 2014), and impacts on bus drivers (e.g., Watkins et al., 2013; Ji et al., 2014).

4. Findings from the Literature Review

Studies presented below are categorized based on their key impacts on passenger behavior: decreases in wait times, decreases in overall travel time because of changes in path choice, increases in use of the transit system (or ridership), increases in satisfaction with transit service, increases in perceived levels of personal safety, and other benefits.

4.1 Decreased Wait Times

The impacts of RTI on passenger wait times are the most common positive finding in the literature.  Accessing RTI at a passenger’s place of origin (e.g., home or work) enables the rider to “time” his or her arrival to a stop to reduce his/her actual wait time.  Additionally, RTI may reduce a passenger’s perceived wait time after reaching a stop because s/he is can check the status of the vehicle and feel more in control of his/her trip.  Specific studies that have examined the impact of RTI on wait times are summarized in Table 1.

[Table 1 near here]

4.1.1 Chronological Summary of Wait Time References

An early study examining the relationship of RTI and passenger wait times was conducted by surveying 295 employees at the University of Michigan in 1994 (Reed, 1995). A series of hypothetical scenarios were used in a conjoint analysis to assess the impact of RTI on the disutility of wait times for bus trips.  The results revealed that the disutility of wait times decreases with the provision of information.  RTI provided via telephone messages had a greater influence on the disutility of wait times than RTI provided on clocks located at bus stops, and both forms of RTI were preferable to printed schedules.

Another study examining the wait time impacts of RTI was conducted in the Hague, Holland (Dziekan and Vermeulen, 2006).  Surveys of passengers on a tramline were conducted in December 2003, one month prior to the installation of at-stop RTI signage.  Three months after RTI signs were installed, a follow-up survey was conducted of the same sample of passengers, and 53 respondents completed both surveys.  Passengers were asked to self-report how long they wait on average for a tram in order to measure perceived average wait time.  On the before survey, the average self-reported wait time was 6.3 minutes, and after, the average wait time was 5.0 minutes, a decrease of 1.3 minutes (20%).  Another survey conducted 16 months after the installation found similar results.

In 2009, a web-based survey of bus riders using RTI provided via personal devices was conducted in Seattle, Washington (Ferris et al., 2010).  A total of 488 RTI users who completed the survey were asked if there was a change in the amount of time they spent waiting for the bus as a result of using RTI.  91% of respondents reported spending less time waiting for the bus, 8% reported no change, and less than 1% reported an increase in wait times.  However, as the authors identified, the survey results were all self-reported and did not include a control group of non-RTI users.

A study conducted of the Clemson area bus system created a simulation model to estimate the passenger wait time impacts prior to deployment of RTI (Fries, Dunning & Chowdhury, 2011).  Pretrip time savings pertaining to timing arrivals at the bus stop and anxiety levels while waiting were evaluated. The results of the simulation model suggest that pretrip travel time savings are likely to be small, and reduction in anxiety levels while waiting are the most significant benefit of RTI.

A seminal study of the relationship between RTI and passenger wait times was conducted in Seattle, Washington (Watkins et al., 2011).  Both non-users and users of RTI were surveyed at bus stops (n = 655) and asked to self-report how long they had been waiting to capture perceived wait times. Self-reported average perceived wait times were 7.5 minutes for RTI users versus 9.9 minutes for non-users, a difference of approximately 30%.  Additionally, observers timing how long those passengers were waiting at the bus stop found that the actual wait times of RTI users were almost two minutes less than non-users.

A behavioral experiment conducted in 2013 evaluated the wait time impacts of RTI on bus riders in Tampa, Florida (Brakewood et al., 2014).  Study participants were randomly divided into a RTI user group that had access to RTI (n=110) and a control group without RTI (n=107).  Both groups were asked to complete a survey before the study began and a second survey after three months.  On both surveys, participants were asked to self-report how long they typically wait for the bus on the route that they ride most frequently.  The results revealed a significantly larger decrease in self-reported usual wait times for the RTI user group (-1.79 minutes) compared to the control group (-0.21 minutes) from the before survey to the after survey, a decrease in ‘‘usual’’ wait times approximately 1.5 minutes.  This represents a 16% decrease from their average wait time (11.36 minutes) from the before survey.

Another study considering the wait time impacts of RTI utilized a dynamic transit model known as BusMezzo (Cats & Gkioulou, 2014).  The model was applied to the “backbone” transit services in Stockholm, Sweden, which include the metro, trunk bus lines, and light rail.  The model evaluated passenger wait time expectations based on prior knowledge of the transit system, accumulated experience riding transit, and RTI.  The findings suggest that passengers will adapt their behavior to shorten wait times.

A study considering the wait time impacts of in-station RTI signage included a before-after survey of heavy rail riders in Boston, Massachusetts in 2012 (Chow et al., 2014).  While passengers waited for their trains in stations with and without RTI signage, 4,118 completed surveys were collected.  Passengers were asked to estimate how long they expected to wait for a train, and the results suggest that, after RTI signage was installed, passengers reduced their wait time estimates by 0.85 minutes on average.  After further controlling for service disruptions, wait time estimates were reduced by 1.3 minutes on average, or 17% of total wait times.

One final study was an onboard survey of passengers in 2012 on two commuter rail lines in Boston, Massachusetts (Brakewood, Rojas, et al., 2015).  Both users and non-users of commuter rail RTI were asked to self-report two different measures pertaining to wait times.  The first was wait time on the day of the survey, which did not have a statistically significant difference between RTI users and non-users.  The second was “usual” wait time, which was almost one minute less for RTI users than for non-RTI users (mean of 7.87 minutes for RTI users versus 8.45 minutes for non-users; n=868 total).  The authors suggest that passengers who consult RTI are able to adjust their wait times on days when the commuter rail experienced delays, and therefore, their “usual” wait times were less than riders who consult only schedules.

4.1.2 Synthesis of Wait Time Studies and Areas for Future Research

In summary, wait times have been studied extensively on both RTI signage and personal devices on multiple transit modes (bus, tram, light rail, heavy rail and commuter rail).  Methods for assessment include surveys of riders, in some cases paired with observations of actual wait times for comparison to stated wait times, as well as simulation models.  There is strong evidence suggesting that RTI provided via signage decreases perceived wait times and that RTI provided via personal devices decreases both perceived and actual wait times.  Subsequently, this may be the most valuable benefit of providing RTI to passengers.  In every study conducted, there was a statistical difference in perceived wait times, with a typical reduction of 20% to 30% or approximately 2 minutes.

One caveat for this type of research is the time and expense associated with surveys to ask respondents about perceived wait times. Future research should make use of location-aware technologies in smartphones or other new technologies such as Bluetooth to measure wait times.  Furthermore, although the average impact of RTI on wait times is well understood, the variance is not.  Future research should differentiate wait time perceptions based on differing headways and differing levels of reliability. Finally, the most substantial gap in knowledge related to these reductions in perceived and actual wait times is the overall impact on riders’ decisions. Travel demand models consistently penalize wait times as approximately 1.5 to 2 times the value of in-vehicle travel times.  However, this research points to wait times with RTI being weighted approximately equally to in-vehicle travel time. Future research is needed to improve the prediction of demand models based on a more accurate understanding of wait times.

4.2 Decreases in Total Travel Times

This section summarizes the impacts of RTI on passenger path choice, because transit riders are likely to choose the route that minimizes their travel time, which could vary based on RTI.  RTI is particularly important when transit service does not follow the posted schedule, as passengers can check RTI and adapt their behavior to choose an alternative mode or route of transit (Carrel et al., 2013).  Studies that have examined the impact of RTI on path choice and corresponding travel times are summarized in Table 2.

[Table 2 near here]

4.2.1 Chronological Summary of Travel Time References

An early study of RTI on passenger path choice and travel times created a simulation model of a single corridor with multiple transit options in Boston, Massachusetts (Hickman and Wilson, 1995).  Passengers were assumed to have access to RTI showing bus departure and running times, and a dynamic path choice model assuming passengers use RTI to improve their travel times was created.  After modeling numerous scenarios, the results suggest that potential travel time savings under perfect information scenarios were only about 3% of the total trip times (total trip times were 34-35 minutes).  Travel time savings under more realistic information scenarios were limited to 1% to 3% of the total trip times, a savings of 0.5 to 1.0 minutes.

Another study evaluating the path choice of passengers utilized a dynamic transit model known as BusMezzo with three components: traffic dynamics, transit operations and passenger demand (Cats et al., 2011).  The model was applied to part of Stockholm’s metro, and various scenarios were evaluated in which differing levels of RTI were provided (from platform-level to system-wide) and varying levels of transit operations (including normal operations and delays).  The findings suggest that providing a comprehensive, system-wide RTI has the potential to lead to shifts in path choice and travel time savings.

Another study used Monte Carlo simulation in two fictitious transit networks (Fonzone & Schmöcker, 2014).  Passengers were assumed to have two different strategies pertaining to travel times: travelers want to arrive at their destination as soon as possible or passenger prefer to stay slightly longer at their current location in order to reduce their overall travel time.  After considering access to RTI, the simulation model results suggest that RTI can reduce travel times by about 20%.

One last study considering the path choice implications of RTI used a discrete event simulation model, and this was applied to the transit network in the small city of Rivera, Uruguay (Estrada, et al., 2015).  Six variations of passenger behavior were evaluated that considered varying levels of information provision, and the results suggest that, compared to no available timetables, having static information can reduce total travel times by approximately 29% and RTI can reduce total travel times by 45%.

2.      Synthesis of Travel Time Studies and Areas for Future Research

Total travel time has relatively few studies compared to the number of wait time studies and all four used simulation modeling techniques.  Results were varied, with studies showing travel time savings of 3% to 45%.  For this reason, additional research is needed to better understand how RTI impacts path choice and travel time.  Future research should consider use of passenger observations, surveys, and/or other methods to measure actual behavioral changes in addition to the simulation models that have been used to date.  Future research could also consider travel time savings when there are varying levels of transit service coverage, e.g., dense networks versus sparser transit networks with fewer routing options.  Similarly, RTI impacts on travel time should be tested with varying levels of transit service frequency and reliability.

4.3 Increased Transit Use

If passengers spend less time waiting and/or decrease their overall travel time by choosing a shorter path, then the provision of RTI may also cause an increase in the frequency of transit trips by existing riders or potentially attract new riders to transit.  Specific studies that have examined the impact of RTI on transit use are summarized in Table 3.

[Table 3 near here]

 

4.3.1 Chronological Summary of Transit Use References

A study conducted from 2006 to 2007 on the University of Maryland shuttle bus network measured changes before and after the implementation of a RTI system (Zhang et al., 2008).  The authors created two fixed effects models of individual travelers’ monthly shuttle trips using survey responses (n=623).  Dependent variables were the natural log of monthly shuttle trips and natural log of monthly campus-based shuttle trips, and independent variables included numerous demographic and car-related variables (e.g., driver’s license and campus parking permit).  Use of RTI was not statistically significant in either model.  One possible explanation the authors identify is that the number of shuttle trips was measured only two weeks after an extensive marketing campaign of the new RTI system, which could have been insufficient time for travel behavior adjustments.

A second study of RTI impacts on transit use was conducted in Thessaloniki, Greece (Politis et al., 2010).  A survey of passengers (n=300) at bus stops with RTI signage was conducted, and respondents were asked if RTI inspired them to make additional bus trips.  A total of 59 respondents (19.7% of the overall sample) stated that they made 103 new trips as a result of the recently installed RTI system.

The previously mentioned survey of RTI users conducted in Seattle, Washington in 2009 also examined self-reported changes in bus trips.  On the survey, users were asked if their average number of transit trips per week changed as a result of RTI.  Over 30% of RTI users reported increases in non-commute trips, while a smaller percentage reported increases in commute trips on transit (Ferris et al., 2010).  A follow-up web-based survey of 5,074 RTI users in 2012 found similar results (Gooze et al., 2013).

A previously discussed study of the Clemson area bus system created a simulation model to evaluate passenger impacts prior to deployment of RTI (Fries, Dunning & Chowdhury, 2011).  Potential mode shifts from existing bus riders were evaluated, and the model results suggest that there would be limited if any changes in mode attributable to RTI.

Two studies conducted in Chicago, Illinois explored potential gains in transit ridership.  First, a stated preference survey was conducted in 2008 before RTI was widely available, and 76.1% of the 92 respondents stated they would increase transit use if RTI were available (Tang and Thakuriah, 2011).  The survey data was then used in a path analysis, and one of the key findings suggests that the psychological effects of providing RTI might lead to transit ridership increase. The second study evaluated the gradual launch of bus RTI by conducting an econometric analysis (Tang and Thakuriah, 2012).  The authors used a linear mixed model in which the dependent variable was average weekday route-level bus ridership per month from 2002 until 2010.  After controlling for unemployment levels, weather, gas prices, population, fares, and service frequency, the authors found an increase of 126 average weekday bus trips per route associated with RTI, or 1.8% to 2.2%.

The previously mentioned behavioral experiment conducted in 2013 in Tampa, Florida also assessed changes in transit use.  On both the before and after surveys, participants were asked to self-report the number of bus trips that they made in the last week.  The results revealed that the change in trips from the before to the after survey was not significantly different between the RTI and non-RTI groups.  However, the authors noted that many bus riders in the study were dependent on transit and had limited ability to increase their trips, as they were already using transit for all or a majority of their trips.  Also, the study participants were recruited from among people already in the sphere of influence of the transit provider; thus, there was no opportunity to analyze the potential of RTI for attracting new riders (Brakewood et al., 2014).

The previously mentioned survey of riders in Boston, Massachusetts also examined transit ridership before and after RTI signage in heavy rail stations in 2012 (Chow et al., 2014) using automated fare collection data in a fixed effects regression model.  The dependent variable was the log of boardings and independent variables included fare changes, station effects, line effects and seasonal effects.  Results suggest that ridership increased by 1.7% as a result of RTI signage in rail stations; however, the authors caution that these results should be treated as “preliminary” because of data limitations.

The most recent study evaluating the impacts of RTI on transit use was an econometric analysis of bus ridership in New York City (Brakewood, Macfarlane and Watkins, 2015).  Panel regression models were used to evaluate a natural experiment in which RTI was gradually made available in different areas of the city between 2011 and 2014.  After controlling for factors such as levels of transit service, fares, weather, and local socioeconomic conditions, the preferred model specification revealed an increase of 118 bus trips per route per weekday associated with RTI, or 1.7%. The results of a second model suggest that ridership increases may only be occurring on larger routes; specifically, the largest quartile of routes (by level of service) had approximately 340 additional trips per route per weekday associated with RTI, or 2.3%.  The authors note that the results are very similar to Tang and Thakuriah (2012).

4.3.2 Synthesis of Transit Use Studies and Areas for Future Research

In summary, numerous studies suggest that RTI may cause passengers to increase transit use. About half of the studies used survey methods to assess changes in transit use by individual riders, with the caveat that many of these studies are self-reported estimates of trips.  The other half used econometric methods to isolate aggregate changes in ridership on systems while controlling for other factors impacting ridership.  It should be noted that the econometric studies were often conducted in large cities with high levels of transit service (e.g., New York and Chicago), and ridership was estimated to increase approximately 2%, a substantial increase in locations where transit ridership is already quite high before RTI was introduced.

Further research is needed on a longer time frame using aggregate methods to assess both additional trips by existing riders and new riders to the system in small to medium size cities to validate results obtained in larger cities.  In addition, variation in ridership impacts based on the level of transit service provided (e.g., frequencies, service area) should be considered in cities of varying sizes. Similar to wait time impacts, the variation in ridership increases with these service levels and varying levels of service reliability are critical to understand for travel demand models, as such models are used in the industry to predict future transit ridership.

 

4.4 Increased Satisfaction with Transit

If transit passengers spend less time waiting and/or adapt their travel choices to reduce their travel time, they may become more satisfied with overall transit service.  Specific studies that have examined the impact of RTI on satisfaction are displayed in Table 4.

[Table 4 near here]

4.4.1 Chronological Summary of Satisfaction References

In the previously mentioned study on the University of Maryland campus (Zhang et al., 2008), respondents were asked about their overall satisfaction with shuttle bus service. These questions were used to create a random effects ordered probit model (n=482).  The dependent variable was satisfaction level on a five point scale, and the independent variables included numerous demographic and car-related variables (e.g., driver’s license and campus parking permit).  The results show that RTI use was associated with a significant increase in overall satisfaction level; RTI use increased the probability of Rating 5 (highest satisfaction level) by 0.071.

The previously mentioned survey of RTI users conducted in Seattle, Washington in 2009 also examined changes in satisfaction with transit service.  Users were asked if their overall satisfaction with public transit had changed as a result of using RTI.  48% of respondents stated that they were much more satisfied, and 44% of respondents stated that they were somewhat more satisfied (Ferris et al., 2010). A follow-up web-based survey found similar results; 51% of RTI users stated that they were much more satisfied with transit and 38% said they were somewhat more satisfied with transit (Gooze et al., 2013).

The previously mentioned behavioral experiment conducted in 2013 in Tampa, Florida also assessed changes in satisfaction.  On both surveys, participants were asked about six indictors pertaining to specific aspects and overall bus service in Tampa, and each indicator was rated on a five-point scale from very dissatisfied to very satisfied. Only two of the variables (satisfaction with how long you have to wait for the bus and satisfaction with how often the bus arrives at the stop on time) increased significantly from the before to the after survey between the control group and the experimental group. However, ratings of overall bus service did not show a significant change between the control and experimental groups from the before to the after survey (Brakewood et al., 2014).

The previously mentioned survey of heavy rail riders in Boston, Massachusetts examined passenger satisfaction with system-wide transit service before and after in-station RTI signage was turned on during 2012.  The results show that after RTI signage was available, passengers had a higher overall rating of the transit agency (3.46 compared to 3.41); however, this difference was not statistically significant (Chow et al., 2014).

4.4.2 Synthesis of Satisfaction Studies and Areas for Future Research

In summary, transit riders who are provided RTI are generally more satisfied with the overall transit service than passengers without RTI.  Although most studies showed a substantial increase in the level of satisfaction, the methods used are entirely self-reported data and often used a sample of only those who use RTI.  Presumably, those who did not find it useful would discontinue RTI use and not be included in the survey.  Future studies of passenger satisfaction should be done on a traditional experimental design basis with a treatment and control group before and after implementation. This will allow comparison between satisfaction scores of those with RTI and without. In addition, research is needed to understand the ranking of RTI in terms of importance among other transit components such as increased service or stop amenities.  Many agencies ask such questions in their passenger surveys, but the reports are not widely circulated in the industry or compared for universal understanding of customer preferences.

4.5 Increased Perception of Safety

Because passengers spend less time waiting at stops and stations, RTI may increase passenger perceptions of personal security when riding transit, particularly at night.  A rider facing an unknown wait time may feel vulnerable without knowledge of when the vehicle will arrive.  A synthesis of perceived personal safety literature is displayed in Table 5.

[Table 5 near here]

 

4.5.1 Chronological Summary of Safety References

The previously mentioned study in the Hague, Holland examined perception of security by passengers waiting at tram stops (Dziekan and Vermeulen, 2006).  On both surveys, passengers were asked to rate their level of perceived security at the boarding stop on a scale from 1 (very bad) to 10 (very good).  The average security experience went from 7.9 to 7.6, but this difference was not significant.

A previously mentioned study conducted on the University of Maryland campus also examined perceptions of personal security.  Respondents were asked to rate their feelings of security at night and during the daytime on a five point scale, and these responses were used in two random effects ordered probit models.  The results show that RTI use had a positive effect on feelings of security at night (significant at the 0.1 level) but did not have a significant effect on feelings of security during the daytime (Zhang et al., 2008).

Two studies conducted in Seattle, Washington that were previously mentioned provide evidence that RTI increases passenger perceptions of safety when riding the bus.  In 2009, respondents were asked how their perception of personal safety had changed as result of using RTI.  Although 79% of survey respondents reported no change, 18% reported feeling somewhat safer and 3% reported feeling much safer (Ferris, Watkins, & Borning, 2010).  A follow-up web-based survey conducted in 2012 found that 32% of 5,074 RTI users reported feeling somewhat or much safer as a result of using RTI (Gooze, Watkins, & Borning, 2013). Interestingly, the Ferris, et al. (2010) study found a strong correlation between feeling safer and gender.

In the previously mentioned behavioral experiment conducted in 2013 in Tampa, Florida, participants were asked a series of questions about feelings while waiting for the bus, including how safe they feel while waiting for the bus at night and during the daytime.  Feelings of safety during the daytime significantly increased for the RTI user group compared to the control group from the before to the after survey; however, there was not a significant difference regarding safety at night (Brakewood et al., 2014).

4.5.2 Synthesis of Safety Studies and Areas for Future Research

In summary, although safety at transit stops is often considered an important issue, most studies did not find substantial perceived safety improvements.  About half of the studies used a self-reported survey ratings asking respondents if they felt safer as a result of having RTI.  In these cases, only a small percentage of riders reported feeling safer (about 20%).  In other studies with a before and after rating of feeling of safety, most did not find a statistical difference between RTI users and riders without RTI or between before and after situations once RTI was implemented.  Riders with RTI have the option to leave a stop that feels unsafe until they know a vehicle is coming; however these situations are likely rare on most systems and may therefore be difficult to identify in the study results.

The potential for an incident to occur involves both a risk and an exposure component.  Therefore, as mobile RTI becomes more prevalent, the reductions in actual waiting time at stops described previously will reduce the exposure to potential crimes at stops. However, having less presence on the streets may increase crime as the street becomes less active and stops become more isolated. Further study in actual crime rates would be a useful addition to the perception studies previously conducted.

4.6 Other Impacts

The literature includes other possible benefits of RTI, including walking speeds in the vicinity of transit stations, distance walked to access transit stops, transfers between different modes of transit, and perceptions of environmental and traffic impacts of transit service.  The results are summarized in Table 6 and discussed below.

[Table 6 near here]

4.6.1 Walking speed

RTI signage in or around transit stations may impact passenger’s walking speed, particularly when signage says that a transit vehicle will soon arrive (Dziekan and Kottenhoff, 2007).  One study of the subway in Stockholm, Sweden observed passengers entering subway stations and counted the number of passengers running and walking when RTI signage outside the subway station was on.  The RTI displays were then turned off for one day at the selected subway stations and similar passenger observations were conducted.  The results reveal that significantly more people run when the RTI signage was on rather than when the signage was off.  Moreover, when the RTI displayed a low number of minutes, higher numbers of passengers were observed running. The greatest percentage of running passengers (14.5%) was observed when the RTI displays showed 1 minute until the train arrival, as compared to 2.6% when the sign was off.

4.6.2 Distance walked

RTI provided on mobile devices may impact a passenger’s decision of where to board the transit vehicle, which would impact the passenger’s walking distance to access transit (Ferris et al., 2010).  On the previously mentioned 2009 survey of RTI users conducted in Seattle, Washington,78% of respondents reported they were more likely to walk to a different stop based on RTI.

4.6.3 Transfers

RTI may influence a passenger’s decision to transfer between different modes of public transit. A study of Chicago, Illinois used econometric techniques to investigate the impacts of bus RTI on train ridership (Tang, Ross & Han, 2012).  Monthly average weekday train ridership per station from 2005 to 2010 was used as the dependent variable in a mixed linear model.  After controlling for numerous other factors such as socioeconomic factors, transit service factors and weather conditions, the model results suggest that there was an increase of 9.8 train rides per day, or 0.3% of the average weekday train station ridership, for every additional connected bus route that is provided with bus RTI.  The authors argue that this increase in rail ridership may be due to increased intermodal transfer efficiency, which suggests a complementary effect of the provision of bus RTI on connected rail service.

4.6.4 Perception of Environmental and Traffic Impacts

In addition to the direct benefits that RTI systems provide to users, a new direction of research suggests that availability of RTI might also lead nonusers of the transit system to perceive the transit provider as a more progressive, efficient entity.  One recent study investigated this using a two-wave survey of transits users and nonusers conducted to evaluate the impact of a shuttle bus RTI at Ohio State University (McCord et al., 2015).  Surveys were conducted in 2008 (n=4,399) and in 2012 (n=4,741).  Binary logit models were created to assess respondents’ attitudes toward the shuttle bus system helping to (1) promote a “green” campus and (2) reduce the amount of car traffic on campus.  The results suggest that those respondents who noticed the RTI system had a higher probability of providing a positive response for both the environmental and traffic related dimensions, regardless of their usage of the shuttle bus.

4.6.2 Synthesis of Other Studies and Areas for Future Research

In summary, additional studies addressed walking speed and distance, transfers, and perceptions of environment and traffic impacts.  However, with only one study looking at each impact, few broad conclusions can be drawn.  The relationship between walking and transit in particular is not often studied and additional research could be conducted to better understand the personal trade-offs between a walk trip and a transit trip that RTI allows, such as deciding to walk to the next stop for exercise to make use of a delayed bus.  Furthermore, RTI may substantially change the burden of transferring from one mode to another or from one route to another, making grid-type transit networks rather than those designed around one-seat services more convenient.  Further research is needed in the area of transfer impacts as well.

 

4.7 Summary and Comparison of Impacts

Table 7 shows the five most common passengers identified in the literature; other impacts (such as walking distance) are not included in this table due to the limited literature.  In Table 7, a filled circle represents a positive finding in that study, a half-filled circle signifies a finding that is sometimes positive, and an empty circle implies that the study investigated the passenger impact but the results were null, negative, or not statistically significant.  A dash means that the study did not investigate that passenger impact.

As can be seen in Table 7, nine of the nineteen studies examined the wait time implications of RTI, and eight of these studies found positive results.  This implies that wait times are thoroughly studied and have the most supporting evidence of the various impacts.  Only four of the nineteen studies examined travel time implications from path choice, meaning this is a field ripe for further research.  Also from Table 7, ten of the nineteen studies evaluated the impacts of RTI on transit use, and seven of them found positive results.  Notably, many of the studies that found positive results were conducted in large cities with high levels of preexisting transit service.  Five of the nineteen studies examined satisfaction and of those, only three found fully positive results. Similarly, five of the nineteen studies examined perceived safety and of those, only two found fully positive results. Satisfaction, though not discussed as frequently in academic studies, is often addressed in agency reports; safety, however, is another area where future research could address impacts more thoroughly.

[Table 7 near here]

 

5. Conclusions and Future Research

This study compiled a comprehensive literature review of the passenger impacts of real-time information (RTI) provided via signage at transit stops or personal devices. Three key behavioral impacts of RTI were identified: reductions in passenger wait times, reductions in overall travel time due to changes in path choice, and increases in transit use.  Additionally, two important changes in passenger feelings were identified: increases in perceptions of safety and increases in satisfaction with overall transit service.

Of the passenger impacts that were identified, the evidence suggests that the primary benefit to passengers is reductions in wait times associated with using RTI.  Early studies of signage at transit stops found a reduction in perceived wait times, or how long passengers think they have been waiting, on the order of 1.5 minutes (Dziekan and Vermeulen, 2006).  A recent study of RTI provided via mobile devices found a reduction of observed, actual wait times of approximately 2 minutes (Watkins et al., 2011).  Similarly, another recent study of mobile RTI found a reduction of self-reported “usual” wait times of nearly 2 minutes (Brakewood et al., 2014).  Taken together, there is now strong evidence that RTI significantly decreases the perceived and/or actual wait times of passengers, which is particularly important given that passengers generally dislike the waiting segment of transit trips more than being in the transit vehicle.

Numerous areas for future research were also identified throughout the paper and a few key areas are highlighted here.  Although passenger wait times have been studied in numerous real-world environments, there has been limited research pertaining to the total travel time impacts of RTI in actual transit systems.  Simulation models suggest that RTI provides possible travel time savings because of changes in passenger path choices (e.g., Hickman & Wilson, 1995); however, identifying effective methods to capture this in real world environments is a key challenge for future research.  A second noteworthy area for future research pertains to the impacts of RTI on overall transit use; specifically, a handful of prior studies evaluating aggregate ridership impacts found increases in bus ridership of approximately 2% (Tang and Thakuriah, 2012; Brakewood, Macfarlane and Watkins, 2015).  However, these studies were conducted in large cities with vast transit coverage and predominantly high frequency service.  Future research is needed to better understand the impacts of RTI in situations with varying levels of service frequency and availability.

Given the many benefits of RTI discussed in this paper, another important area for future research is comparing these passenger benefits to the transit operator costs of providing RTI.  This has had some limited treatment in the literature, including a cost-benefit analysis of a proposed system in San Luis Obispo, California (Nuworsoo et al., 2009) and an economic evaluation that monetized the passenger benefits of providing RTI in Taipei, Taiwan (Chen, 2012).  However, there is room for additional research that compiles the capital and operating costs of RTI systems to provide a complete comparison of the costs versus benefits after implementation.

Finally, as transit information provision continues to evolve based on the availability of new technologies, many additional areas for related research are likely to emerge. For example, transit providers are beginning to provide customized information about other attributes of the transit system, such as real-time crowding information (Yu et al., 2015; Zhang, 2015), and the passengers impacts of these new information sources will be fruitful areas for future study.

6. Acknowledgements

This work was supported in part by the Region II University Transportation Research Center (UTRC) under grant #49198-20 27.  The authors would also like to acknowledge the contribution of former students Almeria Senecat, Vinh Pham-Gia and Aaron Gooze to early drafts of this paper.

References

1. Bai, Y. & Kattan, L. (2014). Modeling riders’ behavioral responses to real-time information at light rail transit stations. Transportation Research Record: Journal of the Transportation Research Board, 2412:82-92.
2. Barbeau, S., Borning, A., & Watkins, K. (2014). OneBusAway Multi-Region – Rapidly Expanding Mobile Transit Apps to New Cities. Journal of Public Transportation, 17(4):14-34.
3. Brakewood C., Barbeau S., & Watkins K. (2014). An experiment evaluating the impacts of real-time transit information on bus riders in Tampa, Florida. Transportation Research Part A: Policy and Practice, 69:409-422.
4. Brakewood C., Macfarlane G., & Watkins, K. (2015). The impact of real-time information on bus ridership in New York City. Transportation Research Part C: Emerging Technologies, 53:59-75.
5. Brakewood C., Rojas F., Zegras P.C., Watkins K., & Robin J. (2015). An analysis of commuter rail real-time information in Boston. Journal of Public Transportation, 18(1):1-20.
6. Carrel, A., Halvorsen, A., & Walker, J. L. (2013). Passengers’ Perception of and Behavioral Adaptation to Unreliability in Public Transportation. Transportation Research Record: Journal of the Transportation Research Board, 2351:153–162.
7. Cats, O. & Loutos, G. (2016a). Evaluating the added-value of online bus arrival prediction schemes. Transportation Research Part A: Policy and Practice, 86:35-55.
8. Cats O. & Loutos G. (2016b). Real-time bus arrival information system: an empirical evaluation. Journal of Intelligent Transportation Systems: Technology, Planning and Operations, 20(2).
9. Cats O., Koutsopoulos H., Burghout W., & Toledo T. (2011). Effect of real-time transit information on dynamic path choice of passengers. Transportation Research Record: Journal of the Transportation Research Board, 2217:46-54.
10. Cats O. & Gkioulou Z. (2014). Modeling the impacts of public transport reliability and travel information on passengers’ waiting-time uncertainty. EURO Journal on Transportation and Logistics, 1-24.
11. Caulfield B. & O’Mahony M. (2009). A stated preference analysis of real-time public transit stop information. Journal of Public Transportation, 12(3):1-20.
12. Chen D.J. (2012). Measuring the passenger’s benefit of providing the real-time information system of the bus transit. Proceedings of the Annual Meeting of the Transportation Research Board, Washington, DC.
13. Chien, S., Ding, Y., & Wei, C. (2002). Dynamic bus arrival time prediction with artificial neural networks. Journal of Transportation Engineering, 128(5):429-438.
14. Chow, W., Block-Schachter, D., & Hickey, S. (2014). Impacts of real-time passenger information signs in rail stations at the Massachusetts Bay Transportation Authority. Transportation Research Record: Journal of the Transportation Research Board, 2419:1-10.
15. Dziekan, K. & Kottenhoff, K. (2007). Dynamic at-stop real-time information displays for public transport: effect on customers. Transportation Research Part A: Policy and Practice, 41(6):489-501.
16. Dziekan, K. & Vermeulen, A. (2006). Psychological effects of and design preferences for real-time information displays. Journal of Public Transportation, 9(1):1-19.
17. Estrada, M., Giesen, R., Mauttone, A., Nacelle, E., & Segura, L. (2015). Experimental evaluation of real-time information services in transit systems from the perspective of users. Proceedings of the Conference on Advanced Systems in Public Transport (CAPST), 1-20.
18. Ferris, B., Watkins K., & Borning A. (2010). OneBusAway: Results from providing real-time arrival information for public transit. Proceedings of CHI, 1807-1816.
19. Fries, R., Dunning, A. & Chowdhury, M. (2011). University Traveler Value of Potential Real-Time Transit Information. Journal of Public Transportation, 14(2), 29-50.
20. Fonzone, A. & Schmöcker, J-D. (2014). Effects of transit real-time information usage strategies. Transportation Research Record: Journal of the Transportation Research Board, 2417:121-129.
21. Gooze, A., Watkins, K., & Borning, A. (2013). Benefits of real-time transit information and impacts of data accuracy on rider experience. Transportation Research Record: Journal of the Transportation Research Board, 2351:95-103.
22. Hickman, M. & Wilson, N.H.M. (1995). Passenger travel time and path choice implications of real-time transit information. Transportation Research Part C: Emerging Technologies, 3(4):211-226.
23. Ji, Y., He, L. & Zhang, M. (2014). Bus driver’s responses to real-time schedule adherence and the effects of transit reliability. Transportation Research Record: Journal of the Transportation Research Board, 2417:1–9.
24. Lin, W-H. & Zeng, J. (1999). Experimental study of real-time bus arrival time prediction with GPS data. Transportation Research Record: Journal of the Transportation Research Board, 1666:101-109.
25. McCord, M., Mishalani, R., & Ettefagh, M. (2015). Effect of real-time passenger information systems on perceptions of transit’s favorable environmental and traffic reduction roles. Transportation Research Record: Journal of the Transportation Research Board, 2538:102-109.
26. Mehndiratta, S.R, Cluett, C., Kemp, M., & Lappin, J. (2000). Transit watch – bus station video monitors: customer satisfaction evaluation. FHWA Report.
27. Nuworsoo, C., et al. (2009). A Benefit-cost evaluation of smart transit features at small scale transit operations. Proceedings of the Annual Meeting of the Transportation Research Board, Washington, DC.
28. Politis, I., Papaioannou, P., Basbas, S., & Dimitriadis, N. (2010). Evaluation of a bus passenger information system from the users’ point of view in the city of Thessaloniki, Greece. Research in Transportation Economics; 29(1):249-255.
29. Reed, T. (1995). Reduction in the burden of waiting for public transit due to real-time schedule information: A conjoint analysis study. IEEE; 83-89.
30. Schweiger, C. (2003). TCRP Synthesis 48: Real-time bus arrival information systems. Transit Cooperative Research Program of the Transportation Research Board, Washington, D.C.
31. Schweiger, C. (2011). TCRP Synthesis 91: Use and deployment of mobile device technology for real-time transit information. Transit Cooperative Research Program of the Transportation Research Board, Washington, D.C.
32. Tang, L. & Thakuriah, P. (2011). Will psychological effects of real-time transit information systems lead to ridership gain? Transportation Research Record: Journal of the Transportation Research Board, 2216:67-74.
33. Tang, L., Ross, H., & Han, X. (2012). Substitution or complementarity: an examination of the ridership effects of real-time bus information on transit rail in the city of Chicago. Proceedings of the Annual Meeting of the Transportation Research Board, Washington, DC.
34. Tang, L. & Thakuriah, P. (2012). Ridership effects of real-time bus information system: A case study in the city of Chicago. Transportation Research Part C: Emerging Technologies, 22:146-161.
35. Watkins, K., Ferris, B., Borning, A., Rutherford, G.S., & Layton, D. (2011). Where is my bus? Impact of mobile real-time information on the perceived and actual wait time of transit riders. Transportation Research Part A: Policy and Practice, 45(8):839-848.
36. Watkins, K. Borning, A., Rutherford, S., Ferris, B. & Gill, B (2013). Attitudes of bus operators towards real-time transit information tools. Transportation, 40(5):961-980.
37. Yu, B., Li, T., Kong, L., Wang, K. & Wu, Y. (2015). Passenger boarding choice prediction at a bus hub with real-time information. Transportmetrica B: Transport Dynamics, 3(3):192-221.
38. Zhang, F., Shen, Q., & Clifton. (2008). Examination of traveler responses to real-time information about bus arrival using panel data. Transportation Research Record: Journal of the Transportation Research Board, 2082:107-115.
39. Zhang, Y. (2015). Real Time Crowding Information (RTCI) Provision. Master of Science Thesis, Royal Institute of Technology, Stockholm, Sweden.