One of the biggest challenges in the transportation sector is addressing traffic congestion on expressways. In Japan, traffic congestion is often caused by traffic moving from surrounding tourism spots toward the Tokyo metropolitan area; the number of vehicles tend to increase on Sundays and the last day of consecutive holidays. This results in tourists losing time and also negatively impacts tourism as travelers return early to avoid traffic congestion.
A potential solution to reduce traffic congestion is to promote delayed departure for tourists. While the construction of new roads can potentially ease congestion, it is time-consuming and expensive. However, promoting delayed departures is a behavior change approach, which is inexpensive, effective, and feasible. Kawahara [1
] showed that delaying the departure time of vehicles could reduce traffic congestion. A survey of Japanese tourists showed that over 60% of respondents would change their departure time based on traffic congestion information [2
]. Additionally, delaying departure from tourist spots could potentially increase consumer activity, which may not have been initially planned, such as shopping or eating at restaurants. In this study, such unplanned activities are referred to as “additional stopovers”.
In recent years, the increased use of smartphones and mobile applications has made it easier to provide real-time information to consumers. Such applications have been used to promote socially desirable behavior in various fields [3
], including the transportation tourism sectors (e.g., [5
]). If tourists are provided with traffic congestion information tailored to their homeward journey via mobile applications, they may delay their departure to avoid traffic.
Prior studies on this subject examined the Global Positioning System (GPS) trajectory data (mobile probe data) of tourists in the Yamanashi prefecture of Japan and demonstrated that self-driving tourists tend to make additional stopovers to avoid congestion [7
]. These studies provided indirect evidence showing that the real-time traffic congestion information can increase stopovers. However, these studies also reported results of situations where no estimated near-future traffic information was provided. Moreover, the reasons for additional stopovers could not be explained using the existing datasets. Thus, an experimental approach is necessary to verify the effect of providing near-future traffic estimates on reducing congestion and promoting tourism.
This study aimed to provide empirical evidence to demonstrate the effectiveness of providing real-time traffic congestion estimates to increase tourism-related additional stopovers. For this, the authors tested and confirmed the following hypothesis through a field experiment.
Self-driving tourists will make additional stopovers to delay their departure and avoid traffic if near-future traffic congestion estimation is provided.
A mobile application was developed to provide information on real-time traffic congestion estimation and nearby tourism spots, for self-driving tourists in Yatsugatake, Japan, to understand its impact on their stopover behavior. Existing traffic congestion data indicate that congestion is at a peak between 16:00 and 17:00 approximately, and it gradually reduces after that. Therefore, delaying departure times during peak hours would potentially reduce the travel time for returning home. The mobile application offers options such as “depart now” and “depart later”, along with the estimated travel time required to reach the intended destination. The user’s operation logs their travel trajectory, and post-experiment questionnaire results were obtained via the application.
The aim of this study was to provide evidence to demonstrate the effectiveness of information intervention to travel and tourism behaviors, while also providing empirical evidence that actual stopover behavior can be inferred from smartphone application operation logs. This finding eliminates the need for GPS trajectory data to determine the effectiveness of such interventions and eases the process of conducting such experiments from the viewpoint of personal information protection.
The remainder of this paper is organized as follows. Section 2
delves into related previous studies, including tourist behavior studies using GPS trajectory data and intervention studies using mobile applications. In addition, this section describes the advantages of the present study. Section 3
describes the design and implementation of the smartphone application to provide real-time traffic estimation. It also presents methods for field experimentation to demonstrate that actual behavior can be inferred using an application operation log. Section 4
describes the experimental results and discussion. Section 5
presents our conclusions, limitations of the study, and future work.
2. Related Work
As GPS data collection is more accurate and reliable than traditional self-reporting, the use of GPS is common in the field of transportation and tourism [8
]. Various studies have analyzed tourist behavior using GPS trajectory data in the transportation and tourism fields [8
]. Hallo et al. investigated traveler behavior using GPS in case studies of national park visitors [10
]. Connell and Page examined car-based tourism in a national park in Scotland and mapped the itineraries of the tourists [11
]. They indicated that itinerary mapping could help policy makers understand spatial patterns for tourism planning. Newton et al. investigated the spatial–temporal patterns of vehicular stopping behavior along park roads, indicating that such information is valuable to park managers to better understand and manage visitor flow [12
]. Le et al. highlighted the tendency of additional stopover behavior during different departure periods under various congestion conditions, stating the possibility of promoting additional stopovers by providing near-future traffic congestion information [7
]. The findings of these studies can help policy makers or managers promote tourism. Nevertheless, these implications were discovered in observations without actual interventions; therefore, it is difficult to conclude that these interventions are effective. The present study tested the effect of interventions via a field experiment.
Mobile applications are used for interventions in broad research areas such as promoting healthy [3
] and pro-environmental behaviors [13
] and traffic and tourism research [5
]. However, a mobile application for stopover promotion has not been designed thus far. Gabrielli et al. described the user-centered design of a mobile application to promote sustainable behaviors in urban mobility, thereby demonstrating that not all users are motivated by environmental concerns and that users desire concrete rewards that are closely related to the target behavior [5
]. Siuhi and Mwakalonge listed several mobile applications that could contribute to resolving traffic problems [6
]. Certain mobile applications provide real-time traffic information and can reduce traffic congestion. For example, Sigalert.com [15
] provides real-time updates on traffic and road speeds for the U.S. 511 Georgia and Atlanta Traffic [16
], which is the official traffic application of the Georgia Department of Transportation (DOT). Colorado Roads [17
] provides real-time information related to highways in the U.S. state of Colorado such as speeds, travel times, road conditions, incidents, and road closures. However, although such applications are commercialized and found in application stores, none of them target the behavior change of promoting stopovers.
In this study, a mobile application to simultaneously provide real-time traffic congestion estimation along with nearby tourism spots was developed. The experiment results showed that if delaying the departure significantly shortened the required travel time for returning home, most users tended to delay their departure. If the estimated travel time could be shortened by 20 min or more, the selection ratio of delaying the departure time was higher than that of departing immediately, and if it was shortened by 30 min or more, 100% of the participants delayed their departure. Thus, it is expected that approximately 40% of people will perform an additional stopover if the returning route is congested and a near-future traffic congestion estimate is provided.
Therefore, the hypothesis that tourists would make additional stopovers to delay their departure is partially supported by the experimental results.
The experimental results show that providing information on estimated near-future traffic congestion and nearby tourism spots promoted stopover behavior. A major contribution of this study is that it empirically confirmed that information interventions are effective in promoting unplanned stopovers when avoiding traffic congestion, especially in the case of tourism-related travel. Moreover, the experimental results revealed that chosen options detected by operation logs showed a high consistency with actual stopover behaviors confirmed by questionnaires and travel trajectory, suggesting the possibility of inferring actual stopover behaviors from smartphone application operation logs. This finding eliminates the need for GPS trajectory data to determine the effectiveness of interventions and eases the process of conducting extensive large-scale experiments in future studies from the viewpoint of personal information protection.
It should be noted that the participants of this experiment had an intention to avoid congestion. It is necessary to reconsider the methodology of the recruitment of participants for future experiments. In addition, the number of active participants was 60, and the experimental period was one month. The number of active participants may seem to be small. The recruitment of participants for this experiment was difficult because it requires personal information such as GPS trajectory data. People hesitated to agree to participate, even if it was clearly stated that the data will be processed such that no individual can be identified. It is expected that the effectiveness of the developed application would enable the increase in the number of users of the application for a more extended experimental period. Therefore, a large-scale and long-term control experiment with more participants is planned in the future.