Information flows in Demand Responsive Public Transport: Interactivity, information, and flexibility in a modern ridesharing service

(1)

Information flows in Demand Responsive Public Transport

Interactivity, information, and flexibility in a modern ridesharing service

Oscar Askfelt Karl Hamnebo

Computer and information science Bachelor thesis

15 hp

Spring term 2021

Supervisor: Jan A. Persson Examinator: Johan Holmgren

(2)

Abstract

The focus of this thesis is to study what and how information flows can be used to improve Demand responsive transport (DRT) systems by understanding potential users and how they could be willing to participate in DRT to a higher degree. The viewpoint of this thesis tends to lean towards a DRT service of a public transport type. This thesis studies users in relation to what interaction and information they perceive to be needed in dealing with a DRT service and the different pros and cons with various approaches. The study gathers information by performing adapted qualitative interviews with a select number of users between the ages of 20-35. The participants give their views on three DRT scenarios and reflect on DRT in general as a concept presented to them through a tangible mocked-up interactive prototype.

The thesis makes several distinct findings. The importance of pricing a DRT service correctly is vital to the users, as several participants in the study relied on pricing for decision-making.

It also finds that the usage of zones as nomenclature is confusing to many users. The services must be dependable and punctual to both attract users, keep users, and build trust among the general populace.

This study shows that DRT services could be a difficult concept to introduce to users. DRT could be introduced as a complement or as an alternative to conventional public transport.

An important factor is a well-designed flow of information in the application to keep the user engaged and involved. It is shown that the usability of the application is a cornerstone for a theoretical DRT service to excel. Context is important where DRT and ridesharing would have a higher success rate. Nighttime in urban areas could be a niche market, due to the irregularity, delay, or interruption of regular public transport services at these hours.

(3)

Popular Science Summary

In this thesis, a study is conducted regarding the informational flows and the interactivity through studying hypothetical DRT services. For this purpose, three different scenarios of DRT services were mocked up in a prototype. Each one had a different mix of approaches to the aspects of time and space in a DRT context. What they each have in common is that they all share the potential of being a shared ride. Ridesharing and DRT, in general, have shown benefits of affecting traffic and transport with less traffic congestion and emissions. To reap the rewards of ridesharing requires increasing the frequency of shared rides, and therefore it is important to understand information and interactions in relation to the users.

To be able to show the subjects what DRT, a rather theoretical and hard-to-grasp concept is, a mock-up was created. The mock-up allowed the users to get a more hands-on experience of the concept as a whole. The mock-up was used during the interviews, and the user's interaction with it served as a basic talking point for what the user would want both information and interaction-wise.

The users were queried on what information and what interaction they would like in a DRT service, and this was analyzed to reflect both the user’s point of view and the strengths of the actual service. The subjects in this survey were limited to 20-35-year-olds. The study points to what direction a DRT service should keep in mind to attract users of this age, how to keep them interested and increase their flexibility. In this study, it is shown that interaction is in high demand when it comes to these sorts of services. Information and interaction often intertwine with each other, and many participants in the study mentioned both of them as key elements.

Summaries of the interviews, keywords picked and diagrams are used to represent the collected data. These are analyzed, and emerging themes are pinpointed. The themes revolve mainly around interactivity, usability, information, and dependability. The themes may be used in further research to establish a foundation for more in-depth knowledge of how a DRT service should operate.

(4)

Dictionary

DRT - Demand Responsive Transport. A flexible ridesharing mode of transportation that is based on the demand of the users.

CPT - Conventional Public Transport.

VS - Virtual Stop. A digital place where the DRT service can pick the user up.

(7)

1. Introduction

1.1 Background

Demand responsive transport (DRT) is a form of transport where the users’ needs are taken into greater consideration for the time and location of the transport. It is often a service where the rides are shared. The extent of freedom for the user in terms of location and time varies between different DRT options. Historically, DRT has had many different

interpretations and implementations over the decades, and many with some basis in providing transport for people with special needs [1],[2]. DRT and ridesharing are highly affected by the modernization of our society and new opportunities arise from the fact that almost everyone walks around with a smartphone allowing for higher optimization of the service in real-time. This new progress with technology has reignited the spark for DRT [1], [3].

A major facette of DRT services is the impact on the environment. With greater

consciousness of the climate, DRT has the potential of alleviating the stress put on nature by reducing the amount of traffic and decreasing the emission of carbon dioxide and other pollutants. Environmental emissions are however a complex problem to solve where ridesharing services have the potential of having a positive effect. An increase in people traveling together and sharing rides leads to a decrease in the environmental impact of traffic. DRT and ridesharing come in many different forms and could be an important tool in affecting issues like travel cost, travel time, traffic congestion, fuel consumption, and air pollution [4]. Using ridesharing with transfers between transport has been shown to reduce vehicles’ total distance traveled by 30% [5], which is good when considering that ridesharing over the last 30 years has seen a decrease by 10% in work-related transport [4].

Many of the modern DRT services have often been initialized by commercial companies, such as Uberpool and Lyft, but they are more motivated by economic gains rather than societal concerns. For publicly subsidized DRT the cost of operation has often been higher compared to traditional public transport in urban areas [1]. The increased price may be acceptable since DRT also leads to further mobility and accessibility. With the reignition of DRT, much of the focus has shifted to more urban settings and not solely on the utility aspect in rural areas [1].

1.2 DRT in this study

In our project, DRT is used as an alternative or complement to public transport. As in many other cases of DRT systems, the type of vehicle in mind is a smaller bus or minibus with about 4-6 seats to make transport more mobile and adaptable than a large conventional bus [2], [6].

While the level of comfort is a selling point for some DRT systems, the scope of this study is to study information flows in DRT in relation to public transport or other modes of

transportation and will not try to approach travelers needs based on comfort rather than

(8)

accessibility, availability, and adjustability. Feedback regarding comfortability and cost was however interesting and often became relevant in tradeoff scenarios.

Though the scope of what constitutes DRT is still up for debate and a highly subjective topic, the fact remains that it still is a more modern means of transportation as opposed to

conventional public transport (CPT). As DRT has evolved into a more novel phenomenon with more modern technology it is crucial to try and understand the different aspects of the topic, hence there is also a need to take a closer look from the user’s perspective and how users' interactions could be understood and interpreted. Studying users and their needs makes it possible to understand the information flows and see what makes it attractive, available, and accessible, while also leading to a broader userbase. As with many other systems, the systems themselves can be well-engineered and yet unused if no one is unable to understand them. If the outcome is a desire for more flexibility for the users in regard to choosing their locations, a better time management system, or whatever form it may take, then there is a better understanding of the issue.

1.3 Potential future

The potential benefits of enabling ridesharing more efficiently could be too great to go unaddressed and finding how to increase the frequency of shared rides are of interest to the whole society. A major issue with DRT in the past has been how to effectivize the ridesharing part of the service and minimize empty or half-empty busses driving around for the majority of the time. Many dynamic ridesharing services have the same problem of needing

instantaneous coordination between different parties, schedules, and costs or pricing [4]. As people are getting more accustomed to using smartphones as part of traveling, the hopes are to create a platform for interaction and information that allows for greater flexibility for DRT services and potentially increases efficiency by a higher rate of ride-sharing. Looking at what aspects of time and space are important to the users allows for the operators to take these things into account as well. In some simulations, the travel time is even decreased with DRT without extra cost and there is the possibility to replace CPT with more efficient forms of DRT altogether [2].

When analyzing the phenomenon of travel we decided to primarily focus on the aspects of place and time. The reasoning was that the basis of any travel is to span a distance from a node of origin to a node of destination, and that spanning of distance consumes time. An argument could be made for including price as one of the prime aspects and while it is an important information block in decision making, it may not necessarily be the deciding factor while considering the interaction.

1.4 Purpose

The main purpose is to contribute to a foundation for further work with DRT and ridesharing services by looking at factors that can increase ridesharing frequency and thereby having a positive effect on the environment. Understanding the user’s flexibility could be a basis for enhanced availability, usability, and efficiency. Currently, there seems to be a gap within DRT-research. This gap consists of the information, information flow, and user-centric

(9)

perspective. We were not able to find any previous research done on this matter that could relate to our purpose.

Our purpose is to find out what information and interactions could make DRT attractive to a certain audience. Studying the information flows and making DRT and ridesharing more attractive would increase the use and enhance the benefits of such services. Better information flows could also potentially fill vehicles in transit reducing empty driving for operators increasing ridesharing. Information flows will contribute to better usability and interaction by determining what information users need at any certain point when interacting with the service. A goal is also to investigate what interactivity the users would want and need, and connect this interactivity to the general flow of information and make it clear and usable. The flexibility of our studies’ subjects using DRT is also being investigated.

DRT has conventionally been used for transporting people with special needs and has been tested and studied to a certain degree in that capacity. We see a need to widen the audience if the intention is to make use of DRT as a complement to CPT in a more societal and

broader context. Adults in the range of 20-35 years old are a demographic that frequently uses CPT, are early adopters, and are flexible. The situation with the pandemic made us have to choose participants from our immediate surroundings, which were mostly students in this age span. Though rather arbitrarily selected, the sample still constitutes a cross-section of an important demographic.

1.5 Research questions

1. What interactive information flows can be used to improve DRT systems?

2. What information and interaction do users perceive as needed when exploring DRT?

2.1. How can a DRT service increase user flexibility?

3. What pros and cons do users perceive in relation to the interactivity of DRT services?

The main focus is how to improve DRT systems and what information is needed to do so.

Our approach is through the user’s reasoning around what and how information and interactions could be used in this context. What kind of information and interactions are perceived as needed pertains to the user's wishes to be accommodated from the operator's perspective to enable a functioning and more efficient system for both parties. Looking at how flexible users are can be used to create a more attractive service with increased ridesharing and improving DRT.

To explore what information flows are needed to satisfy the users of a DRT system, users will be questioned on flexibility from their perspective and how the service can accommodate them in order to increase the ridesharing frequency. The goal is to then put that data into the perspective of how a service could benefit from these information flows and operate to achieve a higher grade of success.

(10)

1.6 Related work

1.6.1 Evaluating DRT experiences

Mapping the current experiences collected from DRT and the viability of the system as such has been a topic for many different studies. Various trials and simulations, both past and present, have been used to create a better understanding of the concept. Many of the studies are often based on simulations of a service, and too few have a user experience perspective capturing mostly technical aspects.

In Pettersson’s review of DRT experiences, an insightful study is made about whether new technology can improve DRT [1]. By compiling data from different sources and conducting interviews to get a more in-depth understanding the study evaluates different kinds of DRT systems as well as defines the different types of DRT systems prevalent at this time in history [1].

Some constraints on the phenomenon to try and scope what DRT is found in Petterson’s evaluation, as well as previous research. Among the constraints for Petterson’s study is a period of being more recent than 2012, the booking of the trip has to be made digitally, automatic real-time matching of supply and demand, open for the public, as well as the trips, are meant to be shared collectively [1]. These constraints are by no means universal, but give a common ground when discussing DRT.

To get an oversight of some of the operational strategies being used right now, a

comprehensive compilation is found in Pettersson’s study, e.g. as a mode of transport from door to door, from zone to zone, or from a zone to a set point of destination or vice versa with a reversed flow [1].

In their research to unearth DRT systems viability, Ronald et al. set their focus upon the spatial aspect of such systems, while other works of research have had their gaze set upon the number of trip requests as a determining factor [7].

Looking at the traffic simulations that already exist, Ronald et al. found none that was able to represent the dynamic nature of DRT satisfyingly, leading them to develop their own model for simulation [7]. In that simulation environment they got the expected result that the waiting time increased when the zone of operation is larger, as well as with a higher number of passengers. Longer trips also meant a higher amount of sharing, while shorter trips meant vice versa [7].

With the variant of the relationship of many-to-one, i.e. passengers traveling from a zone to a single node of destination, Ronald et al. found that while effective on route to the destination there was a high frequency of empty busses on their way back. Though this also depends on the demand for the route in the opposite direction [7].

Navidi et al. shows that during various levels of demand in a simulation, DRT can compete with CPT in user performance as well as the cost for the operator [3]. In another study made

(11)

by Navidi et al. looking closer at the comparison of ad hoc DRT and CPT, their results indicate that mobility could also be increased by replacing CPT with DRT [2].

Looking at an earlier version of a DRT service Nelson and Phonphitakchai found that many negative experiences were related to failures of the service, e.g. delays [8]. Another pain point for users was that the service areas and pick-up nodes were too far away, though this was not always because of operators being unreasonable [8]. Creating a clearer information flow could perhaps have lessened the frustration for the user.

1.6.2 Niches and complementing CPT

Alonso-González et al.’s study about the potential of DRT aims to create a systematic tool for evaluation and to ensure how best to use DRT as a complement to traditional linear transport methods such as buses, trains, and subways [9]. Their study incorporates commercial DRT services such as Lyft, Uber, and Via. According to Jiang et al., such services have also replaced large parts of the commercial vehicle-for-hire market, i.e more conventional taxi services are turning into the more novel ones like Lyft and Uber etcetera [10]. The scope of Alonso-González et al.’s study includes services in a span of several different countries, including England, Italy, and Finland. In Finland, for example, the first fully automatic DRT service called Kutsuplus was in operation between the years of 2012-2015 running alongside CPT allowing for different routes and more flexibility [9], [1].

Doing a case study around 2012, Nelson and Phonphitakchai picked up on the attitude that many users of LinkUp as a DRT service expected it to work as a taxi business [8]. In

comparison, another user-perceived notion was that DRT was superior to CPT but that may be accredited to it being a complement, relying on CPT to cover all other use cases [8].

A niche market for DRT is the rural areas where CPT can not offer the same level of service.

A factor here is waiting time where Avermann and Schlüter found that it had a large impact on satisfaction [11]. In the same study, they could however not make the same connection whether the ride was shared or not affected satisfaction [11].

1.6.3 Potential users

As with all systems and maybe especially more novel ones, a vital part is to understand the users that could potentially utilize it. By using data there is a possibility to discover how to carve out a corner of the market for DRT, and how to find users for DRT services. New technology allows for new ways of communicating what is expected or needed from a service seen from a user-centric perspective.

Understanding the users are important since user experience and user acceptance are crucial components for novel concepts to succeed [6]. In studying the users of new potential modes of transport and if possible to discern different groups of users in a study by Winter et al. they conducted an online survey where users of different backgrounds (though all from dense urban cities in the Netherlands) could answer a stated choice questionnaire to gather data about the perception of new forms of shared transport [12].

(12)

The study indicated that the current mode of transport was the most predominant factor for the perception of different transports [12]. Car commuters already had a personal means of transportation, but some were motivated to share rides with the availability of parking or the cost of parking. Those who use both a car and public transport are highly excited by the possibilities of new modes that can combine the ones they already were using. Commuters by public transport already had means of transport and could possibly be discouraged by a higher cost. The young and pressed for time expressed enthusiasm since it offered new possibilities. Finally, the bicyclists and pedestrians summarily saw a value to new modes, but also possessed means of transport already, ones that were not motorized [12].

Winter et al. also noted that the factor of age in willingness to share rides was inconsistent and showed in their studies that people across all ages were willing to share as well being more closed off to the idea [12].

According to Alonso-Gonzalez et al., a fair percentage of those using DRT options presented and studied in their research would have been able to walk or bicycle the same distance with relative ease. This points to many using DRT services out of comfortability, and not so much out of urgent need [9].

Davison et al. also shows that DRT has the potential to be successful not only in the more traditional rural DRT areas but also in a more urban setting [13]. It also shows the fact that different people have different views on what DRT is, and that it can take many shapes and forms [13]. The report from Davison et al. also portrays how the users are not very content with how DRT is done, and that the majority of people in the surveys and interviews want to change things [13]. This means that it becomes quite evident that large-scale changes in social consciousness as well as in the performance of services are needed for DRT to truly work. The study from 2012 also makes out directions for the future by pointing to the

real-time information exchange online, but their advice is directed at the operators [13]. This could be a product of the technology not being suitable enough for much more at the time.

Being of note is also the importance to address some security and safety concerns passengers can have as well. Ridesharing in DRT comes with an exposition to strangers regarding personal information connected to traveling [4].

The willingness to pay in relation to eco-friendliness has been studied and in some cases where an environmentally friendly design can increase an acceptance of a higher price [14].

Furthermore, surveying also led to the conclusion that users being women and people with less than 800€ disposable income were more accepting of a higher price as well [14].

1.7 Method

To increase the level of palpability and tangibility for interviewees and potential users to better understand the concept of DRT and ridesharing we created an interactive mock-up prototype. To explore information flows through the prototype, which served to underline and showcase DRT, we could extract various aspects from different scenarios and study the interaction between a service and a user. To approach the topic from different angles the prototype had three differentiating scenarios. Analyzing the user's input discerns their

(13)

perception, reasoning, and what factors potentially could be used to improve DRT for both operators and users. For the operators, the result could be used to make the service more flexible and fill more seats, while for the users they could be accommodated and make their experience more satisfactory.

The interviews created were adapted with the current pandemic situation in mind and taking into account the remote nature of communication. Limiting the study to potential users gave more space for creativity in examining the topic independently from operators.

For creating the prototype we relied on Design Thinking as a method to iteratively create and test the prototype, both as a way to relay the concept to a broader audience but also as a communication device for internal stakeholders.

To analyze the results in a comprehensible and fairly easy-to-understand manner we identified keywords in our summations of the user interviews, and then found emerging themes that could indicate certain tendencies within our target audience.

We included price in the study relativized to other transports since it is an important factor in making decisions. The inclusion of price was a necessity, but we tried not to delve too deep into the specifics. From our perspective, it could be wildly misleading and manipulative to try and give certain price points that only operators could calculate to a satisfying accuracy and may be included in further studies in cooperation with such an operator. The stance we took therefore was to not establish any specific numbers whilst discussing price. Price is

compared to other services or relativized in order to quantify it.

1.7.1 Testing with a prototype

Information and interactivity surrounding a DRT system are hard to test or evaluate without an already existing system or a project in its early stages. To create an anchor for this study we decided to mock up a prototype to test out different scenarios for users to have as a reference instead of just grasping DRT as a purely abstract concept. Through the prototype, an understanding of the information and interactions needed or wanted by the test audience could be reached.

The prototype with various scenarios became a mocked stage of a development process where interactive prototype testing is useful to gather data [15]. The goal was to create something that allowed for the exploration of interaction and information that users would need and therefore the focus was more on the information and interaction than the layout and interface of the application.

1.7.2 Adapted Qualitative User Research

When trying to understand the information flows needed for DRT users, it is best to gather as much qualitative information as possible [16]. Qualitative methods, such as our

self-designed interviews, are good for when not much information about the topic is known, such as is the case here [17], [16]. We started designing in-depth, hands-on interviews with reflective questions aided by our mock-up representing a fictitious basic DRT application.

The survey had some quantitative parts as well because inspiration was drawn from mixed

(14)

methodology. Mixed methodology means that both qualitative and quantitative parts are used. This study focused on the qualitative but had some quantitative aspects, such as ratings. The explicit answers to the questions were not directly relevant to the information aspect but rather designed to start a more free-flowing discussion regarding the different subject matters.

The structure of the interview was designed to be hands-on, where at least one of the authors is present as a guide [18]. The user was then provided with instructions by the instructor to perform the task while being allowed to ask questions and receive guidance from the instructor. The important thing to remember was that the mock-up is not what we wanted to evaluate, it was there only to demonstrate the concept of DRT and to highlight the difference between different scenarios.

In these situations, it is vital to establish a good rapport with the subject [17]. If they do not feel comfortable, we would not be able to extract the needed information. Therefore, the tasks and questions are designed to be non-threatening and easy-going. These questions may also provide insight into the informational flows, and what information may be needed to help the users. Upon the completion of a task, the interviewee was asked a set of reflective questions as well as some more general ones. These were designed in an exploratory manner and tailored after the specific variant they are meant to focus on. After all the tasks have been completed, a set of more general questions related to DRT as a service was asked. These were meant to probe the user on what they have retained from the mock-up-variants, and what could be improved and/or introduced to the concept. The questions were also meant to serve as a query to the user’s knowledge on DRT as a concept, and what improvements they felt could be done to further optimize the different variants, or create new variants by mixing and matching our different time- and space variants.

Conducting the interviews remotely also led to performing several of the interviews with only one researcher present, instead of the more optimal and practical way with both present [18]. We did not want to lose any important information though, which is why the interviews took longer to complete. After each answer the information was written down. To avoid making the interviewee feel idle and just wait for the notetaker to note down the answers they were invited to freely explore the mock-up. In a few cases, this extra time exploring the prototype made the subject have a few more realizations and insights.

The sample audience made for the study was based on the principle of diminishing original insights - the larger the sample, the more you hear the same insights over and over again [19]. Since one of the aims of this study is to open up DRT to a more user-oriented research focus it is therefore not to be seen as a complete and comprehensive summation of users.

Such a summation would require a larger audience and a more iterative approach to the interviewing [19]. Due to limitations with contacts, the sample audience was picked from our surroundings which we already were in contact with and we therefore saw a need to

increase the sampling audience size to approximately twelve persons to capture a slightly wider net of insights. When composing a group to interview one intention was to gather a core of users that would have some experience using similar applications or services but also find some users that did not have the same level of experience [15].

(15)

Initially, we did a pilot study before we dove into the real thing. Conducting the first actual test was more informative than the pilot study had been, due to the fact we now had more insight into the problem areas we wanted to explore, as well as practice of the methodology.

During the test it was clear that we had to change the approach a little. Thanks to the familiarity with the subject we needed distance to procure more accurate and serious answers. But it was also clear that the familiarity rendered some interesting results. The subject was much more relaxed than if they had been interviewed by a stranger, and had no problem speaking their mind letting the interviewers have access to their spontaneous thoughts as well as the more reason-driven thoughts.

To quantify the results from the study, and be able to analyze our findings, we selected keywords from our interview summations. With the aid of these distinct keywords, we could expand them into themes that originated from the different variations in the study. The themes could then be used to indicate further work and to see what the subjects thought about in a more general sense.

1.7.3 Design thinking

When narrowing down the understanding of the project in planning and creating the

prototype, Design Thinking was a tried and proven method to approach such problems. The five-stage model of Design Thinking suited this study very well since one of its strengths is that “It’s extremely useful in tackling complex problems that are ill-defined or unknown” and is useful to understand human needs [20]. The nature of this thesis can be abstract and adopting this method allowed for an approach to this problem with a well-defined structure, making it into something comprehensible. Other methods looked at proved less suitable for our project, since they lacked either a sufficient framework to deal with the DRT concepts, or were not as user-centric in nature.

Fig 1. The Design Thinking workflow process going through the five stages in an iterative manner [20].

The five different stages of Design Thinking, fig 1, are empathize, define, ideate, prototype, and test.

● Empathize, to understand the users’ needs and understand the problem.

● Define, to analyze the information gathered and to formulate the problem area.

(16)

● Ideate, to create different suggestions for solutions.

● Prototype, to try and figure out which ideas were viable.

● Test, to test the prototype and analyze for further refinement and defining the scope or scrapping it altogether.

In the spirit of Design Thinking, we worked in an iterative cycle where the stages bled into each other and enabled quick reworks or course corrections to make the prototype better.

Creating the prototype as a visual means of communication helped conform our understanding of the problem, as well as a talking point when speaking to both our supervisor and the test subjects.

After our first cycle of iteration we conducted a minor informal pilot study to test and see what kind of results could be gained from performing interviews based on the prototype such as it was at the time. This yielded a lot of new ideas and perspectives sprouting from the subjects as well as a recurring need to refine questions and user script, but also some data on how the prototype itself should or could be tweaked to clarify the overall concept of DRT and the different variations of it. Using this feedback we continued as we could better empathize with potential users and also define the problem more narrowly with ideas that were implemented in our prototype.

1.7.4 Creating the prototype

To create the prototype we used Figma.com, a free collaborative interface online designing tool. By using Figma we could create the illusion of an interactive app presenting a clickable prototype, but with some shortcomings. Syncing the changes directly and being able to work in the same workspace at the same time made it suitable for remote collaboration as well.

When presenting the project to supervisors and test subjects we were able to use the built-in function to emulate an interactive phone in a web browser, see fig 2.

Fig 2. The presentation function of Figma allows the users to interact with the prototype through a browser window without doing any coding. The picture is an assembly of four different frames the participants could interact with.

(17)

As it is mainly a design tool, it does not allow for too advanced interactions, e.g. we did not find a way to simulate zooming in or out on a map which would increase immersion and interactivity. Another weak point of not developing our own mobile application is the fact that it is hard to simulate interaction over a period of time. The solution to the time frame where a user would put away the phone to later receive some sort of notification to extend the

interaction and delve deeper into the willingness to compromise different flows of information with each other was simply to put a “waiting” frame as a placeholder and ask the users to embrace some light roleplaying. This solution does break the immersion of using a “real”

application, but so does the entire design of the prototype to some extent. The design of the prototype should not be a major problem as we want to unearth the interaction and not have the user judge the graphical design. Though we are aware of the effect that aesthetics can play in user experience [15], [21]. One of our leading graphic design principles was to give the prototype a quite bland appearance to draw the attention away from the aesthetics.

Another part is language, where we used both bland language and light-hearted language to maintain the interviewee’s attention. Where it made sense, as in the case of the map, we used realistic content (i.e. a real map from google maps) [15]. To create an understandable route with a need for motorized transport the route went from Lomma to Malmö.

To avoid an unbearable load of work creating forks in the interactions with the prototype we decided to use a more strict flow of interaction for the user. This led to having to create a user script for the prototype but allowed for the coverage of a few of the exponential multitudes of potential paths where the flow was always going from an origin point to a destination.

When first trying to implement the three different variations we had agreed upon prior, the pre-booking with a node to node system, the check-in with zone to location system, and finally the subscription with a zone to zone system, we started out with a version with them as three different alternatives from the starting screen. We tried to inform the subjects about when and how they changed and the differences between them. This did not really work out as the subjects were confused and mixed them all together, often thinking they were different views of the same application. This assumption was not necessarily wrong, only

unnecessary in our data gathering. To avoid confusion, distinct splash-screens between the scenarios were put up where we also gave some information on the variations. The

information in the splash-screen was intended to be coupled with an oral presentation and guiding as well. To make the distinction even more obvious between the variations and to cut the connection of them being the same application we added three simple buttons labeled

“1”, “2”, and “3” to launch the different pathways.

During the creation of the variations in the prototype we focused mostly on accentuating the information flows and the difference in decision-making between them. But of course we also had to make the mock-up visually acceptable and usable. We had several different iterations of all the respective variations. Differences in color, icons, and locations were noticeable design distinctions between the iterations. Otherwise, we strived towards a uniform and bland look, as to not confuse the test subjects any further than they already were. The intention when designing the prototype was to draw attention away from the graphical

aspects and enhance it as a test environment. This confusion stemmed from the concepts of DRT already being conceptual and abstract in nature.

(18)

1.7.5 Defining the aspect of space

When starting the process of creating the interactive prototype we sketched different

scenarios of various takes on DRT where our two main aspects were place and time. These are the two main aspects that are integral for the users and most other information should be handled by the operator side of the system.

The main place aspects ended up being node to node, zone to location, and zone to zone.

The most comfortable option would be a door-to-door service, but it can be quite

resource-intensive driving all the way to the door and could also inflict on privacy issues.

Door-to-door is not the definition of node to node but was not excluded from the discussion.

The basis for the model of node to node is user-designated nodes of origin and nodes of destination. There are a few different concerns when designing for such an approach,

exactly where the nodes should be placed in regards to the user’s position, whether it should be a radius around the pick-up/drop-off spot, or if there should be predefined nodes i.e.

similar to conventional bus stops or virtual stops (VS). Sometimes a location is not accessible for a vehicle at all.

Traditionally zone to zone is a higher level of abstraction. It is directing the transportation flow from one zone to another node/zone. In the details though, the transportation is always from node to node in some way but in a larger spatial context, meaning that the traveler might need to walk a longer distance compared to the node variation. To adapt the concept to make it more user approachable, the nomenclature of the zone concept is also used as a drop-off area within a set zone of destination.

Fig 3. To the left, a picture of zone to zone in an operative way where areas of travel occur between. To the right, one of the zones split into smaller drop-off zones for further

interaction.

Zone to location is a transportation flow from an area of residence to a more central node connected either to a business hub or further travels with CPT. It is also meant as a

(19)

representation for the inverse location to zone relationship. Here lies a great opportunity and responsibility for the operators to create well-defined zones and nodes that make sense to the passengers or it will break down quickly. There is also the possibility to control the flow to a central node or several different nodes for greater destination accuracy for the passengers.

Zone to location has been remarked upon as a suitable solution for commutes, as there is a need for many to travel back and forth to work [1]. In the afternoon/evening the flow would be from a central node spreading out to an area of residence.

In the case of a trial in Sydney, where they used a many-to-one relationship, the destination node was a connection to a busy bus line with frequent departures, meaning that a strategic placement of a central node could also alleviate the time aspect [1]. Not having to rely on a stricter time limit gives the operator more freedom. One of the weaknesses a many-to-one model carries with it is how the reversed flow of transportation from the one point is not in sync and leads to empty vehicles being unutilized - an imbalance in passengers’ direction exists based on time-of-day [7].

Zone to zone is the more liberated version of zone to location since it does not control the flow as much by constraining the destination to a predefined node. Zone to zone offers the possibility to break down the operational zone into smaller pieces that can act as drop-off areas similar to a destination node with a larger radius around it. Often with regular public transport, the drop-off spot is in the vicinity of the end destination which might make this into an analogous translation for zone to zone where the passenger has to complete the last distance by foot on their own.

Naming the variants to separate them from each other in the prototype also led to some internal discussions. Using the established terms did not capture the full extent of the experience of the variants but rather led to confusion regarding what constitutes a zone and questions along that train of diversions. Conducting the interviews in Swedish also meant the words could be construed differently in translation. The term node or location became “plats”

(place) as to not evoke too much of a technical nature. The zone became “område” (area) to use a more familiar term. Zone was also removed in the first interaction in the case of zone to zone as the user instead was thrust straight into the close surroundings of the user to start the subscription model, though zone crept into the interviews as a synonym sometimes as it is descriptive.

1.7.6 Defining the aspect of time

In regards to the time aspect, we worked out three different takes to engage with three approaches to time. Time is harder to capture and present to a user as it is hard to know how to present what timespan could be deliverable without basing it upon a real system.

Pre-booking for exact travel times is very straightforward in a conventional way and lies close to the neighboring taxi service. The user makes a booking ahead of the travel and then gets a confirmatory message. As with all traffic, some deviation from the scheduled time should be anticipated and allowed. Implementing a standard on how far ahead in time the booking has to be made depends on the operator/system.

(20)

Check-in is a more loose model for time planning. The passenger would simply check-in as a way to signal the system that a passenger is ready to be picked up at a chosen

destination. The analogous representation would be going to the nearest bus stop to catch the next passing bus - but in our scenario, the bus would guaranteedly arrive sooner or later.

Another perspective on this would also be to offer a ride through interaction with the users and at the same time also give operators a way to fill vehicles already in motion.

The subscription model is more closely linked to a commute. If the user has a regular travel pattern the idea is to formalize it into a subscription for commuting to the place of education or work. The reasoning behind this was that the user has a window wherein they are open to travel, but not as exact as directly booking a specific time. This approach lies closer to the concept of mobility as a service by being more flexible to reduce the distinction between public and private transport systems [1], [22].

1.7.7 Interactive aspects of the prototype

Both place and time offer different windows to a more interactive travel experience.

Approaching DRT from the user’s perspective and disregarding the operator’s point of view we wanted to get a spread of ideas that subjects could be open to.

The interactions connected to place were often put into relation to the map in the prototype to have something in common when communicating the reasoning. It was more easily available to ask whether or not a test subject was willing to accept a pick-up node further away when viewing the distance on the map. The interactions examining place could be how far one would be willing to walk to get to a pick-up node if it would be acceptable to change the pick-up spot, and how large an uncertain drop-off zone could be among other things.

The closely-knit relation between time and space often forced the temporal to complement the spatial one. When prodding if the subjects could move to another spot, the following reasoning is often depending on the time it takes to cross the distance. The more standalone version of interactions with time was how long a user was willing to wait, and how precise the time for pick-up/drop-off would have to be of interest for the user.

Another kind of interaction to evoke a reaction was also prolonging the traveling or waiting time to either accommodate more passengers or reducing the cost of the trip. The possibility of sharing the ride would also reveal thoughts on how the users perceived DRT, either closer to a taxi service or as a public transport service. As an alternative approach to ridesharing in one scenario, the sharing option was involuntary.

1.7.8 Combining the aspects of space and time in the prototype

To evaluate the different variations, some combinations of the space and time aspects were needed. The underlying rationale behind zone to zone often being used as a means of commuting between a residential area to an area of work made it a prime candidate for combining it with the subscription model in the time aspect. The presumption was that

connecting subscriptions to travels that are made on a regular basis would make the concept understandable and grounded in relatable routines. Commuting is after all traveling from a

(21)

place of residence to a workplace or place of education which often happens routinely - but put in different wording and context here, making it harder to grasp.

On the other hand, connecting node to node with pre-booking was a way to create an introductory path for the user as it lies close to conventional modes of transport. It is a concept that is a lot easier to fathom since it practically is the procedure of booking a taxi.

Nor is CPT too far away from it either. Creating a combination this close to an already existing concept also has its drawbacks as it was unclear how well the users would understand the differentiation.

The last variant is somewhere between the two others as it entails both traveling from a zone but using a check-in mode instead of a subscription. It was more of a mixed approach since zone to location could be well suited for commuting too, but the checking in may not be a very suitable function in a commuting service. The idea was to create a scenario where participants would check-in when in their habitual zone signaling they wished to travel to one of the set locations.

Node to node Zone to location Zone to zone

Pre-booking X

Check-in X

Subscription X

Table 1. This table shows the combination of the different aspects of space and time used in the prototype to create various scenarios.

(22)

2. Result

2.1 Node to node

Fig 4. Three frames from the prototype illustrating the node to node scenario. In sequence:

choosing a node of origin, choosing a node of destination, and confirming the nodes from a zoomed-out perspective.

2.1.1 Keywords

Punctuality; Price; Communication; Cheap; Delays; Easy-to-use; Smooth; Comfortability;

Useable; Shared; GPS-dots; Guide; Interactivity; Dependability;

2.1.2 Summary of node to node interviews

A majority of the people interviewed responded that it was important for the service to be punctual. Looking at fig 5 and 6, pick-up time was important but a slightly higher frequency answered that drop-off should be exact or categorized as exact. Some concerns identified by the participants were that the trip may be delayed a lot and that it did not feel competitive.

The concept itself was perceived as brand new and had a difficult learning curve. However, many also saw potential and thought it could be a smooth way of traveling with a high level of personal comfort because of the system with pick-up and drop-off nodes that the users pinpoint themselves, thereby creating a trip of their design. A recurring topic was the

(23)

uncertainty of the concept and how many different aspects depended on the price point and how dependable and reliable it would be.

Fig 5. In this diagram, we see how the participants responded regarding punctuality or what they foresee as punctuality of pick-up time

Fig 6. In this diagram, we see how the participants responded regarding the punctuality of drop-off time

Information the users would like to have available and access to in the application was different interactions in the form of deals and offers where you can ride with others, suggestions to get better prices, and places where many travel from - so-called hot spots.

These interactions may end up resulting in a reduced price or a different pick-up time. An

(24)

interactive map with information concerning what vehicles are nearby, how many seats are open, and estimated costs were also high in demand.

Many participants also stated that they would like to know whether the trip was shared or not, preferably as soon as possible. Another aspect was the safety aspect of traveling alone or at certain times. In order to ride in a smaller vehicle with a driver and potential

co-passengers, information about the driver and potentially the passengers were requested to increase the sense of safety.

Most respondents also mentioned a tutorial of some kind, at least upon a first installation.

General pricing, based on distance or time, was also interesting to a vast majority as well as general information about the current traffic and if there were expected delays and so forth.

The make and model of the vehicle picking you up, and perhaps the license plate number was also requested by many to distinguish it from other vehicles.

Waiting times were often requested to be fairly low, in the area around 15-30 minutes, with a few outliers indicating they could wait much longer. Many participants compared this service to a taxi service. Generally, the interviewees felt they could pay a little more than for a bus or other comparative public transports, but less than for a taxi due to the sharing aspect and the potential extra pick-ups that might occur.

When it comes to making this variant of DRT attractive almost everyone mentioned the pricing. Offers and discounts such as getting the first trip for free or half-price to prove that the service would turn out dependable and safe were also mentioned a lot. Commercials and ads in social media were also a common answer taking more of a marketing approach.

(25)

2.2 Zone to location

Fig 7. Three frames from the prototype illustrating the zone to location scenario. In

sequence: choosing the zone of origin, choosing the location represented by Malmö C in this case, and confirming the transport with a view of the assigned VS.

2.2.1 Keywords

Confusing; Cheap; Price; Pointless; Complicated; GPS-dot; Interactivity; Real-time; Shared;

Notifications; Depends; Time; Offers;

2.2.2 Summary of zone to location interviews

Summarizing the participants of the study’s view on waiting times revealed that they had a fairly similar view on the subject. A lot of them answered around 15-20 minutes as the expected and accepted waiting time, a few a little longer depending mainly on whether they waited at home or outside. Waiting in the comfort of the home would lend itself to accept longer waiting times. Time spent waiting could be discussed and negotiated with in terms of price, but also good communication could serve as a way to keep the users’ content.

Summarily, the test subjects also agreed that the scenario as a whole was a somewhat confusing concept with many different parts that were hard to wrap their head around. Many felt that the zones were a bit confusing, especially since you only have a zone upon

departure, not upon arrival. With the concept of VS, there was also a general concern for delays and other disturbances. Participants mostly liked the system with check-ins, but not

(26)

the zone to location-system. They felt that the difference between this system and CPT was unclear. Reflecting more freely on the concept as seen in fig 8, most responded that this model seemed complicated or confusing above other factors.

Fig 8. In this diagram, we see that many users thought zone to location was complicated and/or confusing when reflecting on the concept. A few stated that the service needed to be cheap while some users had no specific input on it at all.

In regards to interactivity and conveying information, the interviewees wanted GPS dots representing the vehicle coming to pick them up on an interactive map with real-time

updates. The information should also be updated and notifications could be used so that the information would be available without having to keep track of the map all the time. The participants answered that they would like to track the vehicle coming to pick you up from the moment it is assigned and possibly even showing the route it would probably take on the map for them to follow the progress. According to several of the interviewees, they would not mind seeing other vehicles in general on the map. Another feature suggested was to have your location showing up as a GPS dot as we had in the mock-up.

In terms of information in the application, they suggested seeing the estimated time for the pick-up as soon as it was available, a tutorial on how to use this system, and a live map showing current traffic information. The participants also said they would like to see nearby vehicles connected to the service and what kind of vehicle that was coming to pick them up.

Many of the participants said they would not like to go further than 500 meters to get to the pick-up location or around 10 minutes. When asked how big the pick-up zones should be, the majority felt they should be quite small, maybe a kilometer wide. A size that would let the users walk comfortably to any pick-up location they were assigned. If the pick-up spot was made so that the travelers would have to wait outside, the general feeling was that they could wait up to 10 to 15 minutes at the longest, but longer if they could wait at home. The

(27)

interviewees generally would not like to pay much more than for a bus, with a few

exceptions. They felt this was very similar to how a bus works, and therefore did not see a reason to pay much more. The majority felt that being assigned a VS would be okay and some preferred it to choose one themselves, on the condition that the information was conveyed in time.

To make the participants of our study interested in using this service, they all felt that it had to be cheap. Some also mentioned that it had to find a gap to fill, otherwise it would just be a pointless alternative to regular public transport, such as buses. Offers and price reductions for the first few trips was another thing mentioned, to let the service show its dependability.

The participants were divided on whether they would rather travel from a zone, as in our case, or to a zone from a set location.

2.3 Zone to zone

Fig 9. Four frames from the prototype illustrating the zone to zone scenario. In sequence:

choosing the VS of origin based on the zone that the user is in, choosing the zone which splits into smaller drop-off zones, and setting up the schedule for the transport.

2.3.1 Keywords

Delays; Easy; Smooth; Confusing; Dependability; Price; Shared; GPS-dots; Comfortability;

Interactivity;

2.3.2 Summary of zone to zone interviews

The concerns for this scenario were connected to different variations of delays, bad information, and a steep learning curve with the zones. Around half of the interviewees answered that they did not like the zone-system while approximately the other half responded to it in a positive manner and felt it was a simple way of booking a trip. The skeptical respondents felt that this scenario might not work in real life because it was too complicated and would not be attractive for public transport users. The subscription part of

(28)

this variant garnered mostly positive feedback, although coupled with some interjections of difficulty in keeping the estimated time schedule and expressions of uncertainty about how it would work in a live scenario. The subscription was interesting, but perhaps in a different form.

In regards to information within the zone to zone-application, the vast majority of the users wanted a tutorial to learn this novel system. The participants wanted to be able to see estimations of the cost and travel time of the trips. Another thing many of the subjects wanted to see was whether or not the ride was shared or the possibility of it being shared further along the trip.

Fig 10. A vast majority of the subjects specifically mentioned wanting a tutorial in the application.

An interactive map was suggested by several interviewees. This map would be updated dynamically, and provide basic traffic information and possible delays that could occur, or other information that could be of impact for the passenger. Some participants also

mentioned that they would appreciate a mode where they were offered information to see if there were any vehicles close to them that could pick them up, perhaps for a better price since time would be shorter.

The subjects also wanted to see information about their subscriptions in a simple way and to be able to handle their ongoing subscriptions. In one instance the person being interviewed mentioned a wish to see if and when the subscription is ending, the cost of it, and possible obstacles that could occur. They also wanted a tutorial for the pure subscription part of this particular scenario as it was hard to grasp in comparison to previous scenarios.

The participants had a few different ideas about the sizes of the zones, but they were not too far off from each other. A few suggested a more direct measure of the size being about a kilometer across, and some anchored their views in a more familiar way to have the different districts of the city as zones in themselves (as long as they were not unreasonably large).

The general consensus within the subjects was that a user should be able to walk to any pick-up point or point of final destination within the zone with relative ease and comfortability.

(29)

They felt that punctuality was very important. Neither the pick-up nor drop-off time should vary greatly, but it was less acceptable for the drop-off to be late.

They felt this service did not stand out enough or was comfortable enough to deserve a higher pricing. The subjects had some concerns about the punctuality this service could offer in real life, and also the difficulty of actually planning something and then putting it into practice in the form of a subscription was a major hindrance to the majority of them.

Increasing the attractiveness of this alternative the respondents expressed a wish for a cheap and affordable price. The users mostly answered they would like to pay equal to, less, or just a tiny bit more than for a bus. Offers and timed discounts, such as a lower cost for a longer waiting or travel time, would also draw users in according to the data. An alternative was targeting companies and letting them pay for the trip for their employees like a company public transport was also suggested.

2.4 DRT in general

2.4.1 Keywords

Offers; Interactivity; Cheap; Price; Dependability; Smooth; GPS-dots; Climate; Shared;

Together;

2.4.2 Summary of general DRT questions interviews

When asked about which variant was the most liked, in regards to time, the pre-booking variant was by far the most popular among our subjects and many stated that this was very similar to a taxi or Uber service. The participants also gave many suggestions on how to improve this concept, primarily by adding more features and increasing the interactivity.

Some stated that they liked the subscription variant, mainly because it seemed like the most interesting and novel concept.

A majority of the answers also relayed that the context was important when choosing the preferred time variant, and the answer could change based on the scenario. Pre-booking was the overall favorite because it met many of the contextual criterias, but other scenarios could be highly suitable in certain circumstances.

Node to node was the absolute favorite when the interviewees were asked. Zone to location was also mentioned a few times, and zone to zone was mentioned by only a singular

participant. When they said node to node, many also hinted about future interactivity they would like, such as notifications for cheaper rides. Some would also like to see hot spots or areas with increased activity, where faster or cheaper pick-up might occur.

When asked about improvements in general, almost all the participants mentioned pricing. It had to be affordable or cheap to be able to compete with other services to niche itself within the public transport sector. Every user agreed that it should be cheaper than a conventional taxi service, but there was a willingness to pay more for the extra flexibility in relation to CPT.

(30)

An interesting point that came up once was the organization of transport by a company in relation to zone to location. It was described as a company picking up employees and co-transporting them to work.

Punctuality and dependability were also mentioned almost as often as pricing. The

interviewees stated they would have to be enticed by low prices or convinced of the service’s dependability and punctuality by being offered free or discounted rides at the start. To

convince the users of dependability and punctuality, good information was also a common theme. As touched upon before, an interactive map, GPS-positioning, and a functioning modern application to service the users were mentioned to reinforce the reliability. When the subjects were asked how to improve the mock-up from an interaction and information

perspective many proposed more interactivity, as seen in fig11.

Fig 11. In this diagram, we see how many of the interviewees explicitly stated they wanted more interactivity than was included in the mock-up. The Other category did not mention it per se but talked about other topics.

Prior to questions directly addressing possible climate effects, some users pointed to the advantage of ridesharing as a boon for the environment. These users made the connection to possibly decreased emissions of carbon dioxide by cars, busses, and other vehicles.

As seen in fig 12, the vast majority of the participants of our study said that the climate perspective would play a huge role in whether or not they would use a DRT service. The ones who answered that it would not impact their decision did not see how this could affect the climate. One point of view was that CPT already addresses the issue.

(31)

Fig 12. This diagram shows how many of the interviewees stated that the impact on the environment would affect their decision in using a DRT service.

To increase the motivation for the environmental-friendly aspect of ridesharing and DRT, some interviewees suggested using offers and climate-friendly routes and showing users information about how they affected the environment by choosing a DRT service. A more interactive approach, offering to book a seat in a DRT vehicle already driving, at a reduced price could increase interest according to some participants.

When the participants were asked to rate the different aspects of our variations, pre-booking with a node to node-system won decisively. As shown in fig 13 and 14 the preference for the aforementioned aspects was high.

Fig 13. In this diagram, we see how many of the interviewees graded the different location aspects with the highest grade.

(32)

Fig 14. A greater number of participants rated Pre-booking highest in comparison to the other time variants.

(33)

3. Analysis

3.1 Node to node analysis

There were several interesting findings in our interviews in connection to node to node.

3.1.1 Pick-up time and drop-off time

One of our questions in relation to node to node was how punctual the pick-up time needed to be. According to the answers, it is clear that the actual time of pick-up cannot differ greatly from the estimated one. Half of the interviewees stated that the time had to be “very exact”

or similar to a bus in the sense that it can arrive a few minutes later or earlier. A few stated a more precise acceptable would be 15 or 20 minutes. Many also stated that depending on the pricing of the service, a longer wait could be acceptable. Geography was also a factor, as many participants could wait longer if they were in a more rural area. Another factor is the flow of information. If the users would be notified of irregularities and disturbances in the schedule in good time, maybe an hour ahead, the accepted fluctuation could be greater.

While the pick-up time seemed to be a bit flexible, the drop-off time turned out to be less so.

The absolute majority of the participants in our study answered that while it was acceptable to differ a few minutes from the estimated time, it had to be pretty exact. This points to the fact that it is easier to handle the pick-up time as a user and the drop-off is out of their hands making them dependent on a means of reliable transport. Again, the pricing of the service could make a real difference here. If it were to be a cheap service in comparison to other modes of public transport, the participants stated they could easily overlook a much greater delay. Though that could disqualify a cheap DRT service as means of travel for important appointments. In a more open reasoning not specifically connected to the context of the prototype in which the users experienced the drop-off time, the goal of the travel

arrangement had a heavy impact on how exact it had to be. If the transport was of a business- or work-oriented nature the time was of the essence, but if it was more of an entertainment trip or for going shopping the limits were more lax.

3.1.2 Wait-time

Geography also played a part in how long people were willing to wait. Most participants in the study lived in Malmö, a city with good public transport. These interviewees were not satisfied with a longer wait than around 15 to 20 minutes generally. On the other hand, those that originated from places in the countryside, further from a frequent public transport system or where the buses don’t depart as often, were willing to wait much longer and in some cases even hours. This seems to indicate that this system would be well utilized in a more rural and less urban setting.

Tying into this is the fact that many of our interviewees failed to see when they would actually utilize this service in a more urban setting. Unless it was cheap or turned out more reliable than regular public transports or taxis there was no need identified. Some theorized that this service could niche itself during the nighttime. At late or early hours there are not as

Information flows in Demand Responsive Public Transport: Interactivity, information, and flexibility in a modern ridesharing service