Rw3 guillermo velázquez tesina mcs by City Sciences

FINAL MASTER THESIS “PUBLIC TRANSPORT USERS' PREFERENCES AND WILLINGNESS TO PAY FOR A PUBLIC TRANSPORTATION MOBILE APP IN MADRID”

Guillermo Velázquez Romera Andrés Monzón de Cáceres (Tutor) Date: 07/09/2015

Final Master Thesis

Guillermo Velรกzquez

Index 1.

Introduction and purpose of the study.................................................................................. 4

Smartphone Apps for multimodal Public Transport.............................................................. 6

2.1.

Multimodality in public transportation ....................................................................... 6

2.2.

Smart Mobility. Technology. Integration. ITS. Smartphone Apps. ............................... 6

2.3.

Public Transport Mobile Applications. Impact of real time information on travellers .... 7

2.4.

Public transport usersâ&#x20AC;&#x2122; requirements and willingness to pay for public transport apps.... 8

Methodological procedure for the analysis .......................................................................... 9 3.1.

Online survey methodology ...................................................................................... 9

3.2.

Data Analysis. ........................................................................................................ 10

Case study: Madrid .......................................................................................................... 12 4.1.

Sample characterization .......................................................................................... 12

4.1.1. Sociodemographic characterization ..................................................................... 12 4.1.2. Travel behavior and most frequent trip ................................................................ 14 4.1.3. Technological capabilities .................................................................................. 17 5.

Results and discussion ..................................................................................................... 20

Conclusions .................................................................................................................... 38

Aknowledgements ........................................................................................................... 38

References ...................................................................................................................... 39

Final Master Thesis

Guillermo VelĂĄzquez

Introduction and purpose of the study

Cities play a key role in our society; they generate high levels of wealth, employment and productivity, and often serve as the engines of their national economies (OECD, 2013). According to the Green Paper of the European Commission just under 85% of the EU’s GDP is generated in urban areas, which are home to over 60% of the EU’s population (European Commission, 2007). With the growth of urban areas in recent years, the role of urban mobility has become increasingly important. Large volumes of traffic in urban areas produce increased congestion and exponentially growing economic externalities (e.g. congestion costs about 1% of the EU’s GDP each year), social externalities (e.g. 69% of traffic accidents take place in cities) and environmental externalities (e.g. 25% of CO2 emissions in cities are caused by transportation, also traffic-related air pollution has been identified as a public health priority in Europe), which are considered key elements leading to the degradation of the quality of life in cities (Transportation White Paper, 2011). In this line, the actions financed by the EU in research and innovation have been focused on the development of new strategies for urban mobility that aim at reducing these externalities (TRIP, 2013). The current EU Research and Innovation programme, Horizon 2020, poses among urban mobility-related objectives, the following priorities: - Getting a seamless transport system to achieve a better mobility, less congestion and increased safety. - Getting new developments to halve the use of combustion vehicles, promoting nonmotorized and public transport travel. From a technical point of view, to achieve the policy objectives of the EU on urban mobility there is a need of improved ways to balance demand while maintaining the actual capacity of the transport system. Therefore mobility management and integrated planning should be desirable policy objectives (ECTRI, 2011). Today's cities are characterized by more diffuse patterns of mobility, with longer travel distances, increased multimodality, and a continued growth of the level of motorization (e.g. positive growth rates 2008-2013 in all EU countries. Eurostat 2015). Also the profile of the user is changing, with increasing modal offer and available information not only time savings are important for the users, but also comfort, safety and travel time reliability become important, and therefore they should be given greater emphasis in transport policy and performance management. Today, the Smart Cities are presented as a solution to achieve a more sustainable urban development while increasing the quality of life of their citizens through the use of new technologies (Neirotti, 2013). Smart Mobility is based on innovative and sustainable ways to provide transport for the inhabitants of cities, enhancing the use of fuels or vehicle propulsion systems that respect the environment, supported by technological tools and a proactive behaviour of citizenship (Neirotti, 2013). In urban transport, the purpose of the Smart Cities is to develop flexible information systems in real time to support decision-making in the use and operation of different modes of transport generating a positive impact, saving users time and improving efficiency and service quality.

Final Master Thesis

Guillermo Velázquez

In this context, several solution types are being introduced in the world’s cities. One of the most extended being the use of mobile apps for providing the user with contextualized static and real time transport information. They enable the improvement of the abovementioned factors acting on the demand side resulting in more efficient journeys for individual travellers, and improved satisfaction with the service. (Skelley et Al., 2013) with a lower level of investment than that of infrastructure deployment or an increase in the level of service. This study aims to be a first step in the analysis of Madrid’s public transport user’s requests for a public transportation app and of their willingness to pay for such a service. The results complete with a case study the literature on which capabilities are the most required, and sets the ground for an estimation of the utility of transportation mobile applications for the users of a public transportation network. The study is based on a survey that was conducted among PT users in Madrid, containing items that asked about their travel behaviour, their degree of technological skills and capabilities, as well as their main expectations on the possibility of using a new app and their main desired capabilities. In the same manner and with a more commercial approach an inquiry on their willingness to pay was included. The paper is organized as follows. First, a brief review is provided on the state of the art on multimodality in transportation, Smart Mobility, mobile transportation applications and the latest research regarding app features and willingness to pay of the users. Then, section 3 contains a description of the applied methodology for the survey and the data collection process, as well as on the techniques of statistical analysis that have been used for the development of this paper. Then, section 4 details the case study of Madrid. Finally, results and conclusions extracted from the research are shown in section 5, along with some recommendations for the development of this kind of apps, and future lines of research.

Final Master Thesis

Guillermo Velázquez

2. Smartphone Apps for multimodal Public Transport 2.1. Multimodality in public transportation Urban mobility is changing rapidly. Complexity is rising every day with larger cities, multiple transportation modes and new service models available. Effectively exploiting the advantages of multimodality is one of the main challenges of modern urban transportation systems, it being reflected on the EU’s objectives for transportation: obtaining sustainable and seamless transport systems. Differences in density and infrastructure availability within urban areas, as well as changing circumstances and citizen needs make distinct transportation modes more efficient in different areas of the city, at different times and for different purposes. With multiple modes available the number of trips whose optimal solution is a multimodal chain increases. To provide door to door type mobility solutions only a multimodal mobility offer can compete with private vehicles in flexibility, convenience and cost. Hence, in order to meet mobility needs of passengers and goods making the most efficient use of the network, traditional public transport operators must achieve seamless and resilient systems in which the users must be able to easily transfer from one mode of transport to another. Also to achieve sustainability non-motorized mobility, and in particular which does not depend on private vehicles, should be enhanced, while the use of private vehicles should evolve towards being socially and economically less attractive (Neirotti, 2013). However, enhancing multimodal transport does not require significant restructuring of transport networks; instead it’s more related with improving the availability of information and the perceived quality of the service (van Nes, 2002). Therefore, with the triumph of the economics of customer-focused services, public transport operators are challenged to evolve towards the provision of integrated mobility, moving towards schemes where mobility is offered in a way that can balance demand without over dimensioning the capacity of the transport system, providing greater information to the user, increasing resilience to incidents and making the integration of services reality, contributing to the goal of achieving a more livable city (Banister, 2008).

2.2. Smart Mobility. Smartphone Apps. The concept of Smart Cities has been extending since the 90’s in parallel with the development of services through internet. The term Smart City has become in recent times synonymous with extensive use of information technology cities, although a Smart City means much more than that (De Santis et al. 2014). The initial publication "European Innovation Partnership on Smart Cities and Communities" focused efforts in 3 main fields: energy, mobility and ICT (EU, 2013). However, later on, the concept of Smart City has drifted away from pure technology-centered approaches, towards the use of ICT as a vehicle to promote the objectives of cities from the multidimensional perspective.

Final Master Thesis

Guillermo Velázquez

Smart Mobility remains as one of the main dimensions that the concept integrates. With new models that the Smart vision brings like sharing schemes, flexible design of facilities and services, user centric service provision or open access to information in real time, being applied also to the urban mobility field. Urban mobility can make the best use of information technologies to harvest and analyse large volumes of data in a structured and integrated way to improve mobility services, or many other services in the city. One of the main challenges to enhance multimodality, integration of modes both at the physical and at the information level, can be effectively addressed through the use of ICTs, and more particullarly, on the user level, through the use of smartphone APPs. Technological innovations like real-time information and trip planners are proving essential to obtain the maximum benefit from multimodal scheme, as travellers can be unaware of viable modal alternatives. (Kenyon et al. 2002) There are already outstanding examples in this initiatives such as fare integration between modes and operators carried out in London and Madrid, offer open data in Santander, Chicago or San Francisco or private initiatives as Moovit multimodal planners, Citymapper or Waze area. Another example is the Smile initiative undertaken in Vienna which has developed an app for route optimization and allows the purchase of tickets online.

2.3. Public Transport Mobile Applications. Impact of real time information on travellers. In the academic literature there are several studies on the utility of travel time information systems with regard to travel time savings (Toledo and Beinhaker, 2006), as well as on the cost of travel time variability (Bรถrjesson, Fosgerau and Karlstrรถm, 2008) and increased user satisfaction with the service (Dziekan and Kottenhoff, 2007). There is a proved positive impact of improved information of public transport networks on urban economy, mobility and environment as it results in more efficient journeys for individual travellers, and improved satisfaction with the service. (Skelley et Al., 2013) However some studies have shown that currently, information plays a minor role in modal choice when compared to that in journey planning and execution for a chosen mode (Kenyon et al. 2003), today individuals still have to make their travel decisions under uncertain circumstances with respect to the travel time; they are not able to predict the exact travel time or arrival time before starting their trips, given a departure time (Zheng Li et Al, 2010), uncertainty that increases in the case of multimodal chains. Mobile solutions appear as a new solution to tackle these problems, presenting travellers with a personalized set of modal options that include soft aspects like comfort or convenience, which can be determinant to persuade a modal change (Kenyon et al. 2003). The general increase in penetration rates of mobile devices in most cities, mainly smartphone devices, and even in underdeveloped economies, suggest that transport authorities have the opportunity to integrate their offer taking it to a multimodal level in an effective and cost efficient way. Public transportation agencies can play an active role by releasing high quality data, thoughtfully engaging with users and developers, and advocating for high quality, affordable mobile service options. (Moss et al 2011)

Final Master Thesis

Guillermo Velรกzquez

Many transport authorities have already done so, or enabled others to do it for them. Some examples of this are the apps developed by Transport for London, the Region Autonome des Transports Parisiens or Singapore on the public side, or the opening of data to the private sector that has developed apps in cities like San Francisco, Boston or London. However, as some futuristic gurus predict (Gartner, Cisco, IDC), most of these solutions still haven’t reached all the value that could be derived with regard to personalized travel assistants and mobility as a service offerings.

2.4. Public transport users’ requirements and willingness to pay for public transport apps. The design and implementation of this kind of solutions has been mainly empirical with few studies on user preferences for this solutions. More studies exist on the information required; some examples are the ones carried about traffic information needs for real time journey planning like the ones carried out by Chorus et al. (2007) or Zografos et al. (2010) and on multimodal traveller information systems by Moss et al (2011) or Kramers (2014) stating that much of the functionality is lacking in current solutions, if they should support societal goals like sustainability. In this regard some studies are being conducted to address how to encourage sustainable behaviour in user decision making, like the one carried out for a biking social app by Felix et al promoting the four E’s model, Enable, Engage, Encourage, Exemplify (Felix et Al. 2013). To identify main features demanded by the user, static information features or some basic real time ones, may hold a certain degree of correlation in the requirements for the already existent web ones. However some studies have been carried out for mobile solutions like the one of Schaffer et al. (2014) stating that the selection of features and personalization of app offerings for mobility should focus on specific user groups or app purpose and that future designs that improve the presentation of alternatives in the modal split of the routing proposal and functions related to utility need to be addressed. (Schaffer et al. 2014) Regarding the user’s expected utility and willingness to pay for transportation information several studies have been carried out. The information on user willingness to pay is useful as a way to measure utility and therefore, to evaluate the potential benefit derived from the implementation of projects. (Misham et al. 2007) Up to now Studies centered on the impact of implemented real time information systems like the studies carried by Watkins et al. 2011 Our study addresses this problems, by inferring which capabilities are the most necessary for the users in each of the main transit information fields: static information, real time information, visualization, search and in app-utilities. Also some information on which barriers are seen as the most problematic for the adoption of these solutions, (According to the literature Flexibility and Trust act as key elements with regard to UTIS acceptance. (Skelley et Al., 2013)) has been analysed. Finally, a user segmentation approach has been taken to analyse willingness to pay by different user groups, which in the future can lead to modelling users utility and address the need personalized information offering for the users of multimodal chains.

Final Master Thesis

Guillermo Velázquez

3. Methodological procedure for the analysis The process for the elaboration of this study had two main parts: Data collection through an online survey and analytic procedures applied for the data.

3.1. Online survey methodology Several of the items contained in the survey were fixed with regard to the European project OPTICITIES, mainly Likert-scale type questions associated to the Theory of Planned Behaviour that states that attitudes toward behaviour, subjective norms, and perceived behavioural control, together shape an individual's behavioural intentions and behaviours (Ajzen, 1991). Further elements of the online survey and the survey methodology have been based on the practice of social research (Babbie, 2010). A set of questions was designed addressed at observing the users characteristics in the areas of travel behaviour, technological skills and predisposition to adopt both new technological solutions in general and mobile apps in particular. Then a set of questions concerning the users preferences within a transportation mobile app were introduced, as well as questions relative to their willingness to pay for such an application. In the months of February and March 2015, a data collection campaign was conducted in one of the main interchange stations of the city in coordination with the transport authority of Madrid. A personal intercept methodology combined with a web-based survey was used. More than 2.000 users of different ages, in different transportation modes, at different days of the week and hours of the day were intercepted in Avenida de América interchange station, and were asked to fill in the online survey. The combination of personal intercept interviews and online questionnaires allows achieving good quality of data without entailing big expenditures. The process consisted of two steps: - First, ‘Interviewers’ stopped members of specific target groups (public transport users) onsite (in the exchange station), they briefly explained the purpose of the project (transport assistant application to be implemented), and invited them to participate in the survey, providing them with an identification code, while keeping record of them in order to aim to ensure an initial adjustment of the sample. - Second, Respondents, identified through their codes, answered a web questionnaire afterwards (at home, on-board a vehicle, etc.) through a mobile device or a personal computer getting as a reward a direct incentive and/or taking part in a prize draw. Through this process a total of 386 complete answers to the questionnaire were obtained, allowing the use of a representative sample. All the respondents provided contact details to be used in the future in order to proceed with subsequent phases of the study.

Final Master Thesis

Guillermo Velázquez

Figure 1. Location of Avenida de América interchange station

Figure 2 shows the location of the data collection site within the subwaypolitan area of Madrid. The station combines four subway lines (one ring line, three regular lines), eleven urban bus lines, and twelve inter-urban bus lines, as well as park and ride, with a demand of approximately 170.000 travellers per day.

3.2. Data Analysis. In order to analyze the collected information, first a descriptive analysis of the sample has been conducted in terms of sociodemographic characterization, travel behaviour and technological capabilities of the users. Then a verification of the validity of the survey and of the representativeness of the sample with regard to Madrid’s public transport user population has been conducted. A process of post stratification adjustment, used to adjust for sample weights so that the estimated joint distribution of a set of post stratifying variables matches the known population joint distribution (Chen et Al. 2012), was introduced to ensure non response from elderly travellers didn’t bias the results of the study. For user travel behaviour description and most preferred features for the app bar charts have been used to represent qualitative variables (Mode for the most frequent trip, Reason for most frequent trip, Frequency of most frequent trip,…) whilst histogram representations have been used for the quantitative ones (Age distribution, Average duration of most frequent trip,…).

Willingness to pay

Final Master Thesis

Guillermo Velázquez

First a correlation analysis was conducted between theoretical percentages of time saved for the user’s most frequent trip thanks to the app and the amount users were willing to pay for them. Given that both variables are ordinal type a Kendall’s Tau b test was selected for the analysis. (Field, 2013). This procedure is a non-parametric test which definition consists in let (x1, y1), (x2, y2), …, (xn, yn) be a set of observations of the joint random variables X and Y respectively, such that all the values of (xi) and (yi) are unique. Any pair of observations (xi, yi) and (xj, yj) are said to be concordant if the ranks for both elements agree: that is, if both xi > xj and yi > yj or if both xi < xj and yi < yj. They are said to be discordant, if xi > xj and yi < yj or if xi < xj and yi > yj. If xi = xj or yi = yj, the pair is neither concordant nor discordant. (Nelsen, 2001). The Kendall τ coefficient is defined as:

Then, in order to analyse which population traits were relevant regarding the willingness to pay of the users, an analysis was conducted against the traits of gender, age, main mode for the most frequent trip, most frequent trip duration, educational level, income and smartphone use. One-way ANOVA tests were carried out to examine whether there were statistically significant differences in willingness to pay among different user groups for each trait. Depending on the acceptance or rejection of the assumption of equal variances among groups, a Scheffe or GamesHowell post hoc test was used to verify where the difference lied. The questions concerning payments that the users had to answer were:

-Question 1 Consider the following percentages of time saving for your most frequent trip. How much would pay monthly to achieve these savings time? The percentages shown to the users were 10%, 20%, 30%, 40% and 50%. Available answers ranged from 0€/month to 30€/month with 5€ increase intervals. Answering was mandatory with an “I don’t know” option also available.

-Question 2 To benefit from the advantages of the service offered by the app I would be willing to pay: Available options were: Yearly subscription, Monthly subscription, Weekly subscription, Daily subscription, Payment only on download, Payment only for extra services and Not willing to pay.

Final Master Thesis

Guillermo Velázquez

4. Case study: Madrid Madrid is a city of some 3.5 million inhabitants with its subwaypolitan area reaching 6 million. Three ring motorways surround the city: M-30, M-40 and M-50 and it has five fully operational (out of seven projected) multimodal interchange stations that connect the main accesses to the city with the public transportation network. According to the Household Mobility Survey of Madrid Region (CRTM, 2005, about 5 million displacements on public transport occur on a working day in the subwaypolitan area, which covers a developed area of 1,037 km2 . This fact gives an idea of the importance of public urban transport management in a city with the size of Madrid. Madrid is making a great effort to improve its mobility by enhancing public transportation and multimodality, e.g. recently a public bicycle sharing system (BiciMAD) has been introduced in the central area of the city. With this aim, the transport authority has started to work on new solutions that exploit their operation data in order to improve the information services provided to the users and reduce the uncertainty of the user associated to disruptions in the service by unexpected incidents. Following this line the transport authority has taken part in the European project OPTICITIES belonging to the 7th Framework Program in which Madridâ&#x20AC;&#x2122;s main contribution is to emphasize urban experimentation in mobility fields like data processing, use of intensity of traffic data, integration of data related to vehicle traffic and pedestrian demands, analysing of public, scholar and worker transport focused on a better management for the city transit, experimentation on the functions and use of big urban commuting facilities already working in the city, etc. As a pilot city for the project, two transportation tools are being tested, one of them being a mobile APP aimed at improving the real time information on the transport network to the final users. The APP will help them finding the closest stop, providing information on the arrival time of the next service, informing about a disruption on their trip, etc. Such tool will be made available in the near future. An ex-ante survey has been elaborated in order to evaluate the impacts on travel behaviour for the specific tool being tested in Madrid. This research is part of the EU 7 FP project OPTICITIES, which deals with developing effective, scalable and transferrable ITS tools that embrace multimodality, enhance the exchange and interoperability of data and ease the provision and accessibility of mobility services for managers and users.

4.1. Sample characterization 4.1.1. Sociodemographic characterization The survey involved a total of 200 women and 186 men, which represents almost equality in gender distribution within the sample. In terms of age, three groups were established: Young (18-30), adults (30 to 45 years) and senior travelers (50-70 years). The following tables show the distribution of respondents according to their age and sex is shown:

Final Master Thesis

Guillermo VelĂĄzquez

Table 1. Number of surveyed travelers by gender. USERS GENDER Men Women Total

Absolute value 186 200 386

% 48 52 100

Table 2. Number of surveyed travelers by age and gender. USERS AGE GROUP 15-30 31-50 51-70

SEXO Men Women Men Women Men Women

Total

Absolute value 66 93 74 78 46 29 386

% 17 24 19 20 12 8 100

The most limiting group to be targeted in order to achieve a representative sample were senior travellers, as they are less adroit when it comes to the use of technology. In total 75 complete answers were obtained for this group. This set a percentage of senior travellers around 18%, which isn’t accurate with regard to the existent proportions in Madrid’s population. Therefore a weighing of the sample to match the population age proportions has been carried out, in an attempt to reduce the impact of senior traveller population unit non-response (Little et Al. 2005).

According to the population by age published on the last household survey published by the Spanish Statistical Office (2011), in Madrid region the existent proportions for the different groups between 15 and 65 years old (70% of the total population) are the ones presented on Table 1.

Table 3. Age group proportions in Household Survey carried out by INE (Spanish Statistical Office) in 2011, in the selected sample and the weighing factors. AGE GROUP

Household survey

Study sample

Adjustment factor

15-30

28%

49%

0,57

31-50

48%

37%

1,30

51-70

23%

18%

1,33

Population distribution for the selected intervals after the adjustment of the different age groups is shown in the histogram on figure 3.

Final Master Thesis

Guillermo Velázquez

Figure 2. Population distribution for the selected intervals after adjustment.

4.1.2. Travel behaviour and most frequent trip To analyze the travel behaviour of respondents they were presented a series of questions asking detailed information on their most frequent trip. The results to this section of the survey are: Figure 3. Reason for most frequent trip (% of total surveyed sample) 65,3%

14,5%

11,0%

Work

Studies

Final Master Thesis

Leisure (Cultural activities, sports,...)

2,7%

3,1%

2,5%

0,8%

Stroll

Other

Shopping

Picking up / Dropping someone

Guillermo Velรกzquez

Figure 4. Frequency of most frequent trip (% of total surveyed sample)

Figure 5. Histogram of time spent on most frequent trip

Final Master Thesis

Guillermo Velรกzquez

Figure 6. Time spent on most frequent trip (% of total surveyed sample)

Figure 7. Figure 7. Mode choice for your most frequent trip (% of total surveyed sample)

64,8%

17,1%

Multimodal

Metro

13,2%

Bus

1,8%

1,3%

0,5%

0,3%

0,5%

Car (driver)

Train

Motorbike

Walking

Car (passenger)

Other

Figure 8. First mode choice for multimodal chain users (% of total multimodal users)

43,5%

27,1%

10,3%

7,3%

6,6% 2,4%

Bus

Metro

Final Master Thesis

Walking Car (driver)

0,7%

0,3%

Train Car (passenger)Motorbike Public Bike

0,3%

Bike

Guillermo Velรกzquez

Figure 9. Second mode choice for multimodal chain users (% of total multimodal users)

44,3%

35,7%

12,8%

5,1% Metro

Bus

Train

Walking

0,9%

0,8%

0,4%

Bike

Car

Public Bike

As it has been explained, all surveyed users were public transportation users as they were intercepted within a public transportation interchange station. A great weigh of intermodal trips characterizes the sample as expected with the combinations Bus-Subway and Subway-Bus accounting for more than 50% of multimodal trips.

4.1.3. Technological capabilities In order to know the extent to which the users were comfortable with the use of different technological tools, they were asked about both their current device usage and their proficiency level in the use of each of the more broadly extended devices, even if they were not users of some of them.

Table 4. Level of use for technological tools by age (% of users that indicated using each tool for each age group). Young

Adult

Senior

Avg.

76,3%

76,0%

78,0%

Smartphone

80,5% 89,3%

82,9%

73,3%

83,7%

Tablet

44,0%

49,3%

32,0%

43,8%

GPS

20,1%

27,6%

26,7%

24,4%

From the above data we can infer technological accessibility, it being an indicator also of potential reach of the app’s penetration in the different user groups by age. However simple usage doesn’t imply domain of the tool, which can also limit the willingness to use complex applications on it, which can be observed in the results of Table 3.

Final Master Thesis

Guillermo Velázquez

Table 5. Proficiency level for technological tools by device, device usage and age group (% of row) PC

Users

Non users

Age group Young Adults Seniors Young Adults Seniors

Global

Tablet

Users

Non users

Age group Young Adults Seniors Young Adults Seniors

Global

Smartphone

Users

Non users

Age group Young Adults Seniors Young Adults Seniors

Global

1-Very bad 0,8 0,0 1,8 12,9 2,8 11,1 2,3 1-Very bad 0,0 0,0 0,0 7,9 5,2 25,5 6,2 1-Very bad 0,0 0,0 1,8 5,9 11,5 55,0 4,2

Bad 1,6 0,9 5,3 0,0 5,6 11,1 2,6

Bad 1,4 2,7 0,0 9,0 9,1 15,7 6,7

Bad 0,7 0,0 0,0 11,8 15,4 15,0 2,6

Average 4,0 10,3 17,5 6,5 11,1 27,8 9,8

Average 2,9 8,0 29,2 13,5 14,3 25,5 13,2

Average 3,5 4,8 29,1 5,9 34,6 10,0 10,1

Good 14,1 16,4 28,1 16,1 11,1 11,1 16,6

Good 10,0 10,7 12,5 16,9 18,2 13,7 14,0

Good 7,0 17,5 20,0 23,5 11,5 15,0 13,7

5-Very Good 79,7 72,4 47,4 64,5 69,4 38,9 68,7 5-Very Good 85,7 78,7 58,3 52,8 53,3 19,6 59,8 5-Very Good 88,7 77,8 49,1 52,9 26,9 5,0 69,4

Avg 4,7 4,6 4,1 4,2 4,4 3,6 4,5

Avg 4,8 4,7 4,3 4,0 4,1 2,9 4,1

Avg 4,8 4,7 4,1 4,1 3,3 2,0 4,4

As Table 5. shows, technological penetration rates are high among Madrid’s public transport users, with ~78% of users being also users of personal computers and up to 86% stating a good level of proficiency in using them, and ~84% of users being users of smartphones with the same rate stating a good or very good proficiency level on their use. Also 56% of them already employ an application to plan their trips, but 75% showed interest in testing the APP.

Income level of respondents was assessed as monthly income per household. The population was segmented into intervals of € 1,000 income except the interval of 1000-2000 € monthly, which given its large volume was divided into two subsections of 500 € to ease comparability. The results are shown in Figure 11. It is noteworthy that one-third of respondents left this section blank.

Final Master Thesis

Guillermo Velázquez

Figure 10. Histogram of income levels

Final Master Thesis

Guillermo Velรกzquez

5. Results and discussion This section contains the results obtained in the analysis detailed in section 3. First, the main results for preferences regarding App features and contents are explained, structured in the same way as in the survey: Static information features, real time information features, special event information features, search options, in-app services availability and App customization options. Second, the main results for the analysis of the willingness to pay of the users, with the study of relevant differences among segmentations of the population according to gender, age, mode used for the most frequent trip, income, duration of most frequent trip and smartphone usage are included. APP features Users were queried on desired capabilities for the application. Several questions on desired features were raised in which the user had to rank capabilities either by selecting his or her three most desired capabilities or by ranking among a series of options split by static information, real time information, information regarding special events in the network and search options and other in-app services. The main sets of information that users desire to receive through the app are: - Regarding static information: Schedules and timetables of the different transportation modes (selected among the three most desired static information services by 70% of users), then routing information between any two points (61%), visualizing a map with public transportation stops on it (41%), checking information on an specific stop (address, stop number, etc.) (39%) and being able to check graphically the itinerary of a line (37%). Figure 11. Most desired static information capabilities (% of respondents selecting the option among their three most desirable ones).

70%

61% 41%

39%

37%

Schedules and Routing Visualizing a Checking being able to timetables of the information map with public information on check different between any two transportation an specific stop graphically the transportation points stops on it itinerary of a line modes * Other options were visualizing the map of any station, and visualizing a general map of the city.

Final Master Thesis

Guillermo Velรกzquez

Figure 12. Most desired static information capabilities (% of respondents selecting the option as their most desired, second most desired and third most desired) 1st

2nd

3rd

Total

Information about the schedules of public transport services

21%

18%

30%

70%

Planning my trip from any origin to any destination (regardless of my position)

27%

16%

19%

61%

Viewing a map of the city

21%

12%

41%

Viewing public transport stops on a map

19%

12%

39%

Basic information of public transport stops (street adress, stop code, etc.)

11%

18%

39%

Graphically displaying the itinerary of public transport lines

15%

13%

37%

Maps of specific stops / transit stations

13%

100%

300%

Total

* Other options were visualizing the map of any station, and visualizing a general map of the city

- Regarding real time information: Checking information on the closest public transportation stops in real time (selected among the three most desired real time information services by 87% of users), routing information from the users real time location (74%), visualizing in real time the closest public transportation lines (69%), real time information on the closest public transportation services (65%), visualization of the user location on the map (59%). - Regarding special events information:

Suggestions of public transportation alternatives when the user is planning his route (selected among the three most desired services related to events by 66% of users), information on lines and stops affected by the event (62%), alternative route suggestions when the user is affected by an event (54%), real time information on events that may affect the userâ&#x20AC;&#x2122;s trip (50%).

- Regarding search options: Users want the application to let them filter by closest stop (important or very important for 80% of users), filter by closest stop of a given transportation mode (ex. Bus) (79%), filter by public transport stop name (73%), search by closest stop of a given line (74%) and, with less significance, search by public transportation stop code (49%). - Regarding in-app services: Information on the userâ&#x20AC;&#x2122;s transportation card: balance, recharge points and card information points were selected among the three most desired services by 63% of users.

Final Master Thesis

Guillermo VelĂĄzquez

Guidance information: The main desired means to receive guidance information are plain map indications, arrow movement through a map, or through panoramic image strings, being considered as more important than camera visualization or radar indications. Payments through the app: selected among the three most desired services by 37% of users. Point Of Interest (POI) information: selected among the three most desired services by 27% of users.

Figure 13. Most desired in-app services (% of respondents selecting the option among their three most desirable ones) 63% 43%

37% 27%

Information on the user’s transportation card

Real time traffic information

Payments through the app

Point Of Interest information

* Other options were park and ride information integration, BiciMAD information integration, fares information.

App customization The main desired features were: saving favourite stops and lines (73% of users selected it among their three most desirable features), activating alerts on events affecting any of the user’s favourite stops/lines (70%), and personalized information on trip optimization (55%).

Figure 14. Most desired personalized features for real time information (% of respondents selecting the option among their three most desirable ones)

73%

70% 55%

Saving favourite stops Activating alerts on Personalized and lines events affecting any of information on trip optimization the user’s favourite stops/lines * Other options were gamification, social network integration, proximity alerts or being able to receive regular information bulletins.

The possibility of changing the appearance of the app (colours, theme, font size, etc.) was desired by more than 50% of the surveyed users, and considered as not necessary by 24% of them.

Final Master Thesis

Guillermo Velázquez

Willingness to pay In order to infer the willingness to pay of surveyed users, questions presenting different expected amounts of travel time savings thanks to the use of the application, with different possible monthly prices for the app as well as one question confronting different possibilities of payment (monthly, yearly, on download,…) were analysed. The aggregation of the obtained results is shown in Figure 15. First, Kendall’s Tau b test was conducted between theoretical percentages of time saved for the user’s most frequent trip thanks to the app, and the amount users were willing to pay for them, obtained from the first of both items and in order to prove the correlation among the two variables. General Considering all cases surveyed, a correlation analysis between time and price variables has been run. Figure 15 shows a linear construct with the average amounts users are willing to pay monthly for each percentage of time saved in their most frequent trip thanks to the use of the app.

Figure 15. Willingness to pay for % of time saved. 10

€/month

0 0%

10%

20%

30%

40%

50%

% of time saved

The results of the Kendall’s Tau b test are brought here directly from SPSS. It is noteworthy that the number of cases is much larger than the initial sample as each respondent answered five different items containing their willingness to pay for each different amount of time saved.

Final Master Thesis

Guillermo Velázquez

Table 6. Kendall’s-Tau b analysis output on SPSS. Case summary figures. Case Processing S ummary Cases Valid N €/month * % time

M issing Percent

1718a

88,9%

Total

Percent

214,300

11,1%

Percent

1932,300

100,0%

a. Number of valid cases is different from the total count in the crosstabulation table because the cell counts have been rounded.

Table 7. Kendall’s-Tau b analysis output on SPSS: Crosstabulation for % of time saved and €/month €/month * % time Crosstabulation % time 10

€/month

Total

271

212

147

111

821

104

403

236

114

102

348

344

339

1718

Total

Table 8. Kendall’s-Tau b analysis output on SPSS. S ymmetric Measures Value

Asymp. Std. Error

Ordinal by Ordinal

Kendall's tau-b

N of Valid Cases

,351

Approx. T b

Approx. Sig.

,017

19,866

,000

1718

a. Not assuming the null hypothesis. b. Using the asymptotic standard error assuming the null hypothesis.

The result of the test shows that as expected there is some degree of correlation among time savings and willingness to pay for the use of the app for the general population with a value of τ = 0,351 and a significance of the result of 0,000.

Final Master Thesis

Guillermo Velázquez

Sig. is the probability that the relationship obtained could have occurred by chance, and is interpreted as follows: Less than .01 - Highly significant Between .01 and .05 - Moderately significant Between .05 and .10 - Somewhat significant Above .10 - Not significant However the correlation is not very strong, therefore we assume that there must be different population segments in which the willingness to pay is both greater and can be modelled in a more precise way. Hence, as a second step and in order to analyse which population traits were relevant regarding the willingness to pay of the users, an analysis by each of the main traits within the population has been conducted. For each of the sample segmentations we have displayed a figure showing the average amounts that the user groups in it are willing to pay for savings of time thanks to the use of the app. Also a Table that shows their payment modality preferences has been included. Results of the ANOVA tests for the different grouping variables are brought here directly from SPSS. Tables corresponding to post-hoc tests have been placed in ANNEX I.

By Gender An analysis taking the gender variable as grouping factor was carried out. The two groups within the population are Men (48% of the sample) and Women (52% of the sample). The independent variable is the different percentages of time saved, whilst the dependent variable is the amount the users are willing to pay for each of those savings of time. The average values of the dependent variable are presented below split by population group. Figure 16. Willingness to pay for time saved, by gender group. 9 8 7

â&#x201A;Ź/month

6 5 4

Men

Women

2 1 0 0%

10%

20%

30%

40%

50%

% of time saved

Final Master Thesis

Guillermo VelĂĄzquez

A one-way ANOVA test has been conducted to examine whether there were statistically significant differences among the different user groups’ willingness to pay. Post hoc tests are not necessary in this case as the segmenting variable has only two groups.

Table 9: One-way ANOVA test output in SPSS: Sample description split by gender Descriptives N

M ean

Std.

Std. Error

Deviation

-10% time

-20% time

-30% time

-40% time

-50% time

95% Confidence Interval

M inimum

M aximum

for M ean Lower

Upper

Bound

Men

182

1,40

3,569

,264

,88

1,92

Women

167

2,27

4,681

,362

1,56

2,99

Total

349

1,82

4,155

,222

1,38

2,25

Men

178

2,81

4,703

,352

2,12

3,51

Women

169

3,99

5,735

,442

3,12

4,86

Total

347

3,38

5,256

,282

2,83

3,94

Men

178

4,49

5,536

,415

3,67

5,31

Women

166

6,12

6,678

,518

5,10

7,14

Total

344

5,28

6,159

,332

4,62

5,93

Men

178

6,53

6,907

,518

5,51

7,55

Women

161

7,76

7,225

,569

6,64

8,89

Total

339

7,12

7,076

,384

6,36

7,87

Men

175

9,08

8,227

,622

7,85

10,31

Women

163

10,01

8,243

,645

8,73

11,28

Total

338

9,53

8,235

,448

8,65

10,41

Table 10: One-way ANOVA test output on SPSS: ANOVA for each % of time saved by gender. ANOVA Sum of Squares Between Groups AHORRO10

67,035

Within Groups

5937,894

346

17,162

Total

6004,929

347

120,141

Within Groups

9431,054

344

27,416

Total

9551,195

345

227,088

Within Groups

12785,851

342

37,386

Total

13012,938

343

128,810

Between Groups AHORRO30

AHORRO40

M ean Square

67,035

Between Groups AHORRO20

Between Groups

Final Master Thesis

128,810

Sig.

3,906

,049

4,382

,037

6,074

,014

2,584

,109

Guillermo Velázquez

Within Groups

16798,016

337

Total

16926,826

338

72,435

Within Groups

22780,305

335

68,001

Total

22852,740

336

Between Groups AHORRO50

49,846

1,065

,303

The results reveal equal variances assumption is rejected for 10%, 20% and 30% time savings with F(1, 345))= 3,906, p-value of 0,001 for 10% savings, F(1, 347)= 4,382, p-value of 0,037 for 20% savings and F(1, 343))= 6,074, p-value of 0,014 for 30% savings. The assumption is met for 40% and 50% levels with F(1,337)= 2,584 and F(1,336)= 1,065 and p-values of 0,109 and 0,303 respectively. Differences in the sample size for each option are due to respondents selecting “I don’t know” in some of the items. The results revealed statistically significant differences among both groups for the lower tier of the time savings, which implies that in the studied sample men were less willing to pay than women for small percentage savings of time thanks to the use of the app. Differences in willingness to pay among the two groups for the higher end of time saving were found to be not significant. Regarding the modality of payment, results by each variable group are shown on Table 11 with percentages for each option calculated over the total group population in order to ease comparisons.

Table 11: Preferred payment option of the user, split by gender. (% of row).

GENDER

Not willing to pay

Payment on Only for extra download services

Yearly Monthly Weekly Daily subscription subscription subscription subscription

Men

65,9%

16,5%

10,0%

2,6%

4,4%

0,3%

Women

74,9%

12,9%

6,1%

2,3%

2,7%

0,3%

0,7%

By Age group An analysis taking the age group variable as grouping factor was carried out. The three groups within the population are young, comprising population aged between 15 and 30 (28% of the sample), Adults, comprising population aged between 31 and 50 (48% of the sample) and Senior, comprising population aged between 51 and 70 (23% of the sample). Again the independent variable is the different percentages of time saved, whilst the dependent variable is the amount the users are willing to pay for each of those savings of time on their most frequent trip. The average values of the dependent variable are presented in figure X below split by population group.

Final Master Thesis

Guillermo Velázquez

Figure 17. Willingness to pay for time saved, by age group. €12,00

€/month

€10,00

€8,00

Adult

€6,00

Young Senior

€4,00 €2,00 €-

10%

20%

30%

40%

50%

% of time saved

A one-way ANOVA test has been conducted to examine whether there were statistically significant differences among the three different user groups’ willingness to pay.

Table 12. One-way ANOVA test output in SPSS: Sample description split by age group Descriptives N

M ean

Std.

Deviation

Error

95% Confidence Interval for M ean Lower Bound

M inimum

M aximum

Upper Bound

Adult

178

2,21

4,715

,353

1,51

2,91

Young

1,82

4,342

,473

,88

2,76

Senior

1,00

2,197

,236

,53

1,47

Total

349

1,82

4,155

,222

1,38

2,25

Adult

178

3,77

5,721

,429

2,92

4,61

Young

3,49

5,426

,589

2,32

4,66

Senior

2,46

3,789

,414

1,64

3,28

Total

347

3,38

5,256

,282

2,83

3,94

Adult

178

5,58

6,475

,485

4,62

6,54

Young

5,30

6,228

,676

3,96

6,65

Senior

4,59

5,335

,592

3,41

5,77

Total

344

5,28

6,159

,332

4,62

5,93

Adult

174

7,74

7,592

,575

6,61

8,88

Young

7,31

7,192

,786

5,75

8,88

Senior

5,57

5,473

,608

4,36

6,78

Total

339

7,12

7,076

,384

6,36

7,87

Adult

173

9,85

8,443

,642

8,58

11,12

-10% time

-20% time

-30% time

-40% time

-50% time

Final Master Thesis

Guillermo Velázquez

Young

10,79

8,737

,961

8,88

12,70

Senior

7,58

6,925

,763

6,06

9,10

Total

338

9,53

8,235

,448

8,65

10,41

Table 13. One-way ANOVA test output on SPSS: ANOVA for each % of time saved by age group. ANOVA Sum of Squares Between Groups AHORRO10

42,612

Within Groups

5919,705

345

17,159

Total

6004,929

347

98,708

49,354

Within Groups

9452,487

343

27,558

Total

9551,195

345

54,638

27,319

Within Groups

12958,300

341

38,001

Total

13012,938

343

244,057

142,028

Within Groups

16662,641

336

49,591

Total

16926,826

338

462,990

231,495

Within Groups

22389,750

334

67,035

Total

22852,740

336

Between Groups AHORRO30

Between Groups AHORRO40

Between Groups AHORRO50

M ean Square

85,224

Between Groups AHORRO20

Sig.

2,483

,075

1,791

,168

,719

,488

2,864

,061

3,453

,033

The results reveal equal variances assumption is rejected for 50% time savings with F(2, 334))= 3,453, p-value of 0,033. For 40% savings and 10%, F(1, 336)= 2,864, p-value = 0,061 and F(2, 345))= 2,483, p-value = 0,075, which is above the standard 0,05 limit, but yet quite close meaning that there is some signinficant disparity in the means of the different groups. The assumption is met for 20% and 30% levels with F(2,345)= 1,791 and F(2,343)= ,719 and p-values of 0,168 and 0,488 respectively. In order to statistically address which groups held the differences that the indicator had pointed out, post-hoc analysis were ran. First Levene’s statistic was calculated for all of them. Then, depending on the statistic result the Games Howell test was analyzed for Sig. in Levene’s < 0,05, and the Scheffe test was analyzed for Sig. in Levene’s test > 0,05.

Final Master Thesis

Guillermo Velázquez

Table 14. Levene’s statistic test output on SPSS for each % of time saved split by age group Test of Homogeneity of Variances Levene Statistic

df1

df2

Sig.

-10% time

9,307

346

,000

-20% time

6,850

344

,001

-30% time

4,509

341

,012

-40% time

7,776

336

,000

-50% time

2,468

335

,086

Games-Howell test was used for 10% and 40% time savings, as Sig. in Levene’s test < 0,05, while Scheffe test was used for 50% time, with Sig = 0,086 (Table 14.). Post-hoc tests (results are included on Table 31 in ANNEX I) revealed statistically significant differences between Senior travellers and Adult travellers and Young travellers. Young and Adult segments of the sample reported significantly higher willingness to pay for time savings thanks to the transportation app. There were no significant differences between the young and adult groups Regarding the modality of payment, results by each variable group are shown on Table 15 with percentages for each option calculated over the total group population in order to ease comparisons. Table 15. Preferred payment option of the user, split by age group. (% of row). AGE GROUP Young Adult Senior

Not Payment on Only for Yearly Monthly Weekly Daily willing to download extra services subscription subscription subscription subscription pay 70% 8% 10% 2% 8% 1% 1% 68% 19% 7% 3% 3% 0% 1% 76% 12% 9% 3% 0% 0% 0%

By main mode An analysis taking the variable mode used for the most frequent trip as grouping factor was carried out. The four categories with enough representativeness within the population are multimodal, bus only, car only as driver and subway only. Again the independent variable is the different percentages of time saved, whilst the dependent variable is the amount the users are willing to pay for each of those savings of time on their most frequent trip. The average values of the dependent variable are presented on the figure below split by population group.

Final Master Thesis

Guillermo Velázquez

Figure 18. Willingness to pay for time saved, by mode used for most frequent trip. €16,00 €14,00 €12,00

€/month

€10,00

multimodal bus only

€8,00

car only (driver) €6,00

subway only

€4,00 €2,00

€0%

10%

20%

30%

40%

50%

% of time saved

A one-way ANOVA test has been conducted to examine whether there were statistically significant differences among the different user groups willingness to pay.

Table 16. One-way ANOVA test output in SPSS: Sample description split by mode Descriptives N

Only bus

Std.

95% Confidence

Deviation

Error

Interval for M ean Lower

Upper

Bound

M inimum M aximum

2,28

4,841

,750

,77

3,80

,82

1,987

,711

-,86

2,51

1,34

3,873

,480

,38

2,30

M ultimodal

213

1,81

4,213

,288

1,24

2,38

Total

328

1,75

4,190

,231

1,30

2,21

3,55

6,090

,959

1,61

5,49

3,30

3,986

1,426

-,09

6,69

2,54

4,091

,501

1,54

3,54

M ultimodal

211

3,49

5,402

,372

2,76

4,22

Total

326

3,30

5,218

,289

2,73

3,87

3,93

5,688

,896

2,12

5,74

6,60

5,932

2,121

1,56

11,64

Only car (driver) AHORRO10 Only subway

Only bus Only car (driver) AHORRO20 Only subway

Only bus AHORRO30

M ean

Only car (driver)

Final Master Thesis

Guillermo Velázquez

Only subway

3,27

4,622

,572

2,12

4,41

M ultimodal

210

5,94

6,435

,444

5,06

6,82

Total

323

5,17

6,095

,339

4,50

5,83

5,28

6,167

,988

3,28

7,28

10,72

8,456

3,024

3,53

17,91

4,77

5,325

,656

3,46

6,08

M ultimodal

205

7,92

7,400

,516

6,91

8,94

Total

318

7,01

7,035

,394

6,24

7,79

8,01

7,990

1,270

5,44

10,57

15,70

10,935

3,910

6,40

24,99

6,72

6,593

,808

5,11

8,34

M ultimodal

203

10,22

8,279

,581

9,07

11,36

Total

317

9,34

8,147

,457

8,44

10,24

Only bus Only car (driver) AHORRO40 Only subway

Only bus Only car (driver) AHORRO50 Only subway

Table 17. one-way ANOVA test output on SPSS: ANOVA for each % of time saved by mode.

ANOVA Sum of Squares Between Groups AHORRO10

10,066

Within Groups

5710,684

323

17,680

Total

5740,883

326

48,728

16,243

Within Groups

8799,649

321

27,413

Total

8848,377

324

439,064

146,355

Within Groups

11533,282

319

36,154

Total

11972,346

322

728,078

242,693

Within Groups

14973,044

314

47,685

Total

15701,123

317

998,151

332,717

Within Groups

19986,256

313

63,854

Total

20984,406

316

Between Groups AHORRO30

Between Groups AHORRO40

Between Groups AHORRO50

M ean Square

30,199

Between Groups AHORRO20

Sig. ,569

,636

,593

,620

4,048

,008

5,090

,002

5,211

,002

The results reveal equal variances assumption is rejected for 30%, 40% and 50% time savings with F(3, 322))= 4,048, p-value of 0,008 for 30% savings, F(3, 317)= 5,090, p-value of 0,002 for 40% savings and F(3, 316))= 5,211, p-value of 0,002 for 50% savings. The assumption is met for 10% and 20% levels with F(3,326)= 0,569 and F(3,324)= 0,593 and p-values of 0,636 and 0,620 respectively. Differences in the sample size for each option are due to respondents selecting “I don’t know” in some of the items.

Final Master Thesis

Guillermo Velázquez

In order to statistically address which groups held the differences that the indicator had pointed out, post-hoc analysis were ran. First Levene’s statistic was calculated for all of them. Then, depending on the statistic result the Games Howell test was analyzed for Sig. in Levene’s < 0,05, and the Scheffe test was analyzed for Sig. in Levene’s test > 0,05. Table 18. Levene’s statistic test output on SPSS for each % of time saved split by mode Test of Homogeneity of Variances Levene Statistic

df1

df2

Sig.

AHORRO10

1,554

324

,201

AHORRO20

1,932

322

,124

AHORRO30

1,560

319

,199

AHORRO40

5,854

314

,001

AHORRO50

1,528

313

,207

Games-Howell test was used for 40% time saving, as Sig. in Levene’s test < 0,05, while Scheffe test was used for 30% and 50% timesavings, with Sig = 0,199 and 0,207 respectively. (Table 18). Post-hoc tests (results are included on Table 32 in ANNEX I) revealed statistically significant differences between Subway travellers and Multimodal and car as a driver travellers. Car and Multimodal segments of the sample reported significantly higher willingness to pay for time savings thanks to the transportation app. There were no significant differences between the other group combinations. Regarding the modality of payment, results by each variable group are shown on Table 19 with percentages for each option calculated over the total group population in order to ease comparisons.

Table 19: Preferred payment option of the user, split by main mode. (% of row). Not willing Payment on Only for extra Yearly Monthly Weekly Daily to pay download services subscription subscription subscription subscription MODE Multimodal 67 18 9 3 2 0 1 Bus only 73 15 7 2 1 1 0 Car only (driver) 62 31 0 0 7 0 0 Subway only 77 5 7 3 9 0 0

By income group An analysis taking the income variable as grouping factor was carried out. Again the independent variable is the different percentages of time saved, whilst the dependent variable is the amount the users are willing to pay for each of those savings of time on their most frequent trip. The average values of the dependent variable are presented on the figure below split by population group.

Final Master Thesis

Guillermo Velázquez

Figure 19. Willingness to pay for time saved, by income level.

10 9

€/month

1. Les s than 1000€ / month

2. 1000€–1500€ / month 3. 1500€–2000€ / month

4. 2000€-3000€ / month 4

5. 3000€-4000€ / month

6. More tha n 4000€- month

2 1 0

0,1

0,2

0,3

0,4

0,5

% of time saved

A one-way ANOVA test has been conducted to examine whether there were statistically significant differences among the different user groups willingness to pay.

Table 20. One-way ANOVA test output in SPSS: Sample description split by income Descriptives N

M ean

Std.

95% Confidence

Deviation

Error

Interval for M ean Lower

Upper

Bound

M inimum M aximum

1. M enos de 1000€ /

3,17

5,747

,929

1,28

5,05

3,18

5,564

,887

1,39

4,98

1,76

4,654

,706

,34

3,19

1,19

3,504

,471

,25

2,14

2,37

5,074

1,034

,23

4,51

1,63

2,994

,553

,49

2,76

2. 1000€–1500€ / mes 3. 1500€–2000€ / mes AHORRO10

4. 2000€-3000€ / mes 5. 3000€-4000€ / mes 6. M ás de 4000€ / mes

Final Master Thesis

Guillermo Velázquez

Total

230

2,15

4,675

,308

1,54

2,76

5,82

6,196

,953

3,90

7,75

4,09

6,889

1,098

1,87

6,31

2,98

5,496

,834

1,30

4,66

2,25

4,092

,547

1,16

3,35

3,70

6,132

1,286

1,03

6,37

2,89

5,344

,987

,87

4,91

233

3,57

5,717

,374

2,83

4,30

6,56

6,484

1,031

4,48

8,65

6,38

8,472

1,350

3,65

9,12

4,39

5,357

,812

2,75

6,03

4,87

5,558

,743

3,38

6,36

5,68

6,190

1,298

2,98

8,37

4,08

6,005

1,109

1,81

6,35

230

5,31

6,384

,421

4,48

6,14

7,88

7,003

1,095

5,66

10,09

7,14

8,291

1,321

4,46

9,81

7,16

7,413

1,141

4,85

9,46

6,80

6,346

,849

5,10

8,50

7,79

6,458

1,354

4,98

10,60

4,64

5,952

1,164

2,25

7,04

227

6,97

6,992

,464

6,05

7,88

10,36

8,734

1,387

7,55

13,17

1. M enos de 1000€ / 2. 1000€–1500€ / mes 3. 1500€–2000€ / mes AHORRO20 4. 2000€-3000€ / mes 5. 3000€-4000€ / mes 6. M ás de 4000€ / mes Total 1. M enos de 1000€ / 2. 1000€–1500€ / mes 3. 1500€–2000€ / mes AHORRO30 4. 2000€-3000€ / mes 5. 3000€-4000€ / mes 6. M ás de 4000€ / mes Total 1. M enos de 1000€ / 2. 1000€–1500€ / mes 3. 1500€–2000€ / mes AHORRO40 4. 2000€-3000€ / mes 5. 3000€-4000€ / mes 6. M ás de 4000€ / mes Total 1. M enos de AHORRO50

1000€ /

Final Master Thesis

Guillermo Velázquez

2. 1000€–1500€ 39

8,33

8,867

1,423

5,45

11,21

9,86

8,092

1,217

7,41

12,32

9,84

6,979

,934

7,97

11,71

10,41

8,136

1,706

6,87

13,95

6,48

6,492

1,256

3,90

9,06

228

9,34

7,936

,526

8,30

10,38

/ mes 3. 1500€–2000€ / mes 4. 2000€-3000€ / mes 5. 3000€-4000€ / mes 6. M ás de 4000€ / mes Total

Table 21. One-way ANOVA test output on SPSS: ANOVA for each % of time saved, by income. ANOVA Sum of Squares Between Groups AHORRO10

29,531

Within Groups

4855,105

223

21,772

Total

5002,757

228

350,761

70,152

Within Groups

7233,061

227

31,864

Total

7583,822

232

202,154

40,431

Within Groups

9145,500

224

40,828

Total

9347,653

229

194,370

38,874

Within Groups

10866,716

221

49,171

Total

11061,086

226

351,205

70,241

Within Groups

13946,820

221

63,108

Total

14298,025

226

Between Groups AHORRO30

Between Groups AHORRO40

Between Groups AHORRO50

M ean Square

147,653

Between Groups AHORRO20

Sig.

1,356

,242

2,202

,055

,990

,424

,791

,557

1,113

,354

The results reveal equal variances assumption is accepted for all time savings with significance levels of the result above the standard 0,05 limit, although for the 20% saving variable Sig = 0,055 indicating that there is some significant disparity in the means of the different groups. Hence we can conclude for this section that there is no statistically significant difference in willingness to pay for different income level groups. Regarding the modality of payment, results by each variable group are shown on Table 23 with percentages for each option calculated over the total group population in order to ease comparisons. Final Master Thesis

Guillermo Velázquez

Table 22. Preferred payment option of the user, split by income group. (% of row).

INCOME Less than 1000€ / month 1000€–1500€ / month 1500€–2000€ / month 2000€-3000€ / month 3000€-4000€ / month More tan 4000€ / month

Not willing to pay

Only for extra services

Payment on download

Yearly Weekly Monthly subscriptio subscriptio subscription n n

Daily subscri ption

By Duration of most frequent trip

Figure 20. Willingness to pay for time saved, by duration of their most frequent trip. 12

€/month

8 >60 45-60

30-45 15-30

0-15 2

0 0%

10%

20%

30%

40%

50%

% of time saved

A one-way ANOVA test has been conducted to examine whether there were statistically significant differences among the different user groups willingness to pay.

Final Master Thesis

Guillermo Velázquez

Table 23. one-way ANOVA test output in SPSS: Sample description split by duration of most frequent trip. Descriptives N

0-15

M ean

Std.

95% Confidence Interval

Deviation

Error

for M ean Lower

Upper

Bound

M inimum M aximum

2,39

5,595

1,132

,05

4,73

1,61

3,580

,375

,86

2,35

103

2,33

5,046

,498

1,34

3,32

1,71

3,584

,406

,90

2,51

1,08

3,026

,417

,24

1,92

Total

349

1,82

4,155

,222

1,38

2,25

0-15

4,59

6,170

1,284

1,93

7,25

2,89

4,834

,503

1,89

3,89

100

3,77

5,752

,576

2,63

4,91

3,55

5,169

,578

2,40

4,70

2,74

4,671

,651

1,43

4,05

Total

347

3,38

5,256

,282

2,83

3,94

0-15

6,00

5,970

1,243

3,43

8,58

5,02

6,852

,718

3,59

6,44

4,85

6,290

,634

3,59

6,11

5,68

5,821

,651

4,38

6,98

5,61

5,280

,736

4,13

7,09

Total

344

5,28

6,159

,332

4,62

5,93

0-15

8,09

7,210

1,501

4,98

11,20

6,72

6,940

,734

5,26

8,18

6,48

7,126

,718

5,05

7,90

1530 30AHORRO10 45 4560 >60

1530 30AHORRO20 45 4560 >60

1530 30AHORRO30 45 4560 >60

15AHORRO40 30 3045

Final Master Thesis

Guillermo Velรกzquez

4578

7,25

6,979

,790

5,67

8,82

8,43

7,356

1,039

6,34

10,52

Total

339

7,12

7,076

,384

6,36

7,87

0-15

9,86

8,550

1,832

6,05

13,67

8,66

8,053

,849

6,97

10,35

8,76

7,309

,748

7,28

10,25

9,62

8,790

,987

7,65

11,58

12,18

8,876

1,236

9,70

14,66

338

9,53

8,235

,448

8,65

10,41

60 >60

1530 30AHORRO50 45 4560 >60 Total

Table 24. One-way ANOVA test output on SPSS: ANOVA for each % of time saved, by duration of most frequent trip. ANOVA Sum of Squares Between Groups AHORRO10

17,213

Within Groups

5936,077

343

17,306

Total

6004,929

347

94,534

23,633

Within Groups

9456,661

341

27,732

Total

9551,195

345

55,053

13,763

Within Groups

12957,885

339

38,224

Total

13012,938

343

164,046

41,012

Within Groups

16762,780

334

50,188

Total

16926,826

338

489,435

122,359

Within Groups

22363,305

332

67,359

Total

22852,740

336

Between Groups AHORRO30

Between Groups AHORRO40

Between Groups AHORRO50

M ean Square

68,853

Between Groups AHORRO20

Sig. ,995

,410

,852

,493

,360

,837

,817

,515

1,817

,125

The results reveal equal variances assumption is accepted for all time savings with significance levels of the results above the standard 0,05 limit for all of the variables. Hence we can conclude for this section that there is no statistically significant difference in willingness to pay for different income level groups.

Final Master Thesis

Guillermo Velรกzquez

Regarding the modality of payment, results by each variable group are shown on Table 27 with percentages for each option calculated over the total group population in order to ease comparisons.

Table 25. Preferred payment option of the user, split by duration of most frequent trip. (% of row).

Most frequent trip duration >60 45-60 30-45 15-30 0-15

Not Only for Yearly Weekly Daily Payment on Monthly willing to extra subscriptio subscriptio subscri download subscription pay services n n ption 73% 20% 2% 2% 2% 0% 0% 70% 18% 8% 2% 2% 0% 0% 71% 12% 9% 3% 4% 0% 0% 72% 10% 11% 1% 4% 1% 2% 60% 20% 5% 10% 5% 0% 0%

By Smartphone use According to Table 5. 84% of PT users in Madrid are users of smartphones with the same rate stating a good or very good proficiency level on their use. Figure 21. Willingness to pay for time saved, by Smartphone use. €9,00 €8,00 €7,00

€/month

€6,00 €5,00 Yes

€4,00

€3,00 €2,00 €1,00 €0%

10%

20%

30%

40%

50%

% of time saved

Final Master Thesis

Guillermo Velázquez

Table 26. One-way ANOVA test output in SPSS: Sample description split by smartphone use. Descriptives N

M ean

Std.

95% Confidence Interval

Deviation

Error

for M ean Lower

Upper

Bound

M inimum M aximum

2,45

3,999

,518

1,41

3,48

AHORRO10 Yes

289

1,69

4,181

,246

1,20

2,17

Total

349

1,82

4,155

,222

1,38

2,25

4,47

5,538

,733

3,00

5,94

AHORRO20 Yes

290

3,17

5,182

,304

2,57

3,77

347

3,38

5,256

,282

2,83

3,94

6,44

6,454

,864

4,70

8,17

AHORRO30 Yes

288

5,05

6,086

,358

4,35

5,76

344

5,28

6,159

,332

4,62

5,93

7,28

7,097

,950

5,37

9,18

AHORRO40 Yes

283

7,09

7,084

,421

6,26

7,91

339

7,12

7,076

,384

6,36

7,87

9,14

8,119

1,069

6,99

11,28

AHORRO50 Yes

280

9,61

8,271

,494

8,64

10,58

338

9,53

8,235

,448

8,65

10,41

Total

Table 27. One-way ANOVA test output on SPSS for each % of time saved, by smartphone use. ANOVA Sum of Squares Between Groups AHORRO10

44,527

Within Groups

5976,206

346

17,272

Total

6004,929

347

80,569

Within Groups

9470,626

344

27,531

Total

9551,195

345

89,239

Within Groups

12923,699

342

37,789

Total

13012,938

343

1,696

Within Groups

16925,130

337

50,223

Total

16926,826

338

10,641

22842,099

335

68,185

Between Groups AHORRO30

Between Groups AHORRO40

M ean Square

44,527

Between Groups AHORRO20

Between Groups

Sig.

2,678

,068

2,927

,048

2,362

,125

,034

,854

,156

,693

AHORRO50 Within Groups

Final Master Thesis

Guillermo Velรกzquez

Total

22852,740

336

The results reveal equal variances assumption is rejected for 20% time savings with F(1, 345))= 2,927, p-value of 0,048. For 10% savings F(1, 346)= 2,678, p-value = 0,068 which is above the standard 0,05 limit, but yet quite close meaning that there is some signinficant disparity in the means of the different groups. The assumption is met for 20%, 40% and 50% levels with F(1,343)= 2,362, F(1,338)= ,034 and F(1,336)=0,156 and p-values of 0,125, 0,854 and 0,693 respectively.Differences in the sample size for each option are due to respondents selecting “I don’t know” in some of the items. The results revealed statistically significant differences among both groups for the lower tier of the time savings, which implies that in the studied sample Smartphone users were less willing to pay than non-users for small percentage savings of time thanks to the use of the app. Differences in willingness to pay among the two groups for the higher end of time saving were found to be not statistically significant. Regarding the modality of payment, results by each variable group are shown on Table 30 with percentages for each option calculated over the total group population in order to ease comparisons.

Table 28. Preferred payment option of the user, split by Smartphone use. (% of row).

SMARTPHO NE USER Yes No

Not Only for Payment on willing to extra download pay services 70 14 71 16

Final Master Thesis

Yearly Daily Monthly Weekly subscriptio subscri subscription subscription n ption 9 2 4 0 0 6 3 3 1 2

Guillermo Velázquez

6. Conclusions App capabilities Technological penetration rates are high among Madrid’s public transport users, with ~78% of users being also users of personal computers and up to 86% stating a good level of proficiency in using them, and ~84% of users being users of smartphones with the same rate stating a good or very good proficiency level on their use. On tool capabilities and design, the following conclusions can be drawn: - Having a user friendly interface is essential as well as granting that the perception of the user is not for the app being too complex, or many of the potential users will not go through the necessary learning process. - Inclusion of multiple options of filtering and searching for information, as users differ in preferences. - Inclusion of multiple options of visualization of results, and for guidance information as users differ in preferences. - Offering personalization features in order to improve the efficiency in its use and the overall user experience. - Offering payment and subscription-related in-app services, as public transport card office information or mobile payment options. - Ensuring that the services offered cover the most desired ones in each of the fields (static information, real time information and special events information). - Offering the app for free, as most of the users are not willing to pay for the service. Most users are interested in testing the app (77% interested or very interested), and complexity in the daily use of the application or a not user-friendly interface are seen as the main barriers for its adoption.

Willingness to pay As a first conclusion of this part of the study it has been shown that there is a certain correlation between the user’s expected time savings thanks to the use of a mobile app for public transportation and the willingness to pay of the users for it. However, as seen in the literature other factors that the app can improve apart from time saving can also generate utility for the user. With regard to population segmentations the analysis has shown that gender and smartphone use have some degree of relevance in the willingness to pay of the users. With women and non Smartphone users showing a higher predisposition to pay, that is, they expect higher utility from such a service than men or Smartphone users. As expected multimodality is directly correlated with the users’ willingness to pay. When maintaining other variables constant the average amount that users are willing to pay for the app, in all the different timesaving options is greater for multimodal users, which has been proven statistically relevant. This means multimodal users find the greatest utility in this kind of applications.

Final Master Thesis

Guillermo Velázquez

The results show the same trend for private car drivers, although the smaller representation of this segment in the sample calls for caution when interpreting the validity of this result. Segmentations by income level and average length of the most frequent trip had no statistical significance regarding different willingness to pay of the members of different groups. The results of the research show that having to pay for the app is a great barrier, as 70% of the users are not willing to pay for the service, 13% would be willing to pay only at download time, 8% would only pay for premium services and only 5% would pay on a monthly basis.

Future steps: As results show that there is a correlation between time saved for the user and willingness to pay. Hence a model could be inferred that predicts willingness to pay of the population. The work should take into account the influence of factors resulting from the Anova analysis conducted (Sex, smartphone use, age, mode used in the most frequent trip..). Also, an analysis through principal components methodology of the variables used for the user preferences section could be useful in the future in order to reduce the size of the survey, that can affect the quality of the retrieved data.

Final Master Thesis

Guillermo Velรกzquez

7. Aknowledgements This work stems from research carried out for OPTICITIES project addressed towards enhancing smart mobility in Europe and funded by the European Commission under R&D EU’s 7th Framework Program. The authors also acknowledge the collaboration of Madrid’s transport authority (CRTM).

Final Master Thesis

Guillermo Velázquez

8. References Ajzen I., (1991). The theory of planned Behaviour. Organizational Behavior and Human Decision Processes. Babbie R., (2010). The Practice of Social Research. Belmont: Wadsworth, Cengage Learning. Banister, D. (2008). The sustainable mobility paradigm. Transport Policy 15 (2008) 73–80 Börjesson, M., Fosgerau, M., Algers, S., (2008). The income elasticity of the value of travel time is not one number, Proceedings of the European Transport. Conference, Leiden, Netherlands. Chen, Q., Gelman A., Tracy M., Norris F., Galea S., (2012). Weighting Adjustments for Panel Nonresponse. University of Columbia. Chorus, C.G., Arentze, T.A., Timmermans J.P., (2007) Travelers' need for information in traffic and transit: results from a web survey. J. of Intelligent transport systems, 11 (2) (2007), pp. 57–67 CRTM (2005). Encuesta domiciliaria de movilidad 2004 en la Comunidad de Madrid. Retrieved from: http://prueba.crtm.es/media/157705/edm_2004.pdf. (Accessed May, 10, 2015). Dziekan, K., Kottenhoff, K. (2007). Dynamic at-stop real-time information displays for public transport: effects on customers. Transportation Research part A. ECTRI. (2011). ECTRI position on the ec white paper 2011 (Vol. 32). European Conference of Transport Research Institutes. Brussels. ECTRI. (2013). ECTRI suggestions for the first work programmonth of the Transport Challenge in Horizon 2020. European Conference of Transport Research Institutes. Brussels. EUROPEAN COMMISSION (2007). Green Paper – Towards a new culture for urban mobility. EUROPEAN COMMISSION (2011). WHITE PAPER: Road Map to a Single European Transport Area -Towards a Competitive and Resource Efficient Transport System. Brussels. EUROPEAN COMMISSION (2013). European Innovation Partnership on Smart Cities and Communities. EUROPEAN PARLIAMENT (2014). Mapping Smart Cities in the EU. European Parliament. Directorate General For Internal Policies. Policy Department A: Economic And Scientific Policy. EUROSTAT (2015). Eurostat Yearbook. Felix, C., Gay, V., Golliard, L., Johnston, B., Leijdekkers, P., Vaughan, E., Wang, X., Williams, M., (2013). What Can a Mobile App Do To Encourage Cycling?. Second IEEE International Workshop on Global Trends in Smart Cities 2013 Field, A., (2014). Discovering statistics using SPSS.

Final Master Thesis

Guillermo Velázquez

ITF: Peer, S., Koopmans, C., Verhoef, E., (2009). Predicting Travel Time Variability for Cost-Benefit Analysis. Kenyon, S., Lyons, G., (2003). The value of integrated multimodal traveller information and its potential contribution to modal change. Transportation research part F. Kramers, A., (2014). Designing next generation multimodal traveler information systems to support sustainability-oriented decisions. Environmental Modelling & Software Volume 56, June 2014, Pages 83–93. Little, RJ., Vartivarian S. (2005). Does Weighing for non response increase the variance of survey means?. Survey Methodology. 31:161-168 Mishan, EJ., Quah, E., (2007). Cost-benefit analysis Moss M., Mandell J., Qing C., (2011). MOBILE Communications and TRANSPORTATION in Metropolitan Regions The Rudin Center for Transportation Policy and Management Neirotti, P. (2012) Current trends in Smart City initiatives: Some stylised facts. Cities 38 (2014) 25– 36. Nelsen, R.B., (2001), "Kendall tau metric", in Hazewinkel, Michiel, Encyclopedia of Mathematics, Springer, ISBN 978-1-55608-010-4. TRIP (2013b). Thematic Research Summary. Urban Transport. Transport Research and Innovation Portal. Toledo, T., Beinhaker, R. (2006). Evaluation of the potential benefits of advanced traveller information system. Journal of of Intelligent Transportation Systems. Santis, R., Fasano, A., Mignolli N., Villa A., Smart City: the future city?. economia&lavoro 1 (2014): 177-193. Skelley, T., Namoun, A., Mehandjiev N., (2013) The Impact of a Mobile Information System on Changing Travel Behaviour and Improving Travel Experience. Van Nes, R., (2002) Design of multimodal transport networks. A hierarchical approach . TRAIL Thesis Series, Delft University Press, The Netherlands Watkins, K. E., Ferris, B., Borning, A., Rutherford, G. S., & Layton, D. (2011). Where Is My Bus? Impact of mobile real-time information on the perceived andactual wait time of transit riders. Transportation Research Part A: Policy and Practice. Zheng Li David A. Hensher John M. Rose (2010) Willingness to pay for travel time reliability in passenger transport: A review and some new empirical evidence. Zografos, K.G., Androutsopoulos K.N., Nelson J.D., (2010) Indentifying Travelers’ Information needs and services for an integrated international real time journey planning system. 13th International IEEE Annual Conference on Intelligent Transportation Systems

Final Master Thesis

Guillermo Velázquez

Final Master Thesis

Guillermo Velรกzquez

ANNEX I – POST HOC ANALYSIS TABLES

Gender No Post-hoc tests were needed for Gender segmentation

Age group

Table 29. Post-hoc Games Howell and Scheffe tests output on SPSS by age group. Multiple Comparisons Games-Howell Dependent

(I)

Variable

GRUPO.ED GRUPO.EDA AD.COD

(J)

D.COD

Std.

Difference

Error

Sig.

95% Confidence Interval Lower Bound

Upper Bound

(I-J)

Young

,386

,590

,791

-1,01

1,78

Senior

1,210*

,425

,013

,21

2,21

Adult

-,386

,590

,791

-1,78

1,01

Senior

,824

,529

,267

-,43

2,08

,425

,013

-2,21

-,21

Young

-,824

,529

,267

-2,08

,43

Young

,278

,728

,923

-1,44

2,00

Senior

1,308

,596

,074

-,10

2,71

Adult

-,278

,728

,923

-2,00

1,44

Senior

1,030

,720

,328

-,67

2,73

Adult

-1,308

,596

,074

-2,71

,10

Young

-1,030

,720

,328

-2,73

,67

Young

,278

,832

,940

-1,69

2,24

Senior

,990

,766

,401

-,82

2,80

Adult

-,278

,832

,940

-2,24

1,69

Senior

,712

,899

,708

-1,41

2,84

Adult

-,990

,766

,401

-2,80

,82

Young

-,712

,899

,708

-2,84

1,41

Young

,428

,974

,899

-1,87

2,73

Senior

,837

,028

,19

4,14

Adult

-,428

,974

,899

-2,73

1,87

Senior

1,739

,993

,190

-,61

4,09

Adult

-10% time

M ean

Young Adult

-1,210

Senior

Adult

-20% time

Young

Senior

Adult

-30% time

Young

Senior

Adult

2,167

-40% time Young

Final Master Thesis

Guillermo Velázquez

Adult

-2,167*

,837

,028

-4,14

-,19

Young

-1,739

,993

,190

-4,09

,61

Young

-,942

1,156

,694

-3,68

1,79

Senior

2,270

,997

,061

-,08

4,62

Adult

,942

1,156

,694

-1,79

3,68

Senior

3,212

1,227

,026

,31

6,12

Adult

-2,270

,997

,061

-4,62

,08

1,227

,026

-6,12

-,31

Senior

Adult

-50% time

Young

Senior Young

-3,212

*. The mean difference is significant at the 0.05 level.

Mode for most frequent trip

Table 30. Post-hoc Games Howell and Scheffe tests output on SPSS by mode for most frequent trip. Multiple Comparisons Dependent Variable

(I) M ODOcod

(J) M ODOcod

M ean

Std.

Difference

Error

Sig.

Interval

(I-J)

Only car (driver) Only bus

Lower

Upper

Bound

1,458

1,639

,851

-3,15

6,06

Only subway

,941

,834

,735

-1,40

3,28

M ultimodal

,472

,712

,932

-1,53

2,47

-1,458

1,639

,851

-6,06

3,15

Only subway

-,517

1,591

,991

-4,99

3,96

M ultimodal

-,986

1,531

,937

-5,29

3,32

Only bus

-,941

,834

,735

-3,28

1,40

,517

1,591

,991

-3,96

4,99

M ultimodal

-,470

,595

,891

-2,14

1,20

Only bus

-,472

,712

,932

-2,47

1,53

Only car (driver)

,986

1,531

,937

-3,32

5,29

Only subway

,470

,595

,891

-1,20

2,14

1,458

1,033

,504

-1,38

4,30

Only subway

,941

,890

,716

-1,40

3,28

M ultimodal

,472

,804

,936

-1,66

2,60

-1,458

1,033

,504

-4,30

1,38

Only subway

-,517

,857

,930

-3,01

1,97

M ultimodal

-,986

,767

,593

-3,37

1,40

Only bus

-,941

,890

,716

-3,28

1,40

,517

,857

,930

-1,97

3,01

Only bus Only car (driver)

95% Confidence

Scheffe Only subway

Only car (driver)

AHORRO10 M ultimodal

Only car (driver) Only bus

Only bus

GamesHowell

Only car (driver)

Only subway Only car (driver)

Final Master Thesis

Guillermo Velรกzquez

M ultimodal

-,470

,560

,836

-1,93

,99

Only bus

-,472

,804

,936

-2,60

1,66

Only car (driver)

,986

,767

,593

-1,40

3,37

Only subway

,470

,560

,836

-,99

1,93

Only car (driver)

,250

2,046

1,000

-5,50

6,00

Only subway

1,011

1,045

,817

-1,93

3,95

M ultimodal

,060

,900

1,000

-2,47

2,59

-,250

2,046

1,000

-6,00

5,50

Only subway

,761

1,979

,986

-4,80

6,32

M ultimodal

-,190

1,907

1,000

-5,55

5,17

-1,011

1,045

,817

-3,95

1,93

Only car (driver)

-,761

1,979

,986

-6,32

4,80

M ultimodal

-,951

,736

,644

-3,02

1,12

Only bus

-,060

,900

1,000

-2,59

2,47

Only car (driver)

,190

1,907

1,000

-5,17

5,55

Only subway

,951

,736

,644

-1,12

3,02

Only car (driver)

,250

1,718

,999

-4,75

5,25

Only subway

1,011

1,082

,787

-1,85

3,87

M ultimodal

,060

1,028

1,000

-2,67

2,79

-,250

1,718

,999

-5,25

4,75

Only subway

,761

1,511

,956

-4,00

5,52

Games-

M ultimodal

-,190

1,473

,999

-4,94

4,56

Howell

Only bus

-1,011

1,082

,787

-3,87

1,85

Only car (driver)

-,761

1,511

,956

-5,52

4,00

M ultimodal

-,951

,624

,426

-2,57

,67

Only bus

-,060

1,028

1,000

-2,79

2,67

Only car (driver)

,190

1,473

,999

-4,56

4,94

Only subway

,951

,624

,426

-,67

2,57

-2,667

2,349

,732

-9,27

3,94

,666

1,204

,959

-2,72

4,05

-2,009

1,034

,288

-4,91

,90

Only bus

2,667

2,349

,732

-3,94

9,27

Only subway

3,333

2,275

,543

-3,06

9,73

M ultimodal

,658

2,190

,993

-5,50

6,81

-,666

1,204

,959

-4,05

2,72

-3,333

2,275

,543

-9,73

3,06

,852

,021

-5,07

-,28

Only bus

2,009

1,034

,288

-,90

4,91

Only car (driver)

-,658

2,190

,993

-6,81

5,50

Only subway

2,675

,852

,021

,28

5,07

Only car (driver)

-2,667

2,303

,665

-9,79

4,46

,666

1,063

,923

-2,13

3,46

-2,009

1,000

,196

-4,65

,63

M ultimodal

Only bus

Only bus Only car (driver) Scheffe

Only bus Only subway

M ultimodal AHORRO20 Only bus

Only bus Only car (driver)

Only subway

M ultimodal

Only car (driver) Only bus

Only subway M ultimodal

Only car (driver) Scheffe AHORRO30

Only bus Only subway

Only car (driver) M ultimodal

M ultimodal

-2,675

GamesHowell

Only bus

Only subway M ultimodal

Final Master Thesis

Guillermo Velรกzquez

Only bus

2,667

2,303

,665

-4,46

9,79

Only subway

3,333

2,197

,472

-3,74

10,40

M ultimodal

,658

2,167

,989

-6,41

7,72

-,666

1,063

,923

-3,46

2,13

-3,333

2,197

,472

-10,40

3,74

,724

,002

-4,56

-,79

Only bus

2,009

1,000

,196

-,63

4,65

Only car (driver)

-,658

2,167

,989

-7,72

6,41

Only subway

2,675

,724

,002

,79

4,56

Only car (driver)

-5,440

2,706

,259

-13,05

2,16

,515

1,395

,987

-3,40

4,44

-2,642

1,206

,189

-6,03

,75

Only bus

5,440

2,706

,259

-2,16

13,05

Only subway

5,956

2,612

,160

-1,38

13,30

M ultimodal

2,798

2,516

,744

-4,27

9,87

Only bus

-,515

1,395

,987

-4,44

3,40

-5,956

2,612

,160

-13,30

1,38

,977

,016

-5,90

-,41

2,642

1,206

,189

-,75

6,03

-2,798

2,516

,744

-9,87

4,27

Only subway

3,158

,977

,016

,41

5,90

Only car (driver)

-5,440

3,181

,376

-15,53

4,65

,515

1,185

,972

-2,60

3,63

-2,642

1,114

,094

-5,59

,30

Only bus

5,440

3,181

,376

-4,65

15,53

Only subway

5,956

3,094

,295

-4,11

16,03

Games-

M ultimodal

2,798

3,068

,799

-7,27

12,87

Howell

Only bus

-,515

1,185

,972

-3,63

2,60

-5,956

3,094

,295

-16,03

4,11

,834

,001

-5,33

-,99

2,642

1,114

,094

-,30

5,59

-2,798

3,068

,799

-12,87

7,27

Only subway

3,158

,834

,001

,99

5,33

Only car (driver)

-7,691

3,127

,112

-16,48

1,10

Only subway

1,283

1,604

,887

-3,22

5,79

M ultimodal

-2,210

1,388

,470

-6,11

1,69

7,691

3,127

,112

-1,10

16,48

Only subway

8,974*

3,021

,033

,48

17,46

M ultimodal

5,482

2,912

,317

-2,70

13,67

-1,283

1,604

,887

-5,79

3,22

Only car (driver)

-8,974

3,021

,033

-17,46

-,48

M ultimodal

-3,493*

1,129

,024

-6,66

-,32

Only car (driver)

Only bus Only subway

Only car (driver) M ultimodal

M ultimodal

Only bus

Only subway M ultimodal

Only car (driver) Scheffe Only subway

Only car (driver) M ultimodal Only bus

M ultimodal AHORRO40 Only bus

Only car (driver)

Only subway M ultimodal

Only car (driver)

Only subway

Only car (driver) M ultimodal Only bus

M ultimodal

Only bus

Only car (driver)

Only bus AHORRO50 Scheffe

Only car (driver)

Only bus Only subway

Final Master Thesis

-2,675

-3,158

Guillermo Velรกzquez

Only bus

2,210

1,388

,470

-1,69

6,11

Only car (driver)

-5,482

2,912

,317

-13,67

2,70

Only subway

3,493*

1,129

,024

,32

6,66

Only car (driver)

-7,691

4,111

,309

-20,74

5,36

Only subway

1,283

1,505

,829

-2,68

5,24

M ultimodal

-2,210

1,396

,397

-5,91

1,49

Only bus

7,691

4,111

,309

-5,36

20,74

Only subway

8,974

3,993

,196

-4,05

22,00

Games-

M ultimodal

5,482

3,953

,544

-7,54

18,51

Howell

Only bus

-1,283

1,505

,829

-5,24

2,68

Only car (driver)

-8,974

3,993

,196

-22,00

4,05

,995

,003

-6,08

-,90

2,210

1,396

,397

-1,49

5,91

Only car (driver)

-5,482

3,953

,544

-18,51

7,54

Only subway

3,493*

,995

,003

,90

6,08

M ultimodal

Only bus

Only car (driver)

Only subway

M ultimodal Only bus M ultimodal

-3,493

*. The mean difference is significant at the 0.05 level.

Income No Post-hoc tests were needed for income segmentation

Duration of most frequent trip No Post-hoc tests were needed for income segmentation

Smartphone Use No Post-hoc tests were needed for Gender segmentation

Final Master Thesis

Guillermo Velรกzquez