- Email: [email protected]

ScienceDirect Transportation Research Procedia 18 (2016) 180 – 188

XII Conference on Transport Engineering, CIT 2016, 7-9 June 2016, Valencia, Spain

Analysis of multistage chains in public transport: The case of Quito, Ecuador "Efrain Bastidas-Zelaya a, Tomas Ruiz b" a, b

Universitat Politécnica de Valéncia, Escuela Técnica Superior de Caminos, Canales y Puertos, Camino de Vera s/n, Valencia 46022, Spain

Abstract Because of the growth of cities in size and population, people get used to perform several stage trips involving transfers due to advantages such as time or price paid, being multistage trips more attractive compared to single stage trips. In Quito, multistage trips represent one third of total daily trips. This paper seeks to identify main characteristics of multistage trips as well as find relationships and inferences that allow recommendations regarding best practices to policy makers and transport managers. The information used belong to the data collected in the Household Survey Mobility held in Quito in 2011. Based on these data, the present work starts using an analysis with descriptive statistics. The next phase of this research involves the search for a methodology in order to identify correlations between demographic, socioeconomic and transport variables related with traveler´s choice for making or not a transfer. Best methodology found was the use of Binary Logistic Regression (Logit) and specific computer software, with which different statistic's models were performed to find the strongest correlation. The paper ends with conclusions and recommendations as well as suggestions for future research. © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of CIT 2016 Keywords: Multistage chains; bynary logistic regression; Quito public transport

1. Introduction In recent decades there has been a phenomenon of increasing motorization in many countries and regions (Matas 2004, Vassallo 2012 and Beirao 2007). As a result of these situation, the evidence shows that possession of private vehicles soared in several regions of the world, thus in the European Union between 1970 to 2000 private vehicles passed from 62.5 million to 175 million (Ulengin 2006) and associated with the ownership of private vehicles is given a growth of the number of trips and the length thereof , as for example in the Metropolitan Area of Madrid from 1996 to 2004 the number of mechanized trips increased 52% compared to populations growth of just only 14% (Vassallo * Corresponding author. E-mail address: [email protected]

2352-1465 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of CIT 2016 doi:10.1016/j.trpro.2016.12.025

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

2012) or in Great Britain that, over a period of 25 years, the total distance traveled in private vehicles increased by 45% (Ibrahim 2003). The same pattern has occurred in Latin America with expansion of private vehicle fleet in all countries of the region with growth rates ranging from 24% to 277% in the period from 2000 to 2007 (Jiron 2013). As effects of growth on use of private vehicles, congestion increased with negative effects to environment, so many studies and analysis seek to promote public policies to encourage use of public transport and reduce car dependence (Ibrahim 2003, Matas 2004, Ulengin et al. 2006, Beirao 2007, Vassallo 2012, Chowdhury 2012 and Chowdhury 2015). Measures to promote public transport include increasing transport supply through investments in transport infrastructure (Vassallo 2012) and providing a wide range of transport modes in integrated public transport systems (Ibrahim 2003), which are reflected in significant rates of transfers as in the cases of London (70% of trips in the Underground, 30% on buses), New York City (30% travel Subway , 80% in commuter trains), Munich and Paris with transfers at 70% and 40% of all public transport trips, respectively (Guo and Wilson 2011). There are many benefits of integrated systems and transfers, starting from the vision of moving into a network rather than a simple route or line (Clever 1997), potential users gain due to these interconnected routes (Bak et al. 2012), more destination options along with reduced trip times and fares (Bak et al. 2012). But there are differences between the networks planned from design versus those unplanned, and users clearly distinguish it (Chowdhury and assign 2013b), whereby cities like Madrid have focused on correcting failures of planning and since 1986 Madrid transport infrastructures have been planned as a complete system (Matas 2004). However the advantages previously indicated, transfers generated discomfort to users such as interrupt the travel experience, and thus be less competitive than the car that goes from door to door (Guo and Wilson 2011) or the perception that transfers mean a time penalty (Navarrete and Ortuzar 2012). (Navarrete and Ortuzar 2012). Therefore it is necessary to analyze characteristics and attributes of trips and transfers, having been found in literature (Chowdhury and Ceder 2013b) that principal ones are trip time, trip cost, walking time and wait time in order to make a transfer, trip information provided , integrated tariffs, traveler safety, security and comfort in terminal. Beyond these quantitative variables of trip characterization, other studies propose to consider qualitative variables related to the conditions in which the trip is performed such as the existence of intermodal stations, availability of escalators, availability of prior trip information (Navarrete and Ortuzar 2012), conditions of walk transfer inside transport network, transfer waiting related to operation and service, and transfer penalties comprising the ease of finding the path during transfers, provision of traveler safe and security, weather's coverage air conditioning, and seat availability (Guo and Wilson 2011). Finally, there are also demographic and socio-economic variables influencing travelers' decision to make transfers, such as age, income, gender, marital status and household structure (Currie 2011). Not all factors have the same weight in traveler´s decision to make transfers, as some variables are cited as preponderant: trip time, personal safety and transfer time (Vande Walle and Steenberghen 2006), coinciding other authors that safe and security in terminals are preponderant and adding integrated fare, appropriate ticketing system , more comfort in transfer terminals (Chowdhury and assign 2013b). It should be noted certain factors are cited as totally adverse towards transfers, like car availability has strong negative elasticity in relation to trips with transfers (Vande Walle and Steenberghen 2006). Although the literature shows statistics on these variables for some Latin American countries such as Chile, there is no information regarding Quito. The main objective of this article is to present an analysis that allows to understand travel process, as well as obtain a characterization of the multistage chains, to subsequently determine the trends that lead the traveler to decide to perform a multistage trip. The correlation between transfer decision and transport variables will be studied, as well as correlation with demographic and socioeconomic variables. 2. Problem statement and data available Quito is a city with complex conditions for roads and mobility. It is flanked both east and west by mountain ranges that are part of the Andes, and his historic center with narrow streets and ancient churches and buildings has high heritage value, so received by UNESCO the status as World´s Cultural Heritage Site. According to the Transportation Authority of Quito (Secretaría de Movilidad), the city generates 2.9 million daily trips in public transport (1 million trips in Metrobus-Q system managed by the municipality) It is considered that

181

182

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

Quito provides universal coverage of public transport, reaching 95% coverage of bus in the urban area of Quito (EPMMQ 2012). BRT systems have a great importance for the city. There are four BRT lines: Ecovía, Central North Corridor, South East Corridor and South Western Corridor with their corresponding network of feeder lines, in order to form a backbone of 83 km of trunk lines in segregated corridors and 120 weather protected stations or large terminals for passengers transfers (EPMMQ 2012). Nowadays Quito is constructing the first Metro line (PLMQ). In private transport there are 1.5 million daily trips (including cars and taxis) and the rate of motorization rose from 145 vehicles per 1,000 people in 2002, (EPMMQ 2012), to 240 vehicles per 1,000 inhabitants today. The data used in this research comes from the planning done by the Transportation Authority of Quito, intended for PLMQ designs. These data were collected between the years of 2011 to 2012. Given the need to ensure the technical and economic feasibility for the project, the city developed a series of studies to find potential demand that would have the project and thereby calculate the train fleet, the size of stations, necessary equipment, and financial flows of the operation of PLMQ. These activities were carried out by the area of international projects of the Madrid Metro company (Metro de Madrid), due to an agreement signed between the local governments of Quito and Madrid. For the data collection, Madrid Metro organized a package of studies, being the main element an Mobility Household Survey (EDM11 for its acronym in Spanish), which was complemented with field studies in order to obtain mobility data in public transport, as well in private vehicles, use of bicycle and pedestrian flows. Geographic scope of EDM11 comprise Metropolitan District of Quito and urban areas of Mejia and Rumiñahui municipalities, which together form the Conurbation of Quito. Madrid experts established the universe survey as population aged 4 years and older, a sampling of random type, by conglomerate of housing, stratified by sex and age within each administrative district (Metro de Madrid 2012). Sampling frame was taken from the data of the Empresa Electrica Quito, the city's power supplier, resulting in a sample size of 75323 people, a sampling error for the estimated trips number of less than 0,5% for the entire scope of the study and 7.5% for the middle zone (confidence level of 95%). There were defined by Metro de Madrid 240 administrative zones of transport (ZAT). According to previous Madrid’s technical team experiences, EDM11 involved a mix of methodologies. So, the first stage of EDM11 was developed with direct interview on households for data collection and requesting to each member their mobile phone numbers, and a second part was performed with phone calls from a contact center, assisted by a computer that allowed data validation. After completing the survey the number of valid registers increased slightly to 77056 people corresponding to 28573 households, with an average of 2.70 people per household In this research and analysis we took the decision not to expand the database from the sample size to the size of the universe, because by doing so they would be using multiplier coefficients that would alter the raw data obtained from the survey. Therefore we will work exclusively with the values collected in the survey. 3. Methodology The work in this article has two sections. First, an analysis of descriptive statistics to find patterns in the mobility and second by an analysis of traveler's decision about whether or not to make a transfer. It is necessary to use a statistical regression function, and computer analysis of the databases is done with IBM-SPSS software package User decision of making transfer or not, is worked as a binary function where 1 represents the fact to make transfer and 0 when there is no transfer. According to Espino (2007) Theory of Random Utility is used to treat empirically discrete choice, since according to the mentioned theory the researcher assumes that the utility of alternative j for the individual q has the expression (Xjq):

Ʋjq = Vjq + Ɛjq

(1)

Vjq is the representative or systematic utility, Ɛjq is the random term in which unobserved by the researcher effects are included. Both variables depend on the attributes of alternative j and socioeconomic characteristics of the individual q. It is important to consider that the distribution of the random term determines the type of econometric model. There are several logistic regression models commonly used by researchers Single Logit, Multinomial Logit, Mixed Logit, Hierarchical Logit, Probit, etc.) But recommendation of Guo and Wilson (2012) it is to use the simplest

183

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

model, because there is no strong evidence that a complex Logit model works better than a single. Simple Logit was selected for this research. Utility function for determining if transfer is performed or not, shall be composed for some variables such as travel time, travel cost, availability of own vehicle and others. Mathematical formulation proceed as follows: ƲTR = β0 + β1t + β2C + β3d + …

்ܲோ ൌ

(2)

ೆೃ

(3)

ଵା ೆೃ

4. Analysis of results 4.1. Characteristics of mobility in the population according EDM11 The point of departure of mobility observed in the EDM11 is the analysis of who traveled and who did not, since not all the population travels every day. According to the survey, considering 77056 citizens surveyed, 16966 people not travelled (22.08% of respondents), while 60090 people reported they made a trip (77,98%). Of those who said they had traveled a total of 144,205 valid trips records was obtained, with an average of 1.87 trips per person. When analyzing in detail the group of people who have not had mobility on the day of the survey it can be characterized using some variables. The first is by gender: 11,316 people, accounting for 66.70% of those who had no mobility were female, compared with only 5,650 men who did not have mobility. Classification by age group shows that the largest group who did not travel was comprising from 25 to 44 years with 4984 people (29.38%), followed by the group of 60 years or more, with people who did not travel 4636 (27.33%), and the range of 45 to 59 years with 3702 people (21.82%). Continuing with the statistical classification, according to the variable level of schooling there were 8047 people (47.4%) that do not exceed primary school, i.e. not even entered high school. By making a crossanalysis of the data it shows that there is a correspondence between the majority groups who did not travel, who are female, defined themselves as housewives, low level of education (without reaching secondary education) and middleaged life. This is a photograph of a precarious social group, with fewer advantages in society, and coincides with the group of lower mobility in the city. Moving on to analysis of those who made trips, the first important classification is the modal share, in which 144,205 trips recorded in the survey, were made as shown in Table 1, showing that public transport is predominant with 53,21 % and if we add to this school buses and minibuses, we would reach a total of 61% of trips made in all types of buses. On the other hand non-motorized modes (walking and cycling) account for 16% of total trips, while private vehicle registered slightly less than 20% of total trips. Table 1: Modal Share in Quito, according EDM11 Transport Mode Public transport Private motor vehicle (cars) Walking School bus and minibus Taxi Cycling Total

Trips 76727 28478 22496 11344 4646 514 144205

Percent 53,21% 19,75% 15,60% 7,87% 3,22% 0,36% 100,00%

As can be seen, public transport is predominant in Quito´s modal share, but travelers often are captives to the bus system because they do not have their own vehicle. This condition would define that exists "captivity to public transport". It is not considered possible to have captivity to other modes of transport (such as private vehicle), due to

184

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

Quito has universal coverage of public transport, since more than 94% of inhabitants are located in a range of 400 meters away from a bus stop (Metro de Madrid, 2012). So users of private vehicles, cyclists or pedestrians have an alternative: the use of public transport (PT). The next analysis is captivity to PT using different variables already mentioned. First variable to be studied is genre showing there are more captivity to PT in women, while men have greater use of private vehicles. Thus, the gender distribution of those who have own vehicle shows that 57% are men, while only 43% are women. That difference is enlarged when asked if traveling as a driver or passenger, as 2/3 of the drivers of private vehicles are men, while women are more commonly passengers of private vehicles, showing a clear gender segregation. The profile of the PT captive, according to the data collected, perfectly portrays high school and university students. 48% of PT captives are under 24 years old, 42% of them specifically states that its main activity is attend classes, 57.5% mentioned that their travel frequency corresponds to working days and 54% PT captives make their trips during peak hours in the morning and noon, coinciding with entry and exit of classes in most institutions of educational system especially in high school. In the peripheral areas of the city (corresponding to lower economic level) predominates captivity PT, while the sector or area of northern Quito (Central Business District) has predominant use of private vehicles. 4.2. Multistage chains in Quito As explained earlier in this document, transfers are becoming increasingly important as city grows and trips become more complex. Of the collected sample (144,205 trips), those involving a transfer ie comprising several stages are 27366 trips (19%) while single-stage trips are 116839 (81%). The statistical analysis of transfers by using the variables already mentioned, gives some features: Women perform 12% more transfers than men. As can be seen in fig1, the modal split of multistage trips is absorbed almost in its totality by public transport, as other modes do not even reach 1%; meanwhile in single stage trips public transport captured 43% of the modal split.

Figure 1: Modal share comparison between single stage trips and multistage trips, EDM11

In terms of geographical distribution, the trend is very similar to that of captive travelers, as people in peripheries make higher number of transfers while those living in the north of Quito make more single stage trips (without transfer). Regarding age range, the group with most transfers are those travelers between 25 and 44 years (38%) and those who make less transfers are children 4-14 years old with only 8% of multistage trips. These data are consistent with the analysis regarding educational level, because in elementary school predominate single stage trips, while college and university graduates are those who make more transfers. When analyzing the modal chain, it shows that conventional bus is always the most used, although its modal participation falls from 47% for the first transfer to 36% for the fourth transfer. The Metrobus-Q system is the second most used mode, keeping a share ranging from 26% to 30%. The metropolitan area bus is the third most used mode,

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

with modal share from 14% to 17%. Meanwhile, taxi and informal transport (non-legal) have greater use when are needed several transfers, as users living in peripheral areas require perform several travel stages. 4.3. Time and price paid characteristics in multistage trips The following analysis seeks to find the relationship behavior of multistage trips based on variables. It start with the analysis according to the total travel time (including travel time at each stage, waiting time and walking time) then continues with the total walking time. Travel times were grouped into ranges amplitude half an hour each. The distribution of travel times and walk time is presented in Fig 2. Almost 42 % of multistage trips were included in the range from 1 hour to 1 hour and 30 minutes, but considering the range between 1 to 2 hours that percentage rises to 56%, with another 13% of trips exceeding 2 hours. We can conclude trips take too much time in Quito.

Figure 2: Multistage trips distribution a) according to total travel time b) according to total wait time

As for the amount paid for the trip, when it was only one stage 75% of travelers paid $ 0.25, with two stages 55% of users paid $ 0.50, for those who made 3 stages 51% paid $ 0.50, maintaining the trend for trips with more stages. The most common price paid in multistage trips was $ 0.50 (statistical mode), even though the average value was $ 0.42. When analyzing the Metrobus-Q system it shows that the predominant use was for trunk lines systems such as the Trolebus, Metrovía and Ecovía (57% of the stages in this system). In Metrobus-Q the average travel time was 1 hour and 22 minutes, and the average rate was $ 0.33 4.4. Analysis with Binary Logistic Regression After analysis of descriptive statistics, we proceed to an analysis Binary Logistic Regression (Logit simple). The work involved running several models in SPSS statistical software under the following conditions: x The basis on which we work is that of all travelers who have made a trip the day of the survey, regardless of how they used. A total of 144,205 cases. x Several models were run, experimenting with different transport variables, as well with demographic and socioeconomic variables, then it is tabulated in search model had better results. In every model, the dependent variable is the decision by the traveler to make or not transfer (binary 1.0). Concerning the statistical parameters to evaluate variables and thus models, software SPSS delivers a set of parameters providing information, of which the most notable are the following: x Coefficient of determination Square R Nagelkerke, which is found in the Omnibus Test (Block 1), is an indicator of how well the model test indicates the part of the variance of the dependent variable that is explained by the

185

186

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

model. Its maximum value is 1 and the higher the R2, better is the model explained. Statistical significance in Hosmer Lemeshow test, showing that there is statistical evidence in favor of the model and random results are not given. Ideally it should be less than 0.1 x Forecast percentage of transfer variable in the classification matrix which shows what percentage of the predicted values match the observed values and allows assessment of the fit of the regression model. The higher the percentage value, better model fit. Betas (B) shown in the last frame corresponding to "Equation variables table" these are the coefficients associated to each of the variables being evaluated in the regression model. x Standard error of the betas associated with the variability of calculating betas, expected to be small. x Confidence Interval for exponential of betas, which is generated depending on how large has been the standard error of the betas. It is better the more tight is to the central value. Exponential of betas. It is the estimate of the odds ratio of having the outcome for the variable in question. It represents the number of times the dependent variable increases when a unit of the independent variable does. Exp (B) must be different from 1 and give better answers the more you move away from that value. The model who presented the best results was the one that included the following transport variables: travel time, walking time, total price of ticket and availability of own vehicle (captivity to public transport). The first three mentioned were presented in the form of ranges for better management of data. In addition to those transport variables, tests with different models showed that if included demographic and socio-economic variables the results were also improved. So the optimal model included other demographic variables: geographic areas of origin and destination, age and gender; as well as a socio-economic variable: level of education or schooling. The results that we found for this model are as follows: x In block 0 analysis, it indicates 81% probability of success in the results of the dependent variable x In block 1 of the model, the omnibus test indicates that the model make a good explain of the event, because the significance is less than 0.1. The value of R2 Nagelkerke indicates that the proposed model explained 64.6% of the variance of the dependent variable (0.646), an acceptable value. x According to Hosmer and Lemeshow test, the variance explained by the model explains a significant percentage of the dependent variable (Chi square: 836.751; significance 0.000 <0.05) x In logistic regression analysis, block 1 indicates a 91.7% chance of success in the results of the dependent variable regarding whether a transfer is made or not. There is more success with those who do not make transfer. Finally, the table of “equation variables” is given by the SPSS software, which is reproduced in Table 2, with the following analysis for the variables: Table 2: Equation variables and parameters for best model

Travel time Walk time Price paid Availability of car or motorcycle Gender Age Origin zone Destination zone Education level Constant

B

Exp (B)

Significance

0,695 -0,184 2,412 -0,205 -0,111 -0,301 -0,009 -0,059 -0,083 -7,911

2,004 0,832 11,155 0,815 0,895 0,740 0,991 0,943 0,921 0,000

0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000 0,000

Standard Error 0,009 0,010 0,019 0,029 0,020 0,011 0,002 0,002 0,005 0,074

x In travel time variable, Beta is obtained with a positive sign, greater than zero and the value of Exp (B) is 2.04; which shows that there is a directly proportional relationship and that with each unit increase in travel time the

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

x

x

x x x x

187

number of travelers who would transfer against those who would not, that is the longer is the journey, would double more tendency there to realize it through several stages and transfers. The variable walking time, gives a beta with negative sign and Exp (B) is 0.832. This means that if walking time is increased, the number of travelers who would perform transfer will be reduced. Besides being a completely logical result, it is worth remembering that previously it was mentioned from Chowdhury and Ceder (2013) and Vande Walle and Steenberghen (2006) indicating that walking time is one of the variables that most influences the decision of making or not transfers. The variable availability of car or motorcycle also has a Beta with negative sign and value of Exp (B) is 0.815. That indicates when increasing availability of cars or motorcycles, the amount of trips with transfers will be reduced. Vande Walle and Steenberghen (2006) claimed that availability of own vehicle has a negative influence on the selection of trips with transfers The last transport variable is the price paid, where Beta is positive and the value of Exp (B) is 11,155. There is a high correlation between the prices paid and performing transfers, ie, the higher is the price paid for the trip, there is more likely to do it in several stages. Worth commenting that in the Metrobus-Q system the travel Multistage do not increase their price. The gender variable, with a negative sign and Exp (B) of 0.895 shows that if the number of men traveling increases, the percentage of transfers is reduced. This finding is consistent with the revised descriptive statistics, where it was shown that men perform transfers 0.88 times less than women. The variable age has Beta with negative sign, representing if the age increases, the percentage of transfers is reduced. This result match with the group of senior citizens, who have the less willingness to make transfers. The variables level of education, origin zone and destination zone, have less influence than the other variables described above, however, predict results similar to that found in the statistical characterization of mobility behavior. The significance values in all cases were null, and standard error were very close to zero cases confirm the benefits of the selected model.

5. Conclusions It’s found that Quito has had very few studies and research mobility beyond those engineering or feasibility studies for road transport or implementation plans for projects. Formerly there were not found scientific research article on mobility and transfers in Quito. The EDM11 development has provided valuable information to the city, from which data was extracted for this work. Descriptive statistics of data from EDM11 has allowed find relationships between groups of citizens with some mobility variables. There are difficult conditions of mobility for certain social or economic groups, element to be considered by policy makers and planners of public transport. On the analysis of modal public transport chains it is found that a fifth of city trips are made using transfers. The number of travelers is substantially reduced as it increases the number of transfers and stages. Regarding the variables time and price paid in the multistep trips, it shows that there is a correlation between these variables with the performance of multistage travel. Finally the last section dealt with the Logit analysis and variables correlation. The development of a base model on which is learning and making improvements shown as the ideal way in this work. It is noticed that variables related to transport have a preponderant weight in the decision about whether traveler does or does not transfers. Meanwhile in demographic and socioeconomic variables, the preponderant are gender and age of the traveler. Transport planners should use this information for better ergonomic design in buses, bus travel routes, hiking trails suitable for transfers and transfer stations. When doing this study, we found some constraints because of the data collected, because timeouts or walk through each of the steps weren´t collected. It is proposed for future research considering this and others variables such as reliability and trustworthiness of a bus passing every certain period of time and the effects of delays and finally the effect of use of stations designed from the start to be multimodal transfer centers. References

188

Efrain Bastidas-Zelaya and Tomas Ruiz / Transportation Research Procedia 18 (2016) 180 – 188

Beirao, G. and Cabral, J, 2007. Understanding Attitudes towards Public Transport and Private Car: A Qualitative Study. Transport Policy, 14 (6) Currie G, 2011. Exploring the trip chaining behavior of public transport users in Melbourne. Chowdhury, S., Ceder, A. and Schwalger, B., 2015. The Effects of Travel Time and Cost Savings on Commuters' Decision to Travel on Public Transport Routes Involving Transfers. Journal of Transport Geography, 43 Empresa Pública del Metro de Quito (EPMMQ), 2012, Modelo de movilidad para el DMQ, 2012. Quito Ecuador. Espino, R., Martín, J. C., & Román, C., 2007. Análisis de la calidad del transporte de pasajeros. Un estudio de la disposición a pagar por no trasbordar. X Encuentro de Economía Aplicada, Logroño, Spain. Guo, Z., and Wilson, N. H., 2011. Assessing the cost of transfer inconvenience in public transport systems: A case study of the London Underground. Transportation Research Part A: Policy and Practice, 45(2). Ibrahim, M., 2003. Improvements and Integration of a Public Transport System: The Case of Singapore. Cities, 20 (3) Jirón, P., 2013. Sustainable Urban Mobility in Latin America and the Caribbean, 2013. Matas A, 2004, Demand and revenue implications of an integrated public transport policy. The case of Madrid, Barcelona España. Ulengin, F., Et Al, 2007. An Integrated Transportation Decision Support System for Transportation Policy Decisions: The Case of Turkey. Transportation Research Part A-Policy and Practice, 41 (1). Unidad de Negocios del Metro de Quito (UNMQ), 2010, Estudio de Pre factibilidad Proyecto Metro de Quito, 2010. Quito Ecuador. Vande Walle S, 2006, Space and time related determinants of public transport in trip chains, Transportation Research Part A. 40 (2006) 151–162. Vassallo J, Di Ciommo F, García A, 2012, Intermodal exchange stations in the city of Madrid, Madrid – España.