FACULTY OF SOCIAL SCIENCES
Institute of Economic Studies
Modal Split in Major Czech Cities:
Thorough Analysis and Proposal of Policies Leading to Less Car-Dependent
Urban Mobility
Master’s thesis
Author: Bc. Vojtěch Bystřický
Study program: Economics and Finance Supervisor: Mgr. Milan Ščasný PhD.
Year of defense: 2023
The author hereby declares that he compiled this thesis independently, using only the listed resources and literature, and the thesis has not been used to obtain any other academic title.
The author grants to Charles University permission to reproduce and to dis- tribute copies of this thesis in whole or in part and agrees with the thesis being used for study and scientific purposes.
Prague, December 29, 2022
Vojtech Bystricky
This thesis focuses on examining the modal split in the five largest Czech cities. Using data from the first nationwide survey on travel behavior in the Czech Republic called Česko v pohybu, the author identifies the main factors which influence the mode choice of inhabitants of Czech cities. The data were evaluated using multinomial logistic regression. Since modal split studies of a large extent are mostly conducted in Western Europe, the United States or Asia-Pacific region, the main contribution of this thesis is to shed some light also on the travel behavior in the Central Europe, more precisely in the largest cities of the Czech Republic. The author analyzes the impact of socio-demographic variables, such as the respondents’ age, education level or household income, as well as the importance of the variables related to the trip, such as trip purpose or trip distance. Further, the author also provides comparison of the travel behavior between the examined cities. Among other findings, the author finds that the entitlement to discounted public transport coupons through the ownership of a discount card does not have a significant effect on the probability of using public transport. Further, the results also show that higher education level does not lead to greater use of ecologically friendly transport modes, as the author has originally hypothesized. Lastly, based on the results, the author proposes several policies which aim to motivate the residents of Czech cities towards higher use of ecological forms of transport, such as using public transport, walking or cycling.
JEL Classification R40, R41, R42, Q53, Q56, O18
Keywords modal split, multinomial logit, urban traffic, CO2 emissions, environment
Title Modal Split in Major Czech Cities: Thorough Analysis and Proposal of Policies Leading to Less Car-Dependent Urban Mobility
Tato práce se zaměřuje na analýzu volby dopravního prostředku v pěti ne- jvětších českých městech. V této práci byla použita data z prvního celostát- ního průzkumu dopravního chování Česko v pohybu. Za pomocí multinomiální logistické regrese autor identifikuje hlavní faktory ovlivňující volbu dopravního prostředku u obyvatel českých měst. Jelikož jsou tzv. modal split studie (studie zabývající se volbou dopravního prostředku) většího rozsahu prováděny přede- vším na území západní Evropy, Spojených států amerických nebo v oblasti Asijsko-pacifického regionu, autor považuje za hlavní přínos práce rozšíření znalostí této tématiky rovněž na území střední Evropy, konkrétně největších českých měst. Autor analyzuje nejen vliv socio-demografických proměnných jako např. věku respondenta, nejvyššího dosaženého vzdělání nebo příjmu domácnosti respondenta, ale také tím, jaký vliv na volbu dopravního prostředku má např. účel cesty nebo její délka. Autor dále poskytuje porovnání do- pravního chování v jednotlivých městech. Mimo jiné, výsledky této práce ukázaly, že nárok na zlevněné dlouhodobé kupóny na MHD díky vlastnictví slevové kartičky nemá signifikantní vliv na pravděpodobnost používání MHD.
Dále výsledky nepotvrdily autorovu původní hypotézu, že vyšší úroveň vzdělání vede k vyšší tendenci používat ekologičtější dopravní prostředky. Závěrem au- tor na základě výsledků práce navrhuje několik politik, jejichž cílem je motivo- vat obyvatele českých měst k vyššímu využívání ekologických forem dopravy, jako je používání veřejné dopravy, chůze nebo jízda na kole.
Klasifikace JEL R40, R41, R42, Q53, Q56, O18
Klíčová slova volba dopravního prostředku, multi- nomiální logistická regrese, doprava ve městech, emise CO2, životní prostředí Název práce Volba dopravního prostředku ve velkých
českých městech: detailní analýza a návrh politik vedoucích k ekologičtější dopravě ve městech
The author would like to express his gratitude to his supervisor Mgr. Milan Ščasný PhD. for his continuous support, time and valuable advice. The author believes that his comments on the thesis have greatly improved its quality. The author is further extremely grateful to his family, his girlfriend and close friends for their tremendous moral support during the time of writing the thesis. This thesis is part of a project that has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska- Curie grant agreement No. 870245.
Typeset in LATEXusing the IES Thesis Template.
Bibliographic Record
Bystricky, Vojtech: Modal Split in Major Czech Cities: Thorough Analysis and Proposal of Policies Leading to Less Car-Dependent Urban Mobility. Master’s thesis. Charles University, Faculty of Social Sciences, Institute of Economic Studies, Prague. 2023, pages 100. Advisor: Mgr. Milan Ščasný PhD.
List of Tables viii
List of Figures x
Acronyms xi
Thesis Proposal xii
1 Introduction 1
2 Literature review 4
2.1 Factors influencing transport mode choice . . . 4
2.1.1 Socio-demographic variables . . . 4
2.1.2 Mode-related variables . . . 7
2.1.3 Geographical variables . . . 8
2.1.4 Trip-related variables . . . 9
2.1.5 Weather-related variables . . . 10
2.2 Data collection methods in travel surveys . . . 10
2.2.1 Revealed preference method . . . 10
2.2.2 Stated preference method . . . 12
2.2.3 Combining revealed and stated preference methods . . . 14
2.2.4 Method used in our dataset . . . 15
3 Data overview 16 3.1 Dataset used for the analysis . . . 16
3.1.1 Choice of variables for the model . . . 17
3.2 Descriptive statistics . . . 20
3.2.1 Mode choice by cities . . . 20
3.2.2 Mode choice by trip purpose . . . 22
3.2.3 Mode choice by household income . . . 23
3.2.4 Mode choice by education . . . 24
3.2.5 Mode choice by PT subscription and PT discount . . . . 25
3.2.6 Comparison with Western European cities . . . 27
4 Methodology 30 4.1 Discrete choice models in the modal split field . . . 30
4.2 Specification of the model . . . 32
5 Results 35 5.1 Results of the model . . . 35
5.1.1 Comparison of the cities . . . 35
5.1.2 Effect of gender . . . 38
5.1.3 Effect of economical activity . . . 39
5.1.4 Effect of trip distance . . . 40
5.1.5 Effect of PT discount card . . . 40
5.1.6 Effect of public transport subscription . . . 42
5.1.7 Children under 5 years of age in household . . . 43
5.1.8 Effect of bicycle and car ownership . . . 44
5.1.9 Effect of age . . . 45
5.1.10 Effect of education . . . 47
5.1.11 Effect of season of the year . . . 50
5.1.12 Effect of household income . . . 51
5.1.13 Effect of trip purpose . . . 54
6 Discussion 58 6.1 Affordability of public transport for the lowest income group . . 58
6.2 Air-conditioning in public transport vehicles . . . 59
6.3 Low-floor public transport vehicles . . . 60
6.4 Motivating citizens to commute by cycling . . . 62
6.5 Low use of active modes among the youngest age group . . . 62
6.6 Ownership of PT discount card not significant . . . 63 6.7 Higher education does not lead to higher use of ecological modes 63
7 Conclusion 64
Bibliography 70
A Additional figures and tables I
3.1 Variables used in the model . . . 18
4.1 Summary of methods used in previous studies . . . 31
5.1 MNL results - detail of results for variable City . . . 36
5.2 Predicted probabilities for variable City . . . 37
5.3 Marginal effects for variable City . . . 38
5.4 MNL results - detail of results for variable Gender . . . 38
5.5 Marginal effects for variable Gender . . . 39
5.6 MNL results - detail of results for Economical activity . . . 39
5.7 Marginal effects for Economical activity . . . 40
5.8 MNL results - detail of results for Trip distance . . . 40
5.9 Marginal effects for Trip distance . . . 40
5.10 MNL results - detail of results for PT discount card . . . 41
5.11 Predicted probabilities for PT discount card . . . 41
5.12 Marginal effects for PT discount card . . . 41
5.13 MNL results - detail of results for PT subscription . . . 42
5.14 Predicted probabilities for PT subscription . . . 42
5.15 Marginal effects for PT subscription . . . 43
5.16 Marginal effects for car ownership . . . 44
5.17 Marginal effects for bicycle ownership . . . 45
5.18 MNL results - detail of results for variable Age . . . 46
5.19 Predicted probabilities for variable Age . . . 46
5.20 Marginal effects for variable Age . . . 47
5.21 MNL results - detail of results for Education . . . 48
5.22 Predicted probabilities for Education . . . 48
5.23 Marginal effects for Education . . . 49
5.24 Predicted probabilities for variable Season . . . 51
5.25 Marginal effects for variable Season . . . 51
5.26 MNL results - detail of results for Household income . . . 52
5.27 Predicted probabilities for Household income . . . 53
5.28 Marginal effects for Household income . . . 53
5.29 MNL results - detail of results for Trip purpose . . . 56
5.30 Predicted probabilities for Trip purpose . . . 56
5.31 Marginal effects for Trip purpose . . . 57
6.1 Percentage of low-floor PT vehicles in Czech cities . . . 61 A.1 Summary statistics for Mainmode . . . III A.2 Summary statistics for City . . . IV A.3 Summary statistics for Education . . . IV A.4 Summary statistics for Season . . . IV A.5 Summary statistics for Household income . . . V A.6 Summary statistics for Trip purpose . . . V A.7 Summary statistics for Age . . . V A.8 Summary statistics for non-categorical variables . . . VI A.9 Full results of the MNL model . . . VII A.10 Postestimation - Predicted probabilities . . . IX A.11 Postestimation - Marginal effects . . . XII
2.1 Example of a pen-and-paper travel diary . . . 12
2.2 Example of a stated preference scenario . . . 13
3.1 Mode choice by cities . . . 21
3.2 Mode choice by trip purpose . . . 22
3.3 Mode choice by household income . . . 23
3.4 Mode choice by education . . . 25
3.5 Mode choice by public transport subscription . . . 26
3.6 Mode choice by PT discount card ownership . . . 27
3.7 Work trips, Praha vs. Western cities of similar pop. size . . . . 28 A.1 Work trips, Brno vs. Western cities of similar pop. size . . . I A.2 Work trips, Ostrava vs. Western cities of similar pop. size . . . II A.3 Work trips, Plzen vs. Western cities of similar pop. size . . . II A.4 Work trips, Liberec vs. Western cities of similar pop. size . . . . III
A/C Air conditioning
BL Binary logit
CDV Centrum dopravního výzkumu
CZK Czech Koruna
CVP Česko v pohybu
DPMB Dopravní podnik města Brna, a.s.
DPMLJ Dopravní podnik měst Liberce a Jablonce nad Nisou, a.s.
DPO Dopravní podnik Ostrava, a.s.
DPP Dopravní podnik hl. m. Prahy, a.s.
MNL Multinomial logit
MNP Multinomial probit
PID Pražská integrovaná doprava
PMDP Plzeňské městské dopravní podniky, a.s.
pp Percentage points
PT Public transport
RP Revealed preference
SP Stated preference
Author Bc. Vojtěch Bystřický Supervisor Mgr. Milan Ščasný PhD.
Proposed topic Modal Split in Major Czech Cities: Thorough Analysis and Proposal of Policies Leading to Less Car-Dependent Urban Mobility
Motivation During the last decades, a high emphasis has been put on reducing the emission levels of CO2, air quality pollutants and noise in the cities, which are caused mainly by the high number of cars in the urban traffic. Therefore, governments have been trying to come up with policies which should lead to more environmentally- friendly, sustainable and healthier urban mobility. More precisely, these policies should aim to promote “greener” and healthier modes of transport, such as public transport, cycling or walking.
Numerous studies have been dealing with the topic of modal split (ratios of travelling people using certain mode of transport) and factors which are influencing the travelers’ transport mode choices. Such study can be conducted by using either revealed preference approach (i.e., studying the actual decisions of commuters) or stated preference approach (i.e., studying the decisions of people in a hypothetical contingent scenario). Both of these methods then reveal which factors have impact on the real or hypothetical choices of transport mode, which can be used to implement efficient policies, leading to environmental and/or health benefits.
Currently, the studies on modal split are conducted mainly in Western Europe (such as the work of Ton et al. (2019) or Scheiner (2010)), in USA (Kim and Ulfars- son (2008), Gehrke and Clifton (2014)) or in Asia-Pacific area (see Jin et al. (2020)).
However, to my knowledge, not many detailed studies on this topic have been con- ducted in the region of Central or Eastern Europe. This study aims to analyze the travel decisions of inhabitants of major Czech cities (Praha, Brno, Ostrava, etc.) using both revealed preference and stated preference approach. The models used in the study will be using data from a fresh dataset, which will be obtained by a survey conducted in the major Czech cities in the mid of 2022 (within a research project
led by my supervisor, Dr. Scasny). The results of this study will shed a new light on the travel behavior of the Czech commuters and should help with designing the policies leading to greener environment not only in the respective cities.
Hypotheses
Hypothesis #1: One of the strongest motivators for the modal shift towards public transport is increasing the density of public transport network (provid- ing public transport stations closer to people’s starting and final destinations and therefore reducing access and egress time), rather than reducing the price of public transport.
Hypothesis #2: Participants with greater education levels tend to choose more ecological modes of transport.
Hypothesis #3: Increasing the cost of travelling by car (e.g., by increasing the tax on fossil fuels or increasing parking fees) is a significant factor causing modal shift from car to other transport modes.
Hypothesis #4: Reducing the travel time (in-vehicle) in public transport, for example by introducing managed lanes, would significantly increase the attrac- tiveness of public transport.
Methodology The first step of the analysis will be the review of current studies on the modal split. Reviewing the research designs and econometric models, together with the variables (and the controls) used in the previously conducted studies will help us with creating the optimal questionnaires for our survey. Moreover, I also aim to analyze studies dealing particularly with the Integrated Transport Systems, since we are planning to include a scenario with a multimodal transport (that is, combining two or more modes of transport throughout the journey) in the stated preference part of the survey.
After we obtain sufficient information regarding this topic from the previously published literature, we can begin to design the survey which will be used to obtain our dataset. With the help of my supervisor and his research team, our goal is to design an extensive survey, which will be conducted in several major Czech cities.
In this survey, we are planning to use both revealed preference and stated prefer- ence techniques. Using the revealed preference, we will be collecting data on actual travel decisions of the participants. Mainly, we aim to achieve this by asking the participants to fill out a 1-day travel diary, as it is typical in such studies. Using just one day for the travel diary seems to be optimal, due to the participants’ increasing
“fatigue” in the consequent days (Scheiner, 2010). Regarding the stated preference
approach, we will present the participants various hypothetical scenarios, in which they should choose their preferred transport mode based on the presented attributes (such as reliability, time spent, and costs).
After collecting the data, my next step will be analyzing the relationship between one’s choice of travel mode and factors which influence it. This analysis will be done using logistic regression models. Namely, I am planning to firstly use multinomial logit (MNL) that is suitable for modelling choice among more than two alternatives dependent on commuter’s characteristics. Conditional logit (CL) is then suitable to model commuters’ choices dependent on the attributes of the travel modes. I will discuss other econometric models suitable to model unobserved preference het- erogeneity, such as mixed (random parameter) logit (MXL) or hybrid mixed logit.
Regarding the research design, I intend to analyse preferences for at least four pos- sible transport modes: private car, public transport, cycling and walking. I am planning to run the models using the STATA software. After conducting the analy- sis, the final part of the thesis will be dedicated to proposing optimal policies based on the results coming from the research.
Expected Contribution With the help of my supervisor and his team, we are planning to design a rather extensive survey which will be conducted in several ma- jor Czech cities and use this fresh dataset for the modal split analysis. Regarding the contribution of this thesis, the results of my analysis will shed some more light on the travel behavior in the region of Central Europe, which has not been fully analysed yet in terms of the modal split models. Moreover, since vast majority of already con- ducted modal split models in Europe have relied on the revealed preference approach only, the inclusion of the stated preference approach in my thesis as well can provide additional source of information for the policy design. Lastly, I am planning to use the results as a basis for proposing policies which could be implemented in order to reduce CO2 emission levels as well as improving the air quality and noise levels in the cities.
Outline
1. Introduction: I will briefly introduce the topic of modal split to the reader. I will also very briefly explain the principles of the integrated transport systems.
2. Literature review: I will summarize what is already known about the modal split problematics from previously conducted studies, analyzing both studies using revealed preference and studies using stated preference approach. Part of the literature review will be also dedicated to integrated transport systems.
3. Data: I will describe the process by which the dataset was obtained, I will summarize the data and I will discuss the choice of the variables.
4. Methods: I will introduce the methodology used in the thesis, which will be mainly multinomial logit (MNL), conditional logit (CL) and mixed logit (MXL).
5. Results: In this part, I will estimate the models and interpret the estimation results.
6. Discussion and Conclusion: I aim to discuss my findings, and based on the results, I would like to propose optimal policies which should lead towards more sustainable urban transportation.
Core bibliography
Gehrke, S. R., & Clifton, K. J. (2014). Operationalizing land use diversity at varying geographic scales and its connection to mode choice: Evidence from Portland, Oregon. Transportation Research Record, 2453(1), 128-136.
Ding, L., & Zhang, N. (2017). Estimating modal shift by introducing transit priority strategies under congested traffic using the multinomial logit model.
KSCE Journal of Civil Engineering, 21(6), 2384-2392.
Halawani, A. T., & Rehimi, F. (2021). Evaluation of the intention to shift to public transit in Saudi Arabia. Transportation research part D: transport and environment, 94, 102809.
Cheng, L., Chen, X., De Vos, J., Lai, X., & Witlox, F. (2019). Applying a random forest method approach to model travel mode choice behavior. Travel behaviour and society, 14, 1-10.
Jin, F., An, K., & Yao, E. (2020). Mode choice analysis in urban transport with shared battery electric vehicles: A stated-preference case study in Beijing, China. Transportation Research Part A: Policy and Practice, 133, 95-108.
Kim, S., & Ulfarsson, G. F. (2008). Curbing automobile use for sustainable transportation: analysis of mode choice on short home-based trips. Trans- portation, 35(6), 723-737.
Kitamura, R. (2009). A dynamic model system of household car ownership, trip generation, and modal split: model development and simulation experi- ment. Transportation, 36(6), 711-732.
Molinillo, S., Ruiz-Montañez, M., & Liébana-Cabanillas, F. (2020). User char- acteristics influencing use of a bicycle-sharing system integrated into an in- termodal transport network in Spain. International Journal of Sustainable Transportation, 14(7), 513-524.
Nurdden, A., Rahmat, R. A. O. K., & Ismail, A. (2007). Effect of transporta- tion policies on modal shift from private car to public transport in Malaysia.
Journal of applied Sciences, 7(7), 1013-1018.
Sabir, M. (2011). Weather and travel behaviour (Doctoral dissertation, Vrije Universiteit).
Santos, G., Maoh, H., Potoglou, D., & von Brunn, T. (2013). Factors in- fluencing modal split of commuting journeys in medium-size European cities.
Journal of Transport Geography, 30, 127-137.
Scheiner, J. (2010). Interrelations between travel mode choice and trip dis- tance: trends in Germany 1976-2002. Journal of Transport Geography, 18(1), 75-84.
Schwanen, T. (2002). Urban form and commuting behaviour: a cross-European perspective. Tijdschrift voor economische en sociale geografie, 93(3), 336-343.
Ton, D., Duives, D. C., Cats, O., Hoogendoorn-Lanser, S., & Hoogendoorn, S.
P. (2019). Cycling or walking? Determinants of mode choice in the Nether- lands. Transportation research part A: policy and practice, 123, 7-23.
Introduction
During the last decades, a high emphasis has been put on reducing the emission levels of CO2, air quality pollutants and noise in the cities, which are caused mainly by the high number of cars in the urban traffic. In order to solve this issue, many governments and organizations are trying to propose policies, which aim to promote more environmentally friendly, sustainable and healthier forms of urban transport, such as using public transport, cycling or walking. However, when designing such policies, one needs to know which factors exactly influence the individuals’ choice of transport mode, in order to determine the optimal way to motivate the individuals to switch to ecological modes of transport.
In order to identify these factors, the researchers conduct so-called modal split studies. The purpose of such studies is to find out what influences respon- dents’ travel behavior and more precisely, their transport mode choice. So far, a considerable amount of such factors have been already found. First of all, the vast majority of modal split literature concludes that socio-demographic characteristics (gender, age, education, income etc.) serve as key factors when it comes to determining the probability of transport mode choice. Next, most researchers are also dealing with trip-related variables, such as trip purpose (Buehler, 2011), trip distance (Kim & Ulfarsson, 2008), and eventually travel time (Carrone et al., 2020). Further, one must also take into account factors such as ownership of driver’s license (Kim & Ulfarsson, 2008), having a public transport subscription (Ton et al., 2019), the density of the public transport network (Cheng et al., 2019), or eventually the distance to the nearest pub- lic transport station (Nurdden et al., 2007). Some researchers also claim that weather conditions (rain, temperature, wind) have a significant impact on the choice of transport mode (Sabir, 2011). Last but not least, we should definitely
not forget considering the cost of using the particular transport mode (Carrone et al., 2020).
In order to design a modal split study, the researcher can either use a re- vealed preference or a stated preference method. Revealed preference approach consists of observing the real travel decisions of the participants, which is ac- complished mostly by collecting a so-called travel diary (a documentation of trips, which the participant has made during a specified period of time). The stated preference approach, on the other hand, is studying the travel decisions of people in hypothetical contingent scenarios. That is, the participant is given a task of getting from point A to point B along with several transport mode options, each of which is assigned certain parameters, such as cost, travel time, access time or some form of measurement of comfort. Both of these meth- ods have their advantages and disadvantages, as will be discussed later by the author.
So far, majority of the European modal split studies were conducted in the states of Western Europe, such as Netherlands (Ton et al., 2019), Ger- many (Buehler, 2011) or Denmark (Carrone et al., 2020). A considerably large amount of mode choice studies are also taking place in the USA, such as the work of Kim & Ulfarsson (2008) or Gehrke & Clifton (2014). Moreover, some studies regarding this topic are also being conducted in the Asia-Pacific area, namely in China (Jin et al., 2020) or Malaysia (Nurdden et al., 2007), where the stated preference technique is often being used in order to test for the attractiveness of newly introduced modes of transport.
However, to my knowledge, not many large-scale modal split studies have been carried out so far in the area of Central and Eastern Europe. Thus, the aim of this thesis is to shed more light on the travel behavior of inhabitants of Czech cities. More precisely, the author analyzes the factors which influence the transport mode choice of individuals in five largest Czech cities, using data from a nationwide travel survey Česko v pohybu, carried out byCentrum dopravního výzkumu between 2017 and 2019. This dataset provides us with 9858 observations of unique trips in the major Czech cities along with a wide range of variables, allowing for thorough analysis. The data are analyzed using multinomial logit model, which is one of the most widely used approaches for analyzing mode choice.
The results confirm the significance of certain socio-demographic factors, such as income, age or education, which are known from previous research on this topic. At the same time, the author believes that this thesis increases
the understanding of the importance of these factors. Further, our findings suggest that in some aspects, the travel behavior of inhabitants of Czech cities differs from the behavior of Western European citizens, that was documented in the previous studies. The thesis further shows surprisingly high popularity of public transport among Czech citizens as well as the potential need for better promotion of active mobility, especially cycling.
Further, based on the results, the author gives several recommendations on how the Czech citizens could be motivated to use more ecological forms of transport. Among other outcomes, the results suggest that there might be a need for subsidizing public transport for the lowest income group or for equip- ping more public transport vehicles with air-conditioning, while on the other hand, entitlement to discounted public transport fares thanks to a discount card such as ISIC does not seem to play a significant role in the probability of using public transport.
The thesis will be structured in the following way: in Chapter 2, the author would like to discuss what is already known from the previously conducted studies on the topic of modal split as well as to explain the most common data collection methods in modal split surveys. The Chapter 3 will be dedicated to describing the dataset used for the analysis both verbally and graphically.
The Chapter 4 will describe the methodology and in Chapter 5, the results will be presented and compared to previous research. In the Chapter 6, the author discusses the data-driven recommendations that could motivate Czech citizens towards more ecological and healthy forms of mobility and finally, the Chapter 7 provides final summary of the thesis and presents suggestions for further research on modal split in the Czech Republic.
Literature review
Throughout the last few decades, many modal split studies have been already conducted all over the world, using various methods and including several fac- tors which have been proven to have an impact on transport mode choice.
This part of the thesis is dedicated to summarizing what we already know about modal split from the previous literature - the first part of the Literature review aims to discuss all the important factors included in the studies and further, the second part of this chapter will focus on explaining the data col- lection methods that can be used when conducting travel surveys, along with examples from previous research.
2.1 Factors influencing transport mode choice
A considerable number of factors, which were shown to have an impact on the travelers’ transport mode choice, is already known from the current literature.
Just to give a few examples, some studies emphasize the importance of the quality of public transport network (Chenget al., 2019), while other researchers are focusing on factors such as bike network length (Santos et al., 2013) or weather (Sabir, 2011). However, certain variables are common for the vast majority of studies, and those are the socio-demographic characteristics.
2.1.1 Socio-demographic variables
Under the term socio-demographic variable, one can imagine variables such as age, gender, education level, income and many others. It is clear to the vast majority of researchers that such factors can have a significant impact on the mode choice, and therefore they are included in almost every modal split model.
Regarding age, there is a consensus among the researchers that higher age leads to a decreased use of active modes of transport (walking and cycling). A significant decrease in the use of those two transport modes with the increasing age has been found by Kim & Ulfarsson (2008) as well as by Gehrke & Clifton (2014). However, there is a slight disagreement regarding the relationship be- tween age and use of public transport. While Chenget al. (2019) and Nurdden et al.(2007) claim that older age leads to higher percentage of public transport use, Meena et al. (2019) have arrived to an opposite result. Moreover, Santos et al.(2013) have also found lower percentage of public transport users in cities with higher percentages of elderly citizens. Regarding car use, most researchers agree that the ratio of car users is positively correlated with age. This thought is supported for example by the work of Kim & Ulfarsson (2008) or by Ding &
Zhang (2017).
Another socio-demographic variable which appears to have an impact on mode choice is gender. Both Buehler (2011) and Gehrke & Clifton (2014) have found that male travelers are more likely to cycle than women. However, it is notable to mention that both of these studies have been conducted in the USA, where the urban cycling conditions may be generally more difficult than in Europe. On the other hand, the work of Ton et al. (2019), which takes place in the Netherlands, shows that in this geographical area, gender does not serve as a significant variable affecting the use of bicycle. This may possibly be caused by the fact that urban cycling is much more common in the Netherlands than in the USA among majority of citizens regardless of gender.
Further, according to previously conducted studies, being a male also seems to increase the probability of using a private car (Ding & Zhang, 2017) or car-sharing (Carrone et al., 2020). Finally, Halawani & Rehimi (2021), who examine the willingness to switch to public transport in Medina, Saudi Arabia, where the public transport network is not fully built yet, claim that women are more willing to switch to using public transport than men. The exact same finding has been also discovered by Nurdden et al. (2007).
Furthermore, it seems that a higher level of education is correlated with higher use of active transport modes. While Kim & Ulfarsson (2008) claim that people with a college degree are more likely to choose walking as their preferred mode, Gehrke & Clifton (2014) as well as Ton et al. (2019) arrive at a conclusion that higher education leads to an increased probability of using a bicycle. Even the fact that a person is currently being a student apparently has a significant impact on the choice of active transport mode, as Ton et al.
(2019) state that students are more likely to commute by a bicycle than other citizens. Kim & Ulfarsson (2008) have found out that such individuals, whose trip purpose is going to school, have higher probability of walking, cycling or taking a bus and a lower chance of using a car to drive to school. This finding is also supported by Santos et al. (2013), who claim that cities with higher percentages of students have higher ratios of walking and cycling commuters, as well as people using public transport. However, this may be partially caused by the fact that students do not often own a driver’s license yet, or that purchasing and maintaining their own car may be too costly for them. On the other hand, it is also possible that the younger generation is more environmentally aware and therefore refrains from using unsustainable modes of transport.
Concerning income, it is to be expected that people with higher income would be more likely to use a car to commute, since purchasing, owning and maintaining a car is much more costly than using other means of transport.
This thought is in accordance with the findings of Dargay & Hanly (2007) as well as Ding & Zhang (2017), as both of these research teams find a positive relationship between income and private car use. The attractiveness of cars among the high-income earners goes hand in hand with a lower interest of those citizens in other means of transport. Meenaet al.(2019) show a negative relationship between income and use of public transport. Further, Gehrke &
Clifton (2014) not only mention a decreased chance of using public transport within higher income levels, but they have also observed a decrease in cycling and use of high-occupancy vehicles (motor vehicles carrying more than one person). Such result suggests that when people with higher income use a car, they often drive it as a single passenger, while the poorer citizens are more likely to drive with two or more people in a car, if they already decide to use it.
The presence of children in a household also seems to serve as a crucial factor in terms of mode choice. More precisely, majority of studies conclude that households with children are less likely to use public transport. Such relationship has been found for example by Santos et al. (2013) or by Ton et al. (2019). On the other hand, Kitamura (2009) has arrived at an opposite result, claiming that families with children tend to use public transport more.
Regarding the use of active means of transport, both Kim & Ulfarsson (2008) and Maley & Weinberger (2011) conclude that members of households with children are less likely to walk to their final destination. Naturally, with a decrease in use of certain transport modes by such families, an increase in use
of a different mode has to occur. An evidence of this statement can be found in the work of Dargay & Hanly (2007), who claim that families with children are more likely to use a private car.
2.1.2 Mode-related variables
Naturally, factors such as ownership of driver’s license, subscription to pub- lic transport system or the actual ownership of a particular transport mode can have a significant impact on the individuals’ mode choice. Moreover, the comfort, quality and affordability of the individual modes is also crucial. This section of the Literature review will be dedicated to those factors, which the author has decided to collectively label as “mode-related” variables.
Firstly, the owners of a driver’s license logically tend to be more likely to use a car to commute (Cheng et al., 2019), while being less likely to take a bus or to walk (Kim & Ulfarsson, 2008). An even more important factor is the ownership of the car itself. It is clear that the researchers agree on the fact that the travelers who own a car are more likely to use it to commute. Such finding can be observed in the work of Scheiner (2010), Paulley et al. (2006) or Kitamura (2009). The car owners also tend to be less likely to use public transport, to walk or to cycle, which has been confirmed by Scheiner (2010), Kim & Ulfarsson (2008) or Buehler (2011). Maley & Weinberger (2011), who examine the travel behavior of citizens of Philadelphia, USA on their shopping trips, also arrive to a conclusion that car owners are less likely to travel to the shop on foot. Similarly, bicycle ownership also naturally appears to be positively correlated with higher levels of cycling, as confirmed by Carrone et al. (2020).
Further, according to previous research, there are supposedly certain factors which discourage people from using cars. First, an increase in petrol price serves as a significant disincentive when it comes to car use (Dargay & Hanly, 2007). Ding & Zhang (2017) also mention an increase in parking fees as one of the most powerful policies which should lead to a shift from cars to public transport. Moreover, Carrone et al.(2020) claim that parking search time can also be one of the important factors, especially regarding car-sharing use.
Concerning the demand for public transport, the most important factors are allegedly the prices of public transport fares, quality of public transport, density of public transport network as well as the distance from home to the nearest public transport station. Firstly, Santoset al.(2013) and Paulley et al.
(2006) state that higher public transport fares serve as a significantly discourag- ing factor from using public transport, while the latter research team believes that the effect is even more prominent in the rural areas, where the switch from public transport to private car can be made more easily. This team of re- searchers also emphasizes the importance of quality of public transport. Cheng et al.(2019) concentrate on the structure of public transport network and their results appear to be very meaningful - according to these researchers, longer distance from home to a nearest subway or bus station has a negative impact on public transport use, while building more dense public transport network would serve as a great incentive to boost the demand for this service.
The idea of importance of the distance to the nearest public transport sta- tion is further supported by two research teams that are examining the will- ingness to shift to using public transport in the cities of Asia-Pacific region, namely Nurdden et al. (2007) and Kedia et al. (2017). In addition, Ding &
Zhang (2017) also propose the introduction of Managed Bus Lanes (traffic lanes in which the buses are prioritized and therefore are able to reach their destination in much lower time). Logically, owning a public transport subscrip- tion also plays a substantial role when deciding what transport mode to choose.
Tonet al.(2019) and Carroneet al.(2020) unanimously claim that owning such subscription significantly increases the chances of traveling by public transport.
2.1.3 Geographical variables
This section of the review aims to focus on geographical variables such as city size or population density, which also supposedly have an impact on mode choice. First, greater city size leads to higher use of public transport, while the ratios of walking and cycling trips decrease (Schwanen, 2002). This finding is rather intuitive, since the citizens of large cities often have to travel longer distances that are difficult to manage by bicycle or on foot. This issue is further developed by Scheiner (2010), who finds that cycling or walking actually can be more frequent in larger cities, but only for short distance trips. Santos et al.
(2013) are dealing with population size rather than city size, and arrive at a conclusion that inhabitants of cities with larger population size are more likely to use public transport.
Further, some researchers also include population density in their analysis, such as Schwanen (2002) who finds a positive relationship of population den- sity and public transport use, and a negative relationship with proportions of
walking and cycling. An interesting perspective on this issue is also offered by Buehler (2011), who compares the travel behavior in the USA and in Germany.
The above-mentioned researcher claims that although higher population den- sity seems to cause an increase in public transport use in both countries, there seem to be some differences regarding the use of active means of transport.
While higher population density leads to higher proportions of walking and cycling in the USA, the results seem to be the opposite in Germany. Dargay
& Hanly (2007) add that higher population density can also be responsible for lower car use.
2.1.4 Trip-related variables
Next, the author should definitely not forget discussing factors which are related to the specific trip characteristics, such as travel distance, travel time or trip purpose. First of all, majority of the scientists arrive at a conclusion that there is a negative relationship between trip distance and use of active modes of transport. Scheiner (2010) observes a more likely use of car or public transport at the expense of walking, as the trip distance increases. Kim & Ulfarsson (2008) arrive at a fairly similar result, stating that larger trip distance favors the use of bus, while the odds of walking decrease. Further, Maley & Weinberger (2011), whose research work deals mainly with travel behavior on shopping trips, claim that the citizens traveling to a shopping mall are also less likely to walk as the distance to the mall increases.
The same kind of relationship allegedly holds for travel time as well. Firstly, Ton et al. (2019) argue that higher travel time supposedly favors more com- fortable means of transport, such as car or public transport, at the cost of ac- tive transport modes. This thought is further supported by Gehrke & Clifton (2014), who also observe a decrease in walking and cycling as the travel time of a trip increases. In addition to that, a decrease in cycling with higher travel time has been also found by Carrone et al. (2020).
Another factor that is worth addressing is the purpose of the trip. According to the previous literature, it is clear that the modal split differs in some way across different trip purposes. For example, going shopping allegedly increases the chances of traveling by car (Meenaet al., 2019). This finding could possibly be justified by the travelers’ need to transport weighty goods from the shop, which contributes to their decision to use a comfortable mode of transport, such as car. Buehler (2011), who compares the travel behavior in Germany
and in the USA, finds some differences between those two countries regarding the impact of trip purpose. Concerning trips to work, the American citizens seem to favor public transport at the expense of walking and cycling, while the Germans are likely to travel to work by public transport as well as on foot.
When traveling to a shop, a decrease in public transport, cycling and walking occurs in the USA, while the coefficient is positive in Germany for all of these transport modes. This difference in the travel behavior may be caused by the fact that the shops in Germany are often more densely distributed in the cities, while in the USA the citizens usually have to travel larger distances to get to a shopping mall, which discourages them from using active modes of transport.
2.1.5 Weather-related variables
Last but not least, since some scientists claim that weather conditions also serve as a significant determinant of mode choice, especially in the matter of active transport modes, the author has decided to dedicate a short section of the review to these factors as well. The connection between weather and mode choice has been mainly examined by Sabir (2011), who is analyzing the influence of three main weather-related variables: temperature, wind and rain.
Unsurprisingly, Sabir (2011) finds that harsh conditions (strong wind, rain or low temperatures) all lead to a significant shift from cycling to car or eventually to public transport. Similar conclusion has been made by Carroneet al.(2020), who also find a significant decrease in cycling when rain occurs.
2.2 Data collection methods in travel surveys
As the author has already mentioned in the Introduction, travel behavior sur- veys can be conducted using two main techniques. More precisely, these meth- ods are known under the terms revealed preference and stated preference. In this part of the thesis, the author would like to discuss these data collection methods and specific examples of their application in the studies.
2.2.1 Revealed preference method
One of the time-proven methods, which can be used to collect travel behav- ior data, is the revealed preference. Using this technique, the researchers ob- serve real-life travel behavior of the survey participants (the real-life prefer-
ences of the respondents are “revealed” to the researchers, hence the name of the method). This process is usually accomplished using a travel diary, where the respondents tracks all his/her travel activities during a certain time period along with additional information about the trips.
The travel diary can take many forms - it can be a classical pen-and-paper diary, it can be located on a website or it can be filled in using a mobile applica- tion developed for this purpose. However, use of such mobile applications still faces certain issues nowadays, such as high demand on phone battery, issues with GPS signal or non-availability of translations into less widely spoken lan- guages (Meister et al., 2020), which is the main reason why many researchers still prefer traditional pen-and-paper travel diaries or website surveys.
An important aspect of RP design is choice of the period, during which the data are collected. Usually, the travel diaries are designed to collect travel data for 1-3 days. It is generally not recommended to collect the data for a longer period, since the participants may become unwilling to fill out the diary if the period is too long. This “fatigue” of participants then may serve as a potential source of bias (Scheiner, 2010). In case of our dataset, the travel data were collected for only one “decisive day”, in order to avoid this “fatigue bias”.
Further, the date of collection would differ for each interviewed household, allowing us to control for season of year.
The relative simplicity of RP method combined with the fact that it is possible to conduct it using only pen and paper is one of the great advantages of RP surveys, allowing to conduct such surveys also in developing countries, where the access to modern technologies is not so widespread. An example of a simple pen-and-paper travel diary incorporated into a survey conducted in Surat, India by Kedia et al. (2017) is presented in the picture below.
Figure 2.1: Example of a pen-and-paper travel diary
Source: Kediaet al.(2017).
This particular simple example requires the respondent to state the trip pur- pose, member of the household who made the trip, the used transport mode, reason for using the mode, destination of the trip, approximate trip length in kilometers, departure time and travel time in minutes. The author would like to emphasize, that in majority of the surveys, the actual travel diary is also sup- plemented by questions on socio-demographic characteristics of the respondent and possibly also other individuals living in the respondents’ household.
2.2.2 Stated preference method
Including stated preference scenarios in the travel survey is a more recent ap- proach than the more traditional use of revealed preference. According to Kroes & Sheldon (1988), stated preference methods can be defined as tech- niques, which estimate utility function using participants’ preferences in a set of transport options. In other words, the participants are presented with two or more possible options of getting from point A to point B, where each option includes different transport mode with its own attributes, such as time spend inside and outside of vehicle, cost or some measurement of comfort. Based on these attributes, the participant will then choose his/her preferred option. For better understanding, an example of a stated preference scenario used in the study of Carrone et al. (2020) is presented below.
Figure 2.2: Example of a stated preference scenario
Source: Carroneet al.(2020).
In this particular scenario, the participants are required to choose from the four available transport mode options, which differ in their parameters: travel time, cost, access time and eventually parking search time and parking cost.
Further, the respondents are presented the information that the purpose of this particular trip is the commute to their place of work or education and that it is raining. However, the stated preference scenarios can also include other attributes, such as availability of cycling paths, number of transfers in the public transport or some measurement of comfort of the presented modes. On the other hand, it is generally recommended to restrict the number of attributes to a reasonable amount (approximately 3 attributes per transport mode), not only to prevent overloading the respondent with too much information, but also due to increased complexity of the econometric analysis when adding too much attributes.
Further, when designing stated preference scenarios, it is important to present the options in a way that neither of these alternatives is strictly dom- inant nor strictly inferior in terms of the attribute values. Including such alternatives can potentially result in biased estimates (Danafet al., 2019). The only case, when it is suitable to include a strictly dominant alternative, is to test whether the participants are not selecting the options at random. More precisely, if not selecting randomly, all participants should naturally choose the strictly dominant option, according to the neoclassical economic theory of utility maximization. If the other options are being selected, however, it may
serve as an indication that the participants are picking the options at random regardless of the presented attribute values.
One of the potential advantages of using the stated preference technique is its ability to evaluate demand under conditions that not yet exist or that are not yet fully available to all participants (Kroes & Sheldon, 1988).
This property can be useful especially when conducting modal split studies in developing countries, where some transport modes are not available yet either to some citizens or at all. As an example, let the author mention the work of Halawani & Rehimi (2021), whose work is dedicated to examining the willingness to shift to public transport in Medina, Saudi Arabia, where the public transport network has yet been unbuilt at the time of conducting the study. Further, some scientists are using the stated preference technique to examine the willingness to use some futuristic modes of transport, as seen in the work of Ilahi et al. (2020), who have included a not yet existing transport mode which the researchers have labeled as “Urban Air Mobility” in their stated preference survey.
Another advantage of using SP is that the survey designers can set various levels of attributes, allowing for evaluation of these attributes in the models. For example, the researchers can vary the costs of travelling by individual transport modes, which subsequently allows them to test, how important is this cost for the choice makers. Such results are eventually useful when implementing policies, such as subsidizing public transport, taxing cars or increasing price of parking spaces in the cities.
On the other hand, one of the disadvantages of SP is the potential un- realisticness of the hypothetical scenarios. Due to this issue, the responses might not predict actual behavior of respondents, which may cause potential bias (De Corte et al., 2021). Further, conducting stated preference surveys is generally also more demanding both in terms of organization and finance.
2.2.3 Combining revealed and stated preference methods
In order to avoid unrealistic scenarios, it is also possible to use these two meth- ods jointly - the researchers firstly obtain the revealed preference data through a travel diary and then construct the stated preference scenarios based on the RP data, such that the hypothetical scenarios have similar characteristics to routes listed in the travel diary. Such method is thus also occasionally called joint preference. The purpose of this approach is to make the hypothetical sce-
narios more realistic to the respondent, which should lead to more relevant out- puts. This procedure has been used for example in the work of Carrone et al.
(2020), who emphasize that presenting realistic attributes to the respondent serves as one of the most important aspects of stated preference experiments.
2.2.4 Method used in our dataset
Although stated preference is considered to be a more recent method with cer- tain advantages over RP, the author did not unfortunately have access to any stated preference data on travel behavior. The Česko v pohybu dataset (here- inafter also referred to as CVP dataset), which is available to the author, has been collected using travel diaries implemented in the surveys (which, however, provide a rather extensive amount of information), which is characteristic for the revealed preference method. To the author’s best knowledge, in the Czech Republic, there does not exist any publicly available dataset using SP of such scale as the CVP dataset at the time of writing this thesis, therefore using the RP data from CVP appeared to be the best option for the author. The reason of non-existence of a large-scale travel behavior survey using SP in the Czech Republic might be its higher design complexity, which is more demanding both in terms of finance and time.
Data overview
This section is dedicated to describing the dataset used for the analysis both verbally and graphically. In the first section, the author will firm inform the reader about the characteristics of the CVP dataset as well as about the vari- ables used in the model, and in the second section, the author will provide a wide overview of the descriptive statistics of the used variables.
3.1 Dataset used for the analysis
The data used in the author’s analysis were obtained from a survey “Česko v pohybu” (Czechia in Motion), which was conducted in years 2017-2019 on a national level. To the author’s best knowledge, this survey carried out by the Transport Research Centre (Centrum dopravního výzkumu) is currently the most extensive survey focusing on travel behavior in the Czech Republic.
This survey includes data collected from randomly selected 9,419 house- holds, whose residents were asked to fill out an extensive one-day travel diary (in other words, the data were obtained using a revealed preference technique).
More precisely, members of each household were asked to select one specific day, in which they would record all their travel activities along with all impor- tant information, such as the selected transport mode, purpose or duration of the trip. In total, the researchers were able to collect data for 51,434 unique trips along with a rich set of variables related to the respondents.
Since the purpose of this thesis is to analyze travel behavior in large cities, the author has decided to use solely the travel data from five largest Czech cities (Praha, Brno, Ostrava, Plzeň and Liberec). Namely, only such observations, for which both the Starting point and the Final point of the trip was located
in that particular city were filtered, ensuring that the whole trip was carried out within the city. Moreover, children under 18 years of age were removed from the dataset by the author, since these children do not often have the option to choose their preferred mode by themselves - often, they are told to use certain mode by their parents, for example, they might be told to take a bus to school. The inability of choosing the means of transport on the basis of their own choice could lead to biased results. Further, the author has decided to remove the trips that were made by a transport mode categorized as “Other”
due to very small amount of such observations. After removing all unwanted observations, our final dataset consists of 9,858 trips made by 3,688 unique respondents.
The CVP dataset consists of 4 individual datasheets - Households, Cars, Individuals and Trips, each including information on the respective topic. All of the variables used in our models were either directly obtained from these datasheets, or calculated using information from those datasheets. Some of the variables were also slightly modified (for example, in case of categorical variables, by grouping categories with low amount of observations and similar characteristics, as described in the subsection below). Moreover, a brief section of this thesis is also dedicated to comparison to other European countries. For this purpose, a dataset “City statistics (urb)” from Eurostat (2022) has been used.
3.1.1 Choice of variables for the model
Regarding the choice of variables for our model, the author tried to follow the structure of variables used in previous research on modal split, as well as to attempt to introduce some new variables that could be obtained from the CVP dataset. As the author has already mentioned, some of the categories were grouped into one due to low amount of observations. In case of the dependent variable mainmode, the original categoriesPublic Transport, Regional Bus and Train were grouped into one category denoted as Public Transport. Moreover, Walking and Cyclingwere grouped into category Active modes(also due to low amount of observations concerning the Cycling category).
Further, concerning theTrip purposevariable, categoriesGoing to work and Business Trip were grouped into Going to work and finally, Leisure time and Going to a restaurant were grouped into Leisure time. The final list of used variables is shown in the Table 3.1 below. The sample thus includes a dependent
variable “mainmode” which can take on 4 different categories, which will be regressed on total 14 independent variables.
Table 3.1: Variables used in the model
Variable Description Categories
mainmode Dependent variable, indi- cating the main transport mode used for the trip
Car-driver Car-passenger Public transport Active modes city City, in which the particular
trip was made
Praha Brno Ostrava Plzen Liberec gender Dummy variable for gender,
equal to 1 if respondent is female
Male Female
age Categorical variable for re- spondent’s age
19-29 years 30-39 years 40-49 years 50-59 years 60-69 years 70+ years Not stated education Categorical variable for re-
spondent’s highest educa- tion level
Unfinished primary Primary
Secondary with vocat. certificate Secondary with final exam Higher professional
University
Variable Description Categories income Household monthly income
(CZK)
less than 15000 CZK 15001-22000 CZK 22001-30000 CZK 30001-50000 CZK More than 50000 CZK Not stated
purpose Purpose of the trip Going home Going shopping Going to work
Going to school / university Leisure time
Arrangements Other
season Categorical variable for sea- son of the year, in which the trip was conducted
Spring Summer Autumn Winter pt_discount Dummy variable equal to 1
if the respondent is a holder of discount card for public transport (ISIC, ZTP etc.) p_caravail Dummy variable equal to 1
if the respondent has access to a car
p_bikeavail Dummy variable equal to 1 if the respondent has access to a bicycle
p_ptavail Dummy variable equal to 1 if the respondent owns a long-term PT subscription chldrn_u_5 Number of children under 5
years of age living in respon- dent’s household
Variable Description Categories p_ea Dummy variable equal to 1
if the respondent is econom- ically active
distance Measured distance from the Starting point to the Final point, in kilometers
3.2 Descriptive statistics
The final dataset used for the analysis consists of trips made in the five largest Czech cities, which were filtered from the raw data and cleaned. The proce- dure of preparing the data is described in the chapter above. After filtering and cleansing, the data consist of 9,858 trips made by 3,688 unique individu- als. Most of these trips were made in Praha (5,985 trips), followed by Brno (1,652 trips), Ostrava (1,221 trips), Plzeň (618 trips) and Liberec (382 trips).
Although walking and cycling trips will be grouped into one category in the model due to low amount of observations for cycling trips, the author has de- cided to keep these two transport modes divided in the descriptive statistics in order to illustrate the low levels of cycling trips in the Czech cities. In this chapter, the author would like to describe both verbally and graphically the structure of the data and how the modal split differs with respect to various variables. Further, the author would also like to compare graphically the travel behavior in Czech cities with similarly sized cities in Western Europe.
3.2.1 Mode choice by cities
Firstly, the author would like to show, how modal split differs across the five examined cities. A graph describing mode choice by cities is presented below.
Figure 3.1: Mode choice by cities
Source: Author’s calculations based on Centrum dopravního výzkumu (2022).
From this graph, we can observe the increasing popularity of car at the expense of public transport with the decreasing city size (the cities are displayed in a descending order of both population and area). Out of all cities, Praha has the highest percentage of trips made by public transport (46 %), while the percentage of trips by car as a driver and as a passenger are equal to only 20 % and 3 %, respectively. On the other hand, Liberec has the highest percentage of trips made by car (47 % as a driver and 11 % as a passenger), while only 19
% of all trips in Liberec were made by public transport.
This pattern is consistent with Scheiner (2010), who has also discovered that larger cities are generally characterized by lower use of cars and higher popularity of public transport. Regarding cycling, the current literature sug- gests that larger cities tend to have lower ratios of trips by bicycle (Schwanen, 2002; Scheiner, 2010), which can be also visible in our data. Namely, only 0.84
% of total trips were made on a bicycle in Praha, while the ratio of cycling trips in Brno, Ostrava and Plzen is equal to 1.82 %, 2.21 % and 1.94 %, respectively.
Interestingly, the smallest city in our data, Liberec, serves as an exception in this case, since only 0.26 % of trips in this city were made on a bicycle. Ac- cording to the author, one of the possible reasons might be the hilliness of this city, which may discourage people from using active modes of transport.
3.2.2 Mode choice by trip purpose
Further, it can be also expected that mode choice would differ with respect to the purpose of the trip. The modal split for different trip purposes is displayed in the Figure 3.2.
Interestingly, we can observe that quite a large proportion of participants goes shopping on foot (46 %), while only a small percentage goes by car (17 % as a driver and 6 % as a passenger). This pattern is in contradiction to Kim
& Ulfarsson (2008), who claim that when going shopping, people mostly tend to choose car over going by foot. However, the author would like to mention that Kim & Ulfarsson’s research was conducted in the USA, where the city infrastructure differs a lot from Europe. In European cities, the shops are usually more densely located than in the USA, therefore the citizens do not have to travel very far to get to a shop.
Figure 3.2: Mode choice by trip purpose
Source: Author’s calculations based on Centrum dopravního výzkumu (2022).
Further, approximately half of the trips to work are conducted by public transport, according to our data. Second most popular transport mode when going to work is car (30 % as a driver and 4 % as a passenger), followed by walking (15 %).
When going to school or university, majority of participants choose either public transport (65 %) or going by foot (25 %). The popularity of ecological
modes when travelling for education is consistent with Kim & Ulfarsson (2008).
In author’s opinion, there might be 2 reasons behind this: firstly, students do not often have access to their own car, since its ownership might be too costly for them. Secondly, the author supposes that students of higher education levels may have greater environmental awareness and therefore decide not to use a car to travel to university.
The author would also like to comment on the low popularity of cycling for almost all trip purposes. For none of the trip purposes does the ratio of cycling trips exceed 2 percent. In author’s opinion, this suggests that cycling in Czech Republic is still perceived as a sports activity rather than a fully- fledged transport mode, as opposed to Western countries such as Germany, Netherlands or Denmark, where the bicycle plays a very important role in city traffic. For example, according to Ton et al. (2019), more than 32 % of trips in the Netherlands are made using a bicycle.
3.2.3 Mode choice by household income
From previous research, we also know of the relationship between mode choice and income. The relationship of these two variables in our data is depicted in the following graph.
Figure 3.3: Mode choice by household income
Source: Author’s calculations based on Centrum dopravního výzkumu (2022).
The current research suggests that car use is positively correlated with income at the expense of use of public transport or active modes, such as walking or cycling (Ding & Zhang, 2017; Dargay & Hanly, 2007; Gehrke &
Clifton, 2014). This trend can be observed in our data as well. Car seems to be used very rarely in the lowest income groups. Interestingly, in the lowest income group (less than 15,000 CZK), people actually use the car more as a passenger than as a driver, suggesting that in many cases, these participants may not have access to their own car. The use of car then increases with each income level, ending up with 35 % of trips made by car within the highest income group (more than 50,000 CZK).
Intriguingly, among all income groups, public transport still seems to be more popular than a car, when travelling through the city. Further, a quite striking fact is that in the lowest income group, majority of the trips were made by foot (52 %), which could be potentially explained by lack of access to other transport modes. Another interesting pattern that we can observe is that the ratio of trips by bicycle increases with the household’s income level. While the percentage of cycling trips is equal to only 0.38 % for participants whose household monthly income is less than 15,000 CZK, this number increases up to 1.43 % for the highest income group.
3.2.4 Mode choice by education
Regarding the impact of education on mode choice, the foreign literature often associates higher education levels with increased use of active modes of trans- port, such as walking (Kim & Ulfarsson, 2008) or cycling (Ton et al., 2019;
Gehrke & Clifton, 2014). The author assumes that this may occur due to the fact that educated people are more environmentally aware and therefore they try to abstain from using car in the city traffic. However, the situation seems to be rather different in our dataset, as displayed in the graph below.
