Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

International Journal of Sustainable Transportation

ISSN: 1556-8318 (Print) 1556-8334 (Online) Journal homepage: http://www.tandfonline.com/loi/ujst20

Attitudes, norms and difficulties underlying road


sharing intentions as drivers and cyclists: Evidence
from the Czech Republic

Sigal Kaplan, Igor Mikolasek, Hana Bruhova Foltynova, Kira H. Janstrup &
Carlo G. Prato

To cite this article: Sigal Kaplan, Igor Mikolasek, Hana Bruhova Foltynova, Kira H. Janstrup
& Carlo G. Prato (2018): Attitudes, norms and difficulties underlying road sharing intentions as
drivers and cyclists: Evidence from the Czech Republic, International Journal of Sustainable
Transportation, DOI: 10.1080/15568318.2018.1471556

To link to this article: https://doi.org/10.1080/15568318.2018.1471556

Published online: 11 Jul 2018.

Submit your article to this journal

Article views: 21

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=ujst20
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION
https://doi.org/10.1080/15568318.2018.1471556

Attitudes, norms and difficulties underlying road sharing intentions as drivers


and cyclists: Evidence from the Czech Republic
Sigal Kaplana, Igor Mikolasekb, Hana Bruhova Foltynovac, Kira H. Janstrupd, and Carlo G. Pratoe
a
Department of Geography, Hebrew University of Jerusalem, Mount Scopus, Jerusalem, Israel; bTransport Research Centre, Czech Ministry of
 ı nad Labem, Czech Republic; dDepartment of
Transport, Brno, Czech Republic; cFaculty of Social and Economic Studies, UJEP, Ust
e
Management Engineering, Technical University of Denmark, Lyngby, Denmark; School of Civil Engineering, The University of Queensland,
Brisbane, Australia

ABSTRACT ARTICLE HISTORY


Cycling-inclusive urban planning is attracting attention worldwide because of the environmental, Received 26 October 2016
health, economic, and transport benefits inherent to cycling from the individual and the societal Revised 23 April 2018
perspectives. The Czech Republic is one of the emerging cycling countries that are trying to Accepted 25 April 2018
promote and support cycling, but cycling mode shares are low because of the poor quality of the
KEYWORDS
scarce infrastructure and the psychological barrier of the perceived lack of safety when cycling in Cycling; hybrid bivariate-
mixed traffic. This study takes a multimodal approach based on the Theory of Planned Behaviour ordered probit model;
and focuses on the factors underlying the willingness to share the road from the perspective of mixed traffic; road
cyclists and drivers: positive attitudes toward cycling, positive attitudes toward cyclists’ behavior, sharing; Theory of
social norms toward cycling and anxiety to share the road. A web-based questionnaire was tailor- Planned Behaviour;
designed and administered in the Czech Republic via social networks, web forums, and the Civinet
network, and a hybrid bivariate-ordered model tested the behavioral framework. Results show that
the willingness to share the road as a cyclist or as a driver relates positively to positive attitudes
towards cycling and cyclists’ behavior, and negatively to the anxiety of sharing the road. Moreover,
mediator effects are observed and a clear relation emerges between the experience on the road
as both a cyclist and a driver on the willingness to share the road as a cyclist. Lastly, results show
that the factors underlying road sharing intentions are related significantly to gender, travel habits,
and perceived personal and infrastructure-related barriers to cycling.

1. Introduction Despite its availability, affordability, and favorable social


norms (Titze et al., 2014; Tapp et al., 2016), cycling has low
Cycling-inclusive urban planning is attracting attention
mode shares because of safety concerns related to lack of
worldwide as cycling contributes to the individual and soci-
cycling infrastructure, negative attitudes toward cycling, low
etal well-being by relieving congestion, reducing environ- respect for cyclists’ needs by other road users, and reckless
mental emissions, preventing, and mitigating health risks behavior by car drivers (Daley et al., 2007; Fishman et al.,
related to physical inactivity, and providing an affordable 2012; Lawson et al., 2013; Chataway et al., 2014). Even more
transport solution that fosters social inclusion (see, e.g., than the actual risk, the perception of potential risk acts as a
Doorley et al., 2015; Murphy & Usher, 2015; Sagaris & psychological barrier, preventing its view as a viable every-
Ortuzar, 2015). Worldwide, bicycle ownership is 42% on day transport mode (Daley & Rissel, 2011; Mu~ noz et al.,
average, with highs of 80% in Scandinavia and northern 2013; Lawson et al., 2015). The perceived risk causes cycling
Europe, values of 60% in North America, Australia, China, and driving stress, affects mode choice, and differs between
southern Europe, and South America, averages of 40% in cyclists and motorists regarding the same incident with the
Russia, eastern Europe, United Kingdom, India, Africa, and latter perceiving higher risk (Chaurand & Delhomme, 2013).
Southeast Asia, and lows of 20% in Central Asia, the middle The fear of cycling relates to the lack of high-quality bicycle
east, and central and north Africa (Oke et al., 2015). infrastructure, especially on major roads and at intersections
Nevertheless, bicycle trip shares are as low as 1% in (Daley et al., 2007; Parkin et al., 2007; Chataway et al.,
Australia and North America and 2% in the United 2014), and does not decrease with the use of safety gear
Kingdom (Bassett et al., 2008) and as high as 18% in (Lawson et al., 2013; Chataway et al., 2014). The strength of
Denmark and 25% in the Netherlands (Oakil et al., 2016), the fear increases for inexperienced cyclists (Daley et al.,
and the cycled-km per person per year vary from 40 in 2007; Parkin et al., 2007; Chataway et al., 2014) and even a
North America and 75 in the United Kingdom to 584 in single cycling experience decreases significantly safety
Denmark and 848 in the Netherlands (Bassett et al., 2008). concerns (Gatersleben & Appleton, 2007).

CONTACT Carlo G. Prato c.prato@uq.edu.au School of Civil Engineering, The University of Queensland, St. Lucia 4072, Brisbane, Australia.
ß 2018 Taylor & Francis Group, LLC
2 S. KAPLAN ET AL.

Given the benefits of cycling, it is important to define as human infrastructure because they create networks in
equitable ways of organizing urban systems in a cycling- which they hold places of meaning and value, and provided
inclusive manner in order to mitigate social exclusion and the example of the CicLAvia event for advocating cycling in
vulnerability (Sagaris & Ortuzar, 2015). The development of Los Angeles that attracted over 100,000 participants and
bicycle infrastructure and dedicated cycling pathways is changed the visibility of cyclists and the social norms in the
often mentioned as a solution that increases the feeling of city. Cyclist visibility and shared road spaces promote the
safety (Chataway et al., 2014), relates to reduced cycling risk “safety-in-numbers effect” as they create valuable social
(Wegman et al., 2012; Kaplan et al., 2014; Kaplan & Prato, norms regarding sharing the road (Lugo, 2013).
2015), and supports the development of utilitarian cycling Understanding that the social climate forms an integral part
with high cost effectiveness even in cities without a strong of the physical urban environment is also useful for address-
cycling tradition (Marques et al., 2015). Nevertheless, ing spill-over effects. Forms of spill-over are the development
promoting road sharing between cyclists and motorists is an of cycling infrastructure in a municipality or a region that
important aspect of sustainable mobility because of three can stimulate cycling in neighbouring municipalities before
main reasons. cycling infrastructure is laid out (Vandenbulcke et al., 2011),
Firstly, cycling infrastructure often lags decades behind and the use of the road network by experienced cyclists to
the formation of a cycling culture in car-oriented countries answer their needs for the quick and direct bicycle commut-
where cyclists are still a minority (Prati et al., 2017). In ing routes (Meng et al., 2014).
Santiago de Chile for example, multiple cycling groups and This study focuses on the factors underlying the willing-
organizations advocated cycling infrastructure over a span ness to share the road as a cyclist and a driver, and its
of 15 years and a multitude of technical committees and relevance derives from the intensity of the road sharing
political roundtables discussed substantial changes to the city experience as manifested in recent studies. The terminology
streetscape by increasing the cycling network from 50 to used to describe the cyclist–motorist interaction by road
200 km, despite a Cycling Master Plan already existed from users and media shows an escalation of the dispute and a far
the 1990s (Sagaris & Ortuzar, 2015). Possibly, the implemen- from pleasant road sharing experience. Common terms are
tation of metropolitan master plans relies heavily on trad- “animosity”, “harassment”, “road wars”, “risk-takers”, and
itional cost-benefit analysis that has severe limitations in “law-breakers”, and both cyclists and drivers describe the
evaluating bicycle projects (van Wee & B€ orjesson, 2015), but road sharing experience as unwelcoming (e.g., Joshi et al.,
simply when cycling infrastructure is lagging behind the 2001; O’Connor & Brown, 2010; Daley & Rissel, 2011;
formation of a cycling culture, sharing the road becomes an Heesch et al., 2011; Lawson et al., 2013; Kaplan & Prato,
important measure for initiating and maintaining a positive 2016). A recent study shows that the reaction of road users
mass effect. to incidents is reactive rather than proactive and results in
Secondly, cycling crash risk in urban environments is refusing to assume responsibility, attributing the responsibil-
associated with infrastructure discontinuity points such as ity to others, and emphasizing the fault of the other party
intersections, bridges without cycling facilities, on-road tram instead of focusing on the incident (Paschalidis et al., 2016).
tracks, shopping center and garage entrances, and parked Along with aggressive and avoidance coping strategies, this
vehicles (Boufous et al., 2012; Kaplan et al., 2014; reaction explains the tendency of cyclist–motorist conflicts
Vandenbulcke et al., 2014). The discontinuity points not to escalate and the mutual anxiety and reluctance to share
only increase the actual risk, but also decrease the degree of the road (Kaplan and Prato, 2016). Solutions like cycling on
comfort in cycling because of the perceived risk (Krizek & sidewalks are not feasible because illegal in several countries
Roland, 2005). Sharing the road deals with risk proactively and unwelcome by pedestrians (Joshi et al., 2001; Daley &
because it requires a safe system approach where the Rissel, 2011), and solutions like setting artificial boundaries
inherent risks of traffic are dealt with by addressing multiple (e.g., a 1.0–1.5 m passing law that was initiated in Australia)
stakeholders according to a shared responsibility vision and create antagonism and deepen the controversy and the
a multi-dimensional prevention and mitigation perspective animosity (English and Salmon, 2016). Notably, in cycling-
(see, e.g., Cycling Embassy of Denmark, 2012; Wegman friendly countries there exists mutual respect between drivers
et al., 2012; Schepers et al., 2017). Safe system principles and cyclists stemming from the long-term cycling tradition
include (i) recognizing the limits of human performance, (ii) and the main role of cycling during childhood. Hence, there
acknowledging the limits of human tolerance, (iii) sharing is a need to understand how to create a cycling-inclusive
the responsibility for safe road use, and (iv) creating a social climate in emerging cycling countries on the basis of
forgiving road transport system by separation and injury mutual understanding across road users as well as a need to
mitigation (Pattinson & Thompson, 2014). go beyond the cyclist–motorist dichotomy and recognize
Lastly, urban traffic free routes as a standalone measure that the two roles are not mutually exclusive.
appear insufficient to encourage a shift from the car to the While previous studies focused on actions and reactions
bicycle in everyday practical journeys and a multifaceted on the road of cyclists and motorists, a recent qualitative
approach combining physical infrastructure with social study (Kaplan & Prato, 2016) focused on the chain of reac-
marketing is preferable (Jones, 2012). This approach views tion that leads to a particular result from a cyclist–motorist
changing the built environment to include more bicycle paths encounter, and identified external and internal stimuli as
and changing human landscapes to include more cyclists as well as coping strategies. This study extends this line of
complimentary measures. Lugo (2013) conceptualized people research by adopting the Theory of Planned Behaviour
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 3

(TPB, Ajzen, 1991), designing a questionnaire and estimating norms toward the behavior, and ease of taking action under
a hybrid model to elicit the attitudes, norms, and difficulties the assumption of volitional control.
of cyclists and motorists sharing the road with the aim of Existing literature on sharing the road supports the
allowing road users to take responsibility and be proactive applicability of the TPB. The effect of attitudes and norms
for their road behavior. The rationale lies in the importance as underlying factors in the decision to share the road is
of creating a meaningful dialogue among road users which evident in the dichotomy between the positive view of
allows them to “first seek to understand and then be under- cycling and its benefits and the fear of traffic as a barrier to
stood” and be better prepared to sharing the road. In this cycling, as well as the contrast between the popularity of
study, a questionnaire requested road users to envision shar- cycling for recreational purposes versus the portrayal of
ing the road as both cyclists and motorists, thus allowing to sharing the road as “road wars”. The behavioral framework
think with the end in mind and create a clear vision of the investigates the hypothesis that the intention to share the
road sharing experience in order to facilitate a win–win road by cyclists and motorists is related to positive attitudes
social climate respecting both the cyclist and motorist toward cycling in general and cyclist behavior in particular,
perspectives. is anchored in the belief regarding cycling tradition becom-
The case-study focuses on the Czech Republic, which is ing the norm in the Czech Republic, and is negatively asso-
marketed by the European Cycling Route Network ciated with the anxiety inherent in mixed traffic situations.
(EuroVelo) as a cycle tourist destination due to the well A tailor-designed questionnaire was prepared according
signed national cycle network and cycle friendly services. As to the hypothesis and was administered via social and pro-
fessional networks, cyclist and motorist web forums, and the
recognized by the Czech Government (2013), there is finan-
Civinet network (Civinet is an EU-supported group of local
cial support for cycling infrastructure but, as an emerging
authorities and organizations interested in sustainable mobil-
cycling country, there are spatial limitations where govern-
ity solutions for urban environments). Scales were designed
ment and municipalities often prefer low-cost solutions that
to measure attitudes and norms toward cycling and sharing
do not provide sufficient separation instead of more thor-
the road on the basis of a standardized interpretation of the
ough, yet time consuming, solutions. Cycling infrastructure
findings from previous qualitative studies (e.g., Joshi et al.,
is therefore lacking continuity and quality, which makes 2001; O’Connor & Brown, 2010; Daley & Rissel, 2011;
cyclists and motorists being forced to share the road in most Heesch et al., 2011; Lawson et al., 2013; Kaplan & Prato,
cities. It should be noted that the emerging cycling is notable 2016) that brought forward the perspective of both cyclists
in the Czech Republic from the average 1.5 bicycles per and drivers. The scales measured: (i) attitudes toward the
household and the 6.3 million bicycles in the country environmental, health, economic and transport benefits of
(Krautova & Librova, 2009), alongside a growing tradition of cycling; (ii) perceptions of cyclist behavior while sharing the
recreational and competitive cycling that is a major past road as an internal stimulus for positive reaction according
time for one third of the population (Vrtalova, 2009). As to the reciprocity principle; (iii) subjective social norms of
plans and investments are at the top of the interests of cycling tradition in the region; (iv) anxiety to share the road
policy-makers (Czech Government, 2013), the major issue is because of the risk of cycling in mixed traffic, the fragility
the low use of bicycles in the large cities that this study of cyclists and the negative emotions for fear related to
looks at from the perspective of understanding what allows a incidents. It should be noted that, in line with the existing
positive climate for cyclists and motorists to share the road literature, the focus is on the experience of sharing the road
and let cycling develop while the infrastructure is not as well as the perception of cycling and cyclists’ behavior
built yet. because cyclists are still a minority (Prati et al., 2017).
The remainder of the paper presents initially the ques- It should be noted that previous studies referred to
tionnaire and the model, then describes the sample and the “cyclists” and “drivers” as two distinct populations according
model estimates, and last discusses the results and draws to their preferred mode, habitual mode, assumed role with
conclusions from the study. respect to specific road incidents, or self-reported empathy
with a specific role (e.g., O’Connor & Brown, 2010; Heesch
et al., 2011; English & Salmon, 2016; Kaplan & Prato, 2016).
2. Methods In contrast, this study deconstructs the cyclist–motorist
2.1. Questionnaire design and data collection dichotomy to address the division and inequality it creates
because these two roles are not mutually exclusive. Evidence
As mentioned, Czech Republic is an emerging cycling exists that for the majority of the population the choice is
country where the rationale of this study applies perfectly practical and relates to functional and situational conditions.
because the infrastructure lags behind and cycling is Kuhnimhof et al. (2012) showed increasing multimodal
developing at a slower pace while cyclists and motorists choices among young adults in Germany and Great Britain
share the road. Creating a social climate where there is and commented that various transport modes can be used
mutual understanding requires the comprehension of the across the same period in parallel with the same or varying
factors behind the road sharing experience, and this study intensity, without the need to identify oneself as a “cyclist”
basis the behavioral framework on the TPB (Ajzen, 1991). or a “driver”. Sigurdardottir et al. (2014) showed that the car
According to the TPB, observed behavior is driven by is strongly correlated with high-order emotional needs and
intentions which are motivated by personal attitudes, social self-identity only for 12% of the population who identify as
4 S. KAPLAN ET AL.

car-enthusiasts, while 70% choose the car for practical rea- variables. The willingness to share the road was then
sons and 18% are car-sceptics that recognise the functional expressed for both cases as a function of the respondents’
value of the car but give a lower importance to it. This study socioeconomic characteristics, habitual travel behavior,
views mode choice as a behavioral trend that can be attitudes toward cycling and e-cycling, and lower and
described along a continuous spectrum, namely a person can higher-order needs from the Alderfer’s model (1969).
at a certain moment cycle or drive and hence empathise As the respondent was the unit of analysis, the dependent
with either perspective. This view of multimodality might variables were likely correlated and the error terms for the
seem trivial at a first glance, but the ability of road users to two models were assumed to capture common unobserved
identify and thus empathize with the perspective of other factors that were related to the same respondent.
road users has important implications regarding road Accordingly, the error terms were assumed to follow a
conflict mitigation and resolution and regarding the choice bivariate normal distribution and the model was specified as
of a proactive sharing solution versus aggressive or avoid- a bivariate-ordered probit model for calculating the probabil-
ance coping strategy. The questionnaire items were phrased ities of the willingness to share the road as a driver and as a
so that the respondents could refer to them from both the cyclist. It should be noted that the multivariate probit
cyclist and driver perspectives. modeling methodology has been previously used in the
The questionnaire items were accompanied by questions travel behavior and the traffic safety context (e.g., Emmerink
eliciting information about: (i) travel habits including mode et al., 1996; Ferdous et al., 2010; Dias et al., 2017; Wang
choice, cycling purpose, and weekly cycling duration; (ii) et al., 2018).
cycling experience in terms of cycling-friendly cities, cycling The specification of the bivariate ordered probit model
during childhood, cyclist–motorist traffic incidents, cycling considered both observable and latent variables related to
on various types of infrastructure from dedicated cycle paths the survey respondents. Observable variables included the
in nature to cycling in mixed traffic; (iii) socioeconomic sociodemographic characteristics of the respondents and
characteristics including age, gender, education, employment, their habitual travel behavior, while latent variables included
and family status; (iv) residential location including region, attitudes, norms, and difficulties of cyclists and motorists
city, location in the city with respect to the core, type, and sharing the road. A latent variable model was specified with
city size. measurement and structural equations, and then was jointly
For this exploratory study, the survey was advertised via estimated with the bivariate model to obtain a hybrid bivari-
social networks, web forums, and the Civinet network ate ordered probit model. This implies that the sociodemo-
because of budget limitations. The snowball recruitment graphic characteristics and the habitual travel behavior
reached over 1000 potential respondents, but of course the might have both a direct (as predictors of the two ordered
convenience sample required to be corrected with a weight- probit models) and an indirect (as predictors of the struc-
ing procedure. Information about the Czech population (e.g., tural equations of the latent variables) effect on the willing-
gender, age, education, family composition, residence city) ness to share the road as a cyclist and as a driver (Vij and
was collected from the OECD databased and the Czech Walker, 2016).
Republic statistics bureau, while additional information Firstly, exploratory factor analysis retrieved the latent
about the daily travel habits were collected from existing traits and contributed to the specification of the measure-
literature (Fromel et al., 2009; Bruha & Bruhova-Foltynova, ment equations of the latent variable model:
2014). Although not perfect, the weighting procedure bal- I n ¼ aX n þ tn ; (1)
anced the convenience sample to a degree of representability
where In is a vector of R indicators corresponding to the
of the population.
survey answers of respondent n, Xn is a vector of L latent
traits of respondent n, a is a vector of R parameters to
2.2. Hybrid bivariate ordered probit model be estimated, and tn is a vector of normally distributed
error terms with a mean vector of zeros and covariance
Given the questionnaire items and the observable individual matrix Wt.
characteristics, a hybrid model was formulated to test the Secondly, the latent traits were related to the observed
behavioral framework because of its ability to model simul- variables in the specification of the structural equations of
taneously latent endogenous constructs, their relations to the latent variable model:
observed exogenous variables, and their relations to the
willingness to share the road as drivers and cyclists. Weights X n ¼ cX n þ fn ; (2)
were calculated via Iterative Proportional Fitting from the where Xn is a vector of L latent traits of respondent n, Xn
marginal probabilities given the sample and the population is a vector of K observed characteristics of respondent n, c is
proportions for gender, age, residential arrangement, a (L  K) matrix of parameters to be estimated, and fn is a
residential location, education, job, and daily travel habit vector of normally distributed error terms with a mean
distribution. vector of zeros and covariance matrix W f .
The willingness to share the road as a driver and as a The bivariate ordered probit model assumed the existence
cyclist was measured on a 5-point Likert scale from of a partitioning of underlying continuous latent variables
“definitely not” to “definitely yes”, and hence two ordered that relate directly to the dependent variables (Dias et al.,
probit models were specified for the two dependent 2017). Having defined the indices p and q and the total
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 5

number of categories P and Q for the two dependent varia- distribution of the latent variables. The log-likelihood
bles, the two ordered probit models assumed the following function was then written as:
form: N X
X P X
Q

fn ¼ bf X n þ bf X n þ en where fn ¼ p if dp1 < fn logL ¼ xn In ðp; qÞPrðfn ¼ p; gn ¼ qÞ; (7)
(3) n¼1 p¼1 q¼1
< dp ; d0 ¼ 1; dP ¼ þ1;
where xn is the weight of observation n as calculated from
the Iterative Proportional Fitting procedure.
gn ¼ bg X n þ bg X n þ gn where gn ¼ q if vq1 < gn The log-likelihood function was coded in PythonBiogeme
(4)
< vq ; v0 ¼ 1; vQ ¼ þ1 2.5 (Bierlaire, 2016) and the maximization of the function
was performed by simulation via 1000 draws from the
The variables fn and gn are the observed categories of the
Modified Latin Hypercube Sampling algorithm (Hess
willingness to share the road as a driver and as a cyclist,
et al., 2006).
while the variables fn and gn indicate the latent propensity
of a respondent n to be willing to share the road as a driver
and as a cyclist (the larger are the values of fn and gn , 3. Results
the higher is the propensity). The vectors bf, bf, bg, and
3.1. Sample characteristics
bg contain parameters to be estimated, the thresholds
dp partition fn in P categories, and the thresholds The questionnaire collected 971 complete responses that
vq partition gn in Q categories. were 84.1% of the initial entries. Table 1 presents the socioe-
The random error terms en and gn of the two models were conomic characteristics for the respondents in the sample
assumed to be normally distributed and, without loss of gen- and the population according to the aforementioned sources:
erality and for identification purposes, the normal distribution the Iterative Proportional Fitting procedure calculated the
of the error terms was standardized (Dias et al., 2017) and a weights xn for each observation by considering the marginal
joint bivariate standard normal distribution was assumed to probabilities over the 10 variables.
capture their correlation. The estimated parameters include The sample required correction with the weighting
the elements of the vectors bf, bf, bg, and bg, the (P – 1) procedure because the recruitment via social media and
thresholds dp (– 1 < d1 <  < dp <  < dP–1 < þ 1) and the professional networks reached a higher number of cyclists
(Q – 1) thresholds vq (– 1 < v1 <  < vq <  < vQ–1 < þ 1). and a significantly higher proportion of men. Another factor
The parameters were estimated by maximizing the log- that required significant correction was the city dimension
likelihood function and defining a binary indicator In ðp; qÞ as most of the sample answered from Prague or Brno. Aside
that was equal to 1 if respondent n chose category p for the from cycling, the distribution of daily use of travel modes
first dependent variables and category q for the second was fairly in line with previous literature from the same
dependent variable, and 0 otherwise. region (Bruha & Bruhova-Foltynova, 2014).
The joint probability of respondent n having the willing- Table 2 presents the perceptions of the barriers to cycling
ness to share the road as a driver and as a cyclist was as well as the experience with (near) crashes as a driver or
expressed as: as a cyclist. Interestingly, conflicts with other road users
were the second most prevalent issue after a very personal
Prðfn ¼ p; gn ¼ qÞ ¼
one (sweating). From the supply side, respondents men-
¼ Prð dp1 < fn < dp and vq1 < gn < vq Þ ¼
tioned the fear of mixed traffic, lack of cycling paths and
¼ Prðdn;p1 < en < dn;p and cn;q1 < gn < cn;q Þ ¼ (5)
lack of end-trip cycling facilities as the three most common
¼ U2 ðdn;p ; cn;q ; hÞ  U2 ðdn;p1 ; cn;q ; hÞ
barriers. It should be noted that most cyclists among the
 U2 ðdn;p ; cn;q1 ; hÞ þ U2 ðdn;p1 ; cn;q1 ; hÞ ;
respondents had some experience in sharing the road, as
where dn,p ¼ dp – bf Xn þ bf Xn, cn,q ¼ vq – bg Xn þ bg Xn, 45% cycle in mixed traffic and 28% use cycle lanes. In order
U2 is the bivariate cumulative normal distribution function, to control for extreme perspectives, the questionnaire elicited
and h is a correlation parameter to be estimated. Given attitudes regarding the trade-offs between dedicated bicycle
that the model specification includes the latent variables infrastructure and parking offer. Most respondents preferred
Xn, the joint probability required integration over their cycle lanes for a scenario with a high volume of cyclists and
distribution (Walker and Ben-Akiva, 2002; Vij and Walker, parking places for a scenario with a low volume of cyclists,
2016): indicating that the respondents did not have extreme
views but have an awareness of the needs of both drivers
Pr ðf n ¼ p; g n ¼ qÞ
ð and cyclists.
  Table 3 shows the intentions to share the road as a cyclist
¼ U2 f n ¼ p; g n ¼ qX n ; X n ; bf ; bf ; bg ; bg ; dp ; vq ; h (6) and as a driver. Interestingly, most respondents would rather
X n
    avoid sharing the road in mixed traffic on major roads as
hI I n X n ; X n ; a; Wt hXn X n X n ; c; Wx dX n ; either driver or cyclist, but a larger proportion would agree
to share the road in mixed traffic on minor roads, thus
where U2 ðÞ is the bivariate probability in Equation (4), showing the importance of road hierarchy, traffic volume
hI ðÞ is the distribution of the indicators, and hXn ðÞ is the and speed in the decision to share the road.
6 S. KAPLAN ET AL.

Table 1. Sample characteristics.


Variable
Gender
Categories Male Female
Sample (%) 70.4 29.6
Population (%) 49.4 50.6
Age
Categories 15–24 years old 25–34 years old 35–44 years old 45–54 years old 55þ years old
Sample (%) 14.1 37.6 26.5 14.2 14.2
Population (%) 19.9 24.5 21.6 16.4 17.6
Living in
Categories Household with adults Household with children Other
Sample (%) 29.4 40.4 30.2
Population (%) 32.1 41.6 26.3
Employment status
Categories Only study Work and study Only work Neither work nor study
Sample (%) 7.5 14.8 74.5 3.2
Population (%) 5.6 11.0 57.7 25.7
Education
Categories Elementary High High with graduation Higher specialized University
Sample (%) 1.2 5.4 30.4 3.5 59.5
Population (%) 14.2 22.2 39.5 4.1 20.0
City size
Categories Below 10,000 10,000-30,000 30,000-50,000 50,000-100,000 Over 100,000
Sample (%) 10.4 6.9 4.2 8.2 65.3
Population (%) 49.2 8.2 12.2 10.5 19.9
Car use
Categories Daily 3–4 times per week 1–2 times per week Sometimes Rarely or never
Sample (%) 29.4 14.4 18.0 19.7 18.5
Population (%) 27.0 17.0 16.0 17.0 24.0
Transit use
Categories Daily 3–4 times per week 1–2 times per week Sometimes Rarely or never
Sample (%) 22.0 11.7 12.7 25.3 28.3
Population (%) 9.0 12.0 11.0 23.0 45.0
Bicycle use
Categories Daily 3–4 times per week 1–2 times per week Sometimes Rarely or never
Sample (%) 15.2 19.1 16.3 24.2 25.2
Population (%) 1.0 2.0 6.0 23.0 68.0
Walk
Categories Daily 3–4 times per week 1–2 times per week Sometimes Rarely or never
Sample (%) 49.5 12.9 10.9 18.2 8.5
Population (%) 48.0 16.0 11.0 10.0 14.0

Table 2. Cycling experience and difficulties with road sharing.


Item Categories (%)
INFRASTRUCTURE USE Countryside paths Separated city Cycle lanes Mixed traffic Mixed traffic Pedestrian areas
FOR CYCLING and cycle paths cycle paths in a city in a city outside a city and sidewalks Do not cycle
65.5 34.2 27.5 45.9 39.3 17.5 12.1
PREFERENCES WITH Prefer separated cycling Do not have a preference Prefer more parking spots
LOW infrastructure to more to separated cycling
CYCLIST VOLUME parking spots infrastructure
20.9 21.9 57.2
PREFERENCES WITH Prefer separated cycling Do not have a preference Prefer more parking spots
HIGH infrastructure to more to separated cycling
CYCLIST VOLUME parking spots infrastructure
55.3 15.6 29.1
PERCEIVED ISSUES Do not like Lack of cycle Fear of Lack of trip Distance to Hilly terrain Infrastructure
WITH BICYCLE cycling paths/lanes mixed traffic end facilities work or school design
COMMUTING (P. 1)
7.7 38.4 36.1 35.4 24.0 22.8 33.1
PERCEIVED ISSUES Lack of Social norm Physical Sweating Conflicts with No access No problem
WITH BICYCLE maintenance and peer condition drivers/pedestrians to bicycle
COMMUTING (P. 2) skills pressure
3.7 4.5 8.2 37.7 17.4 3.4 13.0
(NEAR) Yes, as Yes, as No crash
CRASH EXPERIENCE a cyclist a driver experience
32.5 16.1 56.4

Before model estimation, exploratory factor analysis (via reliability with Cronbach’s Alpha equal to 0.815, and the
Principal Axis Factoring with Oblimin rotation) verified that sample had very good adequacy to factor analysis with
the four hypothesized factors emerged from the data. Initial Kaiser–Meyer–Olkin index equal to 0.936. Further analysis
analysis showed that the items had very good internal confirmed the existence of four factors on the basis of the
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 7

Table 3. Intentions to share the road.


Willingness to share the road Definitely not Most likely not Maybe Most likely yes Definitely yes
AS A DRIVER ON A MAJOR ROAD 30.9 28.7 11.3 15.3 13.7
AS A CYCLIST ON A MAJOR ROAD 41.2 29.5 10.0 10.3 9.1
AS A DRIVER ON A MINOR ROAD 5.5 8.7 10.9 28.5 46.4
AS A CYCLIST ON A MINOR ROAD 7.8 8.9 12.8 28.8 41.7

Table 4. Rotated factor matrix.


Factors
Item Factor 1 Factor 2 Factor 3 Factor 4
Riding a bicycle is healthy 0.728 0.404 –0.105 0.043
Riding a bicycle is good for the environment 0.788 0.482 –0.019 –0.002
Riding a bicycle reduces congestions 0.789 0.567 –0.158 –0.035
Bicycle is an affordable transport mode 0.635 0.427 –0.076 0.071
Riding a bicycle is fun 0.660 0.471 –0.226 –0.011
Cycling is an important sport/recreational activity 0.600 0.352 –0.113 0.068
Bicycle is a pragmatic mean of transport 0.731 0.574 –0.148 0.036
Cyclists should have their own infrastructure 0.219 0.041 0.476 0.051
Bicycles are too slow/fragile to be used on the road –0.378 –0.491 0.643 0.059
Cyclists have the same right to use the road as cars 0.585 0.610 –0.426 –0.086
Cyclists on the road slow the traffic down (R) 0.436 0.604 –0.430 –0.107
Cyclists and cars can share the same roads (R) –0.518 –0.560 0.584 –0.024
Cyclists usually make way for drivers 0.484 0.676 –0.227 –0.025
Cyclists do not respect traffic lights (R) 0.470 0.656 –0.165 0.065
Cyclists drive in wrong direction on one-way streets (R) 0.242 0.493 –0.099 0.031
Cyclists usually signal their left/right turns with their hands 0.454 0.661 –0.126 0.158
Cyclists usually avoid cycling in the middle of the lane 0.369 0.462 –0.104 0.067
Most people want to share the roads without conflicts 0.181 0.223 –0.033 0.360
Most of the cyclists try not to block the traffic 0.542 0.798 –0.308 0.095
Most of the people do not mind the cyclists on the road 0.270 0.491 –0.372 0.315
Cyclists are usually respectful toward the drivers 0.495 0.778 –0.299 0.128
Drivers drive carefully around cyclists –0.269 –0.168 0.035 0.417
Most of the people can ride a bicycle 0.377 0.395 –0.056 0.359
There is a strong cycling tradition 0.307 0.227 –0.033 0.324
Most of the children are taught to ride a bicycle 0.157 0.092 0.034 0.421
I know a lot of people who cycle 0.473 0.360 –0.043 0.278
There is a lot of recreational cyclists 0.362 0.218 –0.065 0.261
As a driver I am/would be afraid to drive around cyclists –0.154 –0.347 0.661 0.050
I am afraid of cyclists-motorist accidents –0.044 –0.228 0.578 0.002
Riding a bicycle in mixed traffic is hazardous –0.322 –0.470 0.804 –0.033
As a driver I am/would be afraid of hitting a cyclist –0.117 –0.334 0.709 0.059
I am afraid to cycle in mixed traffic –0.141 –0.182 0.796 –0.122
I feel anxious sharing the road –0.035 –0.067 0.731 –0.160
Drivers are used to cyclists on the road –0.032 0.017 –0.096 0.450
If there were more cyclists on the roads, drivers would get used to them 0.542 0.630 –0.460 –0.045
There are fewer disputes between drivers and cyclists in cities with many cyclists 0.520 0.582 –0.300 0.017
Cyclists do unexpected manoeuvres because they lack experience with driving or riding on the road (R) 0.266 0.551 –0.269 0.084
Cyclists do not know how to behave on the road because they do not have a driving license (R) 0.341 0.627 –0.280 –0.002
Drivers forget about cyclists on the road (R) 0.463 0.363 –0.038 –0.370
(R) – Reversed coding in the case of negatively phrased items; bold – highest factor loading for each item.

interpretability of the solution and the parsimony of the finding the best model fit while offering the highest model
model: the rotated factor matrix of the solution is presented interpretability, and having parameters associated with
in Table 4. The four factors were named “positive attitudes observed and latent variables relevant to the willingness to
toward cycling” (factor 1), “positive attitudes toward cyclist share the road significant at the 0.10 level.
behavior” (factor 2), “anxiety to share the road” (factor 3), The best model is presented in the following tables:
and “social norms toward cycling” (factor 4). The relations Table 5 presents the estimates of the measurement
between items and factors served as input to for the equations, Table 6 illustrates the structural equations relating
measurement equations in the hybrid bivariate ordered the four latent traits to the observed characteristics of
probit model. the respondents, and Table 7 shows the estimates of the
bivariate ordered probit model expressing the willingness to
share the road on major roads as a driver and as a cyclist as
3.2. Model estimation results
a function of both observed and latent variables. The log-
Following the exploratory factor analysis, the best model likelihood of the presented model is equal to –619.40, the
combined the specification of the measurement equations adjusted McFadden rho-square is equal to 0.532, and likeli-
according to the factor analysis results with the best specifi- hood ratio tests reveal that the model is to be preferred to a
cation of the structural and the ordered probit equations. bivariate model without latent variables (LRT ¼ 43.56, df ¼ 7,
The latter was achieved via an iterative procedure aimed at p ¼ 0.000) as well as two separate models for the willingness
8 S. KAPLAN ET AL.

Table 5. Measurement equations of the latent constructs for the hybrid bivariate ordered probit model.
Estimated t-stat
Positive attitudes toward cycling
Riding a bicycle is healthy 1.000 –
Riding a bicycle is good for the environment 1.067 24.31
Riding a bicycle reduces congestions 1.112 25.33
Riding a bicycle is fun 0.833 21.09
Cycling is an affordable transport mode 0.930 22.30
Cycling is an important sport/recreational activity 0.855 20.51
Bicycle is a pragmatic mean of transport 0.967 21.50
Positive attitudes toward cyclist behavior
Cyclists have the same right to use the road as cars 1.000 –
Cyclists on the road slow the traffic down –0.958 –16.48
Cyclists usually make way for drivers 0.838 15.60
Cyclists do not respect traffic lights –0.914 –16.66
Cyclists drive in the wrong direction on one-way streets –0.739 –13.76
Cyclists usually signal their left/right turns with their hands 0.797 14.53
Most of the cyclists try not to block the traffic 1.200 18.85
Most of the people do not mind the cyclists on the road 0.902 15.81
Cyclists are usually respectful toward the drivers 1.121 17.33
If there were more cyclists on the roads, drivers would get used to them 1.173 19.09
There are fewer disputes between drivers and cyclists in cities with many cyclists 0.921 16.14
Cyclists do unexpected manoeuvres because they lack experience with driving or riding on the road –0.817 –14.61
Cyclists do not know how to behave on the road because they do not have a driving license –0.825 –14.45
Anxiety to share the road
Cyclists should have their own infrastructure 1.000 –
Bicycles are too slow/fragile to be used on the road 2.916 7.13
Cyclists and cars can share the same roads –3.017 –6.71
As a driver I am/would be afraid to drive around cyclists 2.771 6.96
I am afraid of cyclists-motorist accidents 2.414 7.17
Riding a bicycle in mixed traffic is hazardous 3.595 7.19
As a driver I am/would be afraid of hitting a cyclist 2.957 7.64
I am afraid to cycle in mixed traffic 2.866 7.64
I feel anxious sharing the road 2.400 7.01
Social norms toward cycling
Most people want to share the roads without conflicts 1.000 –
Most of the people can ride a bicycle 2.223 8.04
There is a strong cycling tradition 1.612 7.70
Most of the children are taught to ride a bicycle 2.020 7.80
I know a lot of people who cycle 2.212 7.84
There are a lot of recreational cyclists 1.980 7.88
The estimates of the constants and the covariance matrix elements are omitted for brevity.

to share the road as a driver or as a cyclist (LRT ¼ 105.49, looking at travellers as multimodal, rather than fomenting
df ¼ 1, p ¼ 0.000). The correlation between the error terms the dichotomy between drivers and cyclists, is likely the
shows also that the bivariate model is the preferred specifica- correct way as experience with both contributes to a positive
tion as it is equal to 0.652 and it is significantly different view of cycling. Very frequent of using public transport is
from zero at the 0.01 confidence level. related only to attitudes toward cyclists’ behavior, and the
sign of the relation is negative, perhaps because one can be
seen as substitute for the other.
3.3. The relation between TPB constructs and individual
Having experienced cycling during childhood reinforces
characteristics
social norms toward cycling, while having experienced
Travel habits are significantly related to all the TPB traits. Very (near) crashes has a negative relation to positive attitudes
frequent use of the bicycle for either commuting or recreational toward cyclist behavior and a positive relation to the anxiety
purposes is positively associated with positive attitudes toward of road sharing if the crash was experienced as a driver. On
both cycling and cyclists’ behavior and with the formation of a positive note, previous crash experience does not associate
social norms toward cycling. Conversely, the relation is nega- adversely with positive attitudes and social norms toward
tive with the formation of anxiety to share the road. These cycling.
findings are in line with the observation by Kaplan et al. The perceived barriers to cycling are also related to the TPB
(2015) that both utilitarian and recreational cycling relate to constructs. Personal barriers to cycling (e.g., distance, hilliness,
favourable cycling attitudes and norms. Interestingly, cycling fitness, enjoyment) are related to a lesser agreement with
for commuting purposes is related to the anxiety in sharing the cycling positive attitudes and social norms, while infrastructure
road unlike cycling for recreation, possibly because the latter is barriers to cycling (e.g., poor infrastructure, lack of cycle lanes
more flexible in terms of route choices to avoid traffic. and paths, conflicts with road users) are linked to the anxiety
Interestingly, very frequent use of the car is also related of road sharing. Interestingly, both barrier types are associated
positively to having positive attitudes toward cycling and negatively with the perception of cyclist behavior.
cyclists’ behavior, and negatively to feeling anxiety to share The socioeconomic characteristics play a role as well.
the road. These findings suggest that the hypothesis of Both living in small and large cities (with respect to an
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 9

Table 6. Structural equations of the latent constructs for the hybrid bivariate ordered probit model.
Est. t-stat
Positive attitudes toward cycling
Constant 0.401 5.93
Male –0.340 –4.33
Age lower than 34 years old 0.307 2.07
Age between 35 and 54 years old 0.219 2.09
Living in a household with children 0.172 1.99
Living in a city with less than 10,000 inhabitants 0.276 3.20
Living in a city with more than 100,000 inhabitants 0.209 1.76
Very frequent car use 0.091 2.36
Very frequent bicycle use 0.202 2.16
Using bicycle for commuting 0.375 3.21
Using bicycle for recreation or sport 0.169 2.18
Distance from work/school/activities is perceived as a barrier to cycling –0.188 –2.32
Lack of cycle lanes and paths is perceived as a barrier to cycling –0.207 –2.75
Lack of interest or enjoyment is perceived as a barrier to cycling –0.450 –3.25
Peer pressure and social norms are perceived as a barrier to cycling –0.311 –2.18
Positive attitudes toward cyclist behavior
Constant 0.346 5.97
Male –0.172 –2.77
Age under 34 years old –0.176 –2.52
Living in a household with children 0.137 2.13
Very frequent car use 0.143 2.12
Very frequent transit use –0.163 –2.62
Very frequent bicycle use 0.145 4.03
Using bicycle for commuting 0.369 2.14
Using bicycle for recreation or sport 0.641 3.38
Experience with (near) crash as a cyclist –0.164 –2.78
Experience with (near) crash as a driver –0.576 –7.66
Hilly terrain is perceived as a barrier to cycling –0.206 –3.04
Lack of cycle lanes and paths is perceived as a barrier to cycling –0.142 –2.33
Lack of interest or enjoyment is perceived as a barrier to cycling –0.413 –3.32
Anxiety of cycling in mixed traffic is perceived as a barrier to cycling –0.159 –2.47
Conflicts with drivers and pedestrians are perceived as a barrier to cycling –0.207 –3.16
Positive attitudes toward cycling 0.445 8.49
Social norms toward cycling 2.110 4.48
Anxiety to share the road
Constant 0.053 2.02
Male –0.081 –3.23
Age under 34 years old –0.200 –2.51
Living in a household with children 0.052 2.09
Very frequent car use –0.061 –2.34
Very frequent bicycle use –0.195 –4.15
Using bicycle for commuting –0.068 –1.96
Experience with (near) crash as a driver 0.112 4.12
Distance from work/school/activities is perceived as a barrier to cycling 0.050 1.98
Lack of cycle lanes and paths is perceived as a barrier to cycling 0.059 2.47
Poor infrastructure design is perceived as a barrier to cycling 0.073 3.05
Anxiety of cycling in mixed traffic is perceived as a barrier to cycling 0.245 7.24
Conflicts with drivers and pedestrians are perceived as a barrier to cycling 0.110 4.02
With a low number of cyclists, I would prefer increased parking spaces –0.068 –2.49
With a high number of cyclists, I would prefer dedicated cycling infrastructure –0.061 –2.41
Positive attitudes toward cyclist behavior –1.390 –4.22
Social norms toward cycling –2.606 –2.39
Social norms toward cycling
Constant 0.117 2.56
Male –0.115 –3.10
Living in a city with less than 10,000 inhabitants 0.176 2.56
Living in a city with more than 100,000 inhabitants –0.128 –2.35
Very frequent bicycle use 0.147 2.92
Using bicycle for commuting or recreation 0.142 2.78
Experience with cycling as a child 0.255 3.49
Distance from work/school/activities is perceived as a barrier to cycling –0.094 –2.47
Lack of cycle lanes and paths is perceived as a barrier to cycling –0.106 –1.98
Hilly terrain is perceived as a barrier to cycling –0.076 –2.06
Conflicts with drivers and pedestrians are perceived as a barrier to cycling –0.115 –2.02
With a high number of cyclists, I would prefer dedicated cycling infrastructure 0.082 2.27

average size between 10,000 and 100,000 inhabitants) relates household with children is related to positive attitudes
to positive attitudes toward cycling, while the relation is toward cycling and cyclists’ behavior, but also higher anxiety
positive for the former and negative for the latter when it of sharing the road.
comes to the formation of social norms. In other words, From a gender perspective, women have on the one hand
having positive attitudes toward cycling in a larger city does stronger positive attitudes toward cycling and cyclists’ behav-
not translate in having norms about cycling. Living in a ior, as well as higher perception of the existence of cycling
10 S. KAPLAN ET AL.

Table 7. Estimates of the hybrid bivariate ordered probit model. and lower anxiety to share the road, while positive attitudes
Estimate t-stat toward cycling in general are related to the willingness to
Willingness to share the road as a driver share the road as a driver and as a cyclist both directly and
Living in a household with children –0.220 –1.83
Living in a city with more than 100,000 inhabitants –0.212 –1.84
indirectly (via the relation with positive attitudes toward cyc-
Using the bicycle for commuting or recreation 0.270 1.92 list behavior). Notably, three latent traits have a comparable
Very frequent bicycle use 0.179 2.20 direct effect on the willingness to share the road as a driver
Positive attitudes toward cycling 0.348 5.56
Positive attitudes toward cyclist behavior 0.777 14.16 and as a cyclist, while social norms have a positive relation
Anxiety to share the road –0.897 –14.91 to the willingness to share the road as a driver. Secondly,
Social norms toward cycling 0.097 1.86 the relation is not merely a direct one either between the
d1 –1.008 –7.29
d2 0.103 0.75 observed variables and the willingness to share the road.
d3 0.585 4.24 While using the bicycle frequently and for both commuting
d4 1.291 9.33 and recreation seems to increase the intention of sharing the
Willingness to share the road as a cyclist
Living in a household with children –0.263 –2.15 road from both perspectives, the opposite is observed for
Using the bicycle for commuting or recreation 0.371 2.96 living in a household with children that appears to reduce
Very frequent car use 0.217 2.14 the desire of sharing the road from both perspectives.
Very frequent bicycle use 0.406 2.16
Experience with (near) crash as a cyclist –0.195 2.00 Differences exist in terms of living in a large city, which
Experience with (near) crash as a driver –0.205 1.87 relates negatively only to the willingness to share the road as
Positive attitudes toward cycling 0.440 7.13
Positive attitudes toward cyclist behavior 0.422 7.44
a driver, and using the car frequently, which relates posi-
Anxiety to share the road –0.787 –13.09 tively only to the intention to share the road as a cyclist.
v1 –1.994 –14.31 Interestingly, being a frequent bicycle and car user relates
v2 –0.802 –5.77
v3 –0.272 –1.96
positively to the willingness to share the road as a cyclist,
v4 0.430 3.09 again suggesting that the perspective of shared responsibility
Correlation of the error terms as multimodal travellers appears as the one leading to the
H 0.652 6.66
desire to share the road (and, in turn, to more cycling).
When looking at the number of common variables (both
social norms. Women appear also, on the other hand, more observed and latent) that are significantly related to both
anxious than men to share the road. These findings agree equations expressing the willingness to share the road, and
with existing literature on favourable attitudes toward looking at the strong correlation between the error terms, it
cycling being stronger for women (Kaplan et al., 2015), male is clear that there is not a dichotomy in wanting to share
cyclists being less afraid of traffic and less prone to cycling the road as either driver or cyclist. Clearly, there are both
avoidance while having more propensity to cycle in mixed observed and latent factors that affect the willingness to
traffic (Chataway et al., 2014), and women being more likely share the road while the respondents consider only the cir-
to report safety concerns, driver aggression and cycling cumstances to determine whether they are drivers or cyclists.
avoidance (Garrard et al., 2008).
An interesting aspect is that there exist relations between
4. Discussion and conclusions
the latent constructs themselves, namely positive attitudes
and social norms toward cycling in general are related to This study brings forward attitudes, norms and difficulties
positive attitudes toward cyclist behavior, while the latter underlying the willingness to share the road as a driver and
and the social norms are associated with lower anxiety to a cyclist while deviating from the harsh rhetoric regarding
share the road. These findings suggest a mediating effect of “road wars” between cyclists and drivers to promote instead
some latent constructs when considering the relations with a view of shared responsibility. This study takes the multi-
the willingness to share the road. modality perspective and recognizes that the same road user
can relate to both perspectives because being a driver or a
cyclist is related to circumstances. The chosen Czech case-
3.4. The relation between TPB constructs and willingness
study represents an emerging cycling region characterised by
to share the road
growing cycling enthusiasm and lack of cycling infrastruc-
As aforementioned, most respondents were willing to share ture that brings to sharing the road between cyclists and
the road on minor roads as both drivers and cyclists, and motorists. Nevertheless, the behavioral framework and
hence the model focused on the willingness to share the the model follow the TPB as a general human behavior
space on major roads. A couple of considerations apply to paradigm.
the estimates of the hybrid bivariate ordered probit model The results confirm with an advanced weighted model
for both parts related to the driver and the cyclist perspec- that psychological barriers are paramount in forming posi-
tive on sharing the road. tive attitudes toward cycling and cyclists’ behavior, forming
Firstly, the relation is not merely a direct one between social norms toward cycling, mitigating anxiety in mixed
the TPB traitsand the willingness to share the road, as the traffic, and improving the experience of sharing the road.
aforementioned structural equations show that some con- While the anxiety of road sharing is mainly associated with
structs act as mediators. Social norms toward cycling are in perceived infrastructure gaps and lacunas, pro-cycling
fact related to both positive attitudes toward cyclist behavior attitudes and norms are related to personal barriers that are
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 11

related to self-perceptions (e.g., abilities, competencies). The that allow continuity along axes with major commuting
anxiety to share the road is a strong psychological barrier flows. Multimodality would include widespread, well
that surpasses the positive effect of positive attitudes and maintained, and easy to use bike-sharing systems, promoting
norms, as shown by the stronger relation between anxiety the integration of bicycle and public transport as well as
and willingness to share the road from both the driver’s and park-and-cycle in the inner city, and offering employer-
the cyclist’s perspective. based incentives to sustainable transport. Campaigns could
The results show the importance of the social climate in focus on shared responsibility with an emphasis on the
improving the experience of sharing the road. Both the anx- social inclusion of all road users, empathy, kindness, conflict
iety and the willingness to share the road can be bettered by mitigation and awareness that every road user is a driver, a
improving positive attitudes toward cyclists’ behavior, miti- cyclist and a pedestrian at the same time. Campaigns
gating actual conflicts among drivers and cyclists, and form- increasing visibility such as cycling marathons, bike-to-work
ing positive attitudes and social norms. Another important initiatives and bike-to-school programs should aim at
result concerns social inequity, as women reported stronger generating an inclusive ambience where everyone can join
agreement with statements related to the anxiety of sharing and cycle, creating a sense of shared vision, togetherness and
the road, and thus are less willing to cycle in mixed traffic. social inclusion. An example is the “Denmark cycling togeth-
The results support multifaceted solutions to the problem er” approach that includes: (i) design of attractive and pleas-
of encouraging the willingness to share the road in urban ant bicycle infrastructure; (ii) extensive cycling infrastructure
environments. Specifically, the results support the idea that along main corridors with prioritised accessibility to main
providing dedicated cycling infrastructure is important, but city attractions; (iii) the “bike to work” and “all kids bike”
its implementation needs to be part of a “safe system” campaigns that contribute to togetherness through termin-
approach, namely a holistic approach addressing individual
ology and gamification; (iv) the view of the bicycle as an
and social barriers by providing mutually synergetic social,
integral part of the transport system alongside public trans-
institutional, infrastructure, and policy solutions. Moreover,
port, cars, and taxis; (v) soft and compassionate communica-
the “safe system” approach should be part of a wider sus-
tion promoting a shared vision of sustainable mobility.
tainable mobility vision, because the results show that travel
Limitations should be acknowledged in that the sample
habits, perceived personal barriers, regional effects, and
was recruited via word of mouth and was not representative
infrastructure barriers affect the experience of sharing
of the Czech population. However, weights were used in the
the road.
model to make the sample reflect the population and hence
The results support the idea that safety is a shared
the findings are informative because of the weighting
responsibility as the willingness to share the road while driv-
ing or cycling is correlated and both are driven by the same procedure and the rigorous econometric procedure that
underlying factors, led by the anxiety of sharing the road, produced unbiased parameter estimates (Wooldridge, 2009).
but also by the frequency with which respondents cycle and Moreover, further research could investigate the role of pol-
drive. When cycling or driving in mixed traffic, road users lution in the willingness to share the road in light of recent
share the same anxiety, and the aggressive interactions or findings about bicycle lanes at least marginally reducing the
cycling avoidance reported in previous studies are possibly exposure to pollutants (Hatzopoulou et al., 2013).
the result of ill-coping mechanisms with the feeling of anx-
iety. This study shows that the anxiety related to driving in Acknowledgements
mixed traffic is equally important to the anxiety of cycling
in mixed traffic and both should be considered seriously. The authors would like to acknowledge the contribution of two
anonymous reviewers who commented on earlier versions of
The idea of shared responsibility, in this case, concerns also
the manuscript.
conflicts between cyclists and drivers and the perception of
cyclists’ behavior. Conflicts between road users and illegal
manoeuvres not only are dangerous, but also increase the References
overall tension. Considerate behavior mitigates the tension
Ajzen, I. (1991). Theory of planned behaviour. Organizational
and thus indirectly contributes to the willingness of all road Behaviour and Human Decision Processes, 50, 179–211.
users to share the road in mixed traffic. The shared respon- Alderfer, C. P. (1969). An empirical test of a new theory of human
sibility approach is also manifested through the perceived needs. Organization Behavior and Human Performance, 4, 142–175.
importance of cycling lanes for regulating cycling in mixed Bassett, D. R., Pucher, J., Buehler, R., Thompson, D. L., & Crouter,
traffic. S. E. (2008). Walking, cycling, and obesity rates in Europe, North
America, and Australia. Journal of Physical Activity and Health, 5,
The results suggest that emerging cycling regions could 795–814.
become cycling-friendly and that sustainable mobility Bierlaire, M. (2016). PythonBiogeme: ia short introduction. Report
solutions allowing people to overcome their psychological TRANSP-OR 160706, Series on Biogeme. Transport and Mobility
barriers to cycling could increase bicycle mode shares. The Laboratory, School of Architecture, Civil and Environmental
findings from this study suggest that proactive policy direc- Engineering, Ecole Polytechnique Federale de Lausanne, Switzerland.
Boufous, S., de Rome, L., Senserrick, T., & Ivers, R. (2012). Risk factors
tions could aim at measures and campaigns promoting
for severe injury in cyclists involved in traffic crashes in Victoria,
multimodal and sustainable transport solutions as an integral Australia. Accident Analysis and Prevention, 49, 404–409.
part of the trip chain. Infrastructure solutions could include Bruha, J., & Bruhova-Foltynova, H. (2014). Active travel in Prague: The
well-designed bicycle paths along urban roads in a network role of age, gender and education in a heterogeneous population.
12 S. KAPLAN ET AL.

Presented at the 2014 Joint Cluster Meeting “Evolving Perspectives Joshi, M. S., Senior, V., & Smith, G. P. (2001). A diary study of the risk
on Sustainability and Security in Transport”, Beer Sheva, Israel. perceptions of road users. Health, Risk & Society, 3, 261–279.
Chataway, E. S., Kaplan, S., Nielsen, T. A. S., & Prato, C. G. (2014). Kaplan, S., Manca, F., Nielsen, T. A. S., & Prato, C. G. (2015).
Safety perceptions and reported behavior related to cycling in Intentions to use bike-sharing for holiday cycling: An application of
mixed traffic: A comparison between Brisbane and Copenhagen. the Theory of Planned Behavior. Tourism Management, 47, 34–46.
Transportation Research Part F: Traffic Psychology and Behaviour, Kaplan, S., & Prato, C. G. (2015). A spatial analysis of land use and
23, 32–43. network effects on frequency and severity of cyclist-motorist crashes
Chaurand, N., & Delhomme, P. (2013). Cyclists and drivers in road in the Copenhagen region. Traffic Injury Prevention, 16, 724–731.
interactions: A comparison of perceived crash risk. Accident; Kaplan, S., & Prato, C. G. (2016). “Them or Us”: Perceptions,
Analysis and Prevention, 50, 1176–1184. cognitions, emotions, and overt behavior associated with cyclists and
Cycling Embassy of Denmark (2012). Collection of cycle concepts: A motorists sharing the road. International Journal of Sustainable
revision and update of the collection of cycle concepts published by the Transportation, 10, 193–200.
Danish Road Directorate in 2000. Retrieved from http://www.cycling- Kaplan, S., Vavatsoulas, K., & Prato, C. G. (2014). Aggravating and
embassy.dk/2013/08/01/cycle-concepts2012/, Rounborg Grafiske Hus, mitigating factors associated with cyclist injury severity in Denmark.
Holstebro, Denmark. Journal of Safety Research, 50, 75–82.
Czech Government (2013). National strategy for development of cycling Krautova, Z., & Librova, H. (2009). Household consumption and the
transport in the Czech Republic for the years 2013–2020 (in Czech). individualization process from an environmental perspective (in
Government of the Czech Republic, Prague, Czech Republic. Czech). Faculty of Social Studies, Masaryk University, Brno, Czech
Daley, M., & Rissel, C. (2011). Perspectives and images of cycling as a Republic.
barrier or facilitator of cycling. Transport Policy, 18, 211–216. Krizek, K. J., & Roland, R. W. (2005). What is at the end of the road?
Daley, M., Rissel, C., & Lloyd, B. (2007). All dressed up and nowhere Understanding discontinuities of on-street bicycle lanes in urban set-
to go? A qualitative research study of the barriers and enablers to tings. Transportation Research Part D: Transport and Environment,
cycling in inner Sydney. Road & Transport Research, 16, 42–52. 10, 55–68.
Dias, F. F., Lavieri, P. S., Garikapati, V. M., Astroza, S., Pendyala, Kuhnimhof, T., Armoogum, J., Buehler, R., Dargay, J., Denstadli, J. M.,
R. M., & Bhat, C. R. (2017). A behavioral choice model of the use & Yamamoto, T. (2012). Men shape a downward trend in car use
of car-sharing and ride-sourcing services. Transportation, 44, among young adults: Evidence from six industrialized countries.
1307–1323. Transport Reviews, 32, 761–779.
Doorley, R., Pakrashi, V., & Ghosh, B. (2015). Quantifying the health Lawson, A. R., Ghosh, B., & Pakrashi, V. (2015). Quantifying the
impacts of active travel: Assessment of methodologies. Transport perceived safety of cyclists in Dublin. Proceedings of the Institution
Reviews, 35, 559–524. of Civil Engineers – Transport, 168, 290–299.
Emmerink, R. H., Nijkamp, P., Rietveld, P., & van Ommeren, J. N. Lawson, A. R., Pakrashi, V., Ghosh, B., & Szeto, W. Y. (2013).
(1996). Variable message signs and radio traffic information: An Perception of safety of cyclists in Dublin City. Accident; Analysis
integrated empirical analysis of drivers’ route choice behaviour. and Prevention, 50, 499–511.
Transportation Research Part A: Policy and Practice, 30, 135–153. Lugo, A. E. (2013). CicLAvia and human infrastructure in Los Angeles:
English, P., & Salmon, P. (2016). New laws, road wars, courtesy and Ethnographic experiments in equitable bike planning. Journal of
animosity: Cycling safety in Queensland newspapers. Safety Science, Transport Geography, 30, 202–207.
89, 256–262. Marques, R., Hernandez-Herrador, V., Calvo-Salazar, M., &
Ferdous, N., Eluru, N., Bhat, C. R., & Meloni, I. (2010). A multivariate Garcıa-Cebrian, J. A. (2015). How infrastructure can promote
ordered-response model system for adults’ weekday activity episode cycling in cities: Lessons from Seville. Research in Transportation
generation by activity purpose and social context. Transportation Economics, 53, 31–44.
Research Part B: Methodological, 44, 922–943. Meng, M., Koh, P. P., Wong, Y. D., & Zhong, Y. H. (2014). Influences
Fishman, E., Washington, S., & Haworth, N. (2012). Barriers and facili- of urban characteristics on cycling: Experiences of four cities.
tators to public bicycle scheme use: A qualitative approach. Sustainable Cities and Society, 13, 78–88.
Transportation Research Part F: Traffic Psychology and Behaviour, Mu~ noz, B., Monzon, A., & Lois, D. (2013). Cycling habits and other
15, 686–698. psychological variables affecting commuting by bicycle in Madrid,
Fromel, K., Mitas, J., & Kerr, J. (2009). The associations between active Spain. Transportation Research Record: Journal of the Transportation
lifestyle, the size of a community and SES of the adult population in Research Board, 2382, 1–9.
the Czech Republic. Health & Place, 15, 447–454. Murphy, E., & Usher, J. (2015). The role of bicycle-sharing in the city:
Garrard, J., Rose, G., & Lo, S. K. (2008). Promoting transportation Analysis of the Irish experience. International Journal of Sustainable
cycling for women: The role of bicycle infrastructure. Preventive Transportation, 9, 116–125.
Medicine, 46, 55–59. Oakil, A. T. M., Ettema, D., Arentze, T., & Timmermans, H. (2016).
Gatersleben, B., & Appleton, K. M. (2007). Contemplating cycling to Bicycle commuting in the Netherlands: An analysis of modal shift
work: Attitudes and perceptions in different stages of change. and its dependence on life cycle and mobility events. International
Transportation Research Part A: Policy and Practice, 41, 302–312. Journal of Sustainable Transportation, 10, 376–384.
Hatzopoulou, M., Weichenthal, S., Dugum, H., Pickett, G., Miranda- O’Connor, J. P., & Brown, T. D. (2010). Riding with the sharks: Serious
Moreno, L., Kulka, R., … Goldberg, M. (2013). The impact of traffic leisure cyclist’s perceptions of sharing the road with motorists.
volume, composition, and road geometry on personal air pollution Journal of Science and Medicine in Sport, 13, 53–58.
exposures among cyclists in Montreal, Canada. Journal of Exposure Oke, O., Bhalla, K., Love, D. C., & Siddiqui, S. (2015). Tracking global
Science & Environmental Epidemiology, 23, 46–51. bicycle ownership patterns. Journal of Transport & Health, 2, 490–501.
Heesch, K. C., Sahlqvist, S., & Garrard, J. (2011). Cyclists’ experiences Parkin, J., Wardman, M., & Page, M. (2007). Models of perceived
of harassment from motorists: Findings from a survey of cyclists in cycling risk and route acceptability. Accident; Analysis and
Queensland, Australia. Preventive Medicine, 53, 417–420. Prevention, 39, 364–371.
Hess, S., Train, K. E., & Polak, J. W. (2006). On the use of a Modified Paschalidis, E., Basbas, S., Politis, I., Prodromou, M. (2016). “Put the
Latin Hypercube Sampling (MLHS) method in the estimation of a blame on … others!”: The battle of cyclists against pedestrians and
mixed logit model for vehicle choice. Transportation Research Part car drivers at the urban environment. A cyclists’ perception study.
B: Methodological, 40, 147–163. Transportation Research Part F: Traffic Psychology and Behaviour,
Jones, T. (2012). Getting the British back on bicycles: The effects of 41, 243–260.
urban traffic-free paths on everyday cycling. Transport Policy, 20, Pattinson, W., & Thompson, R. G. (2014). Trucks and bikes: Sharing
138–149. the roads. Procedia - Social and Behavioral Sciences, 125, 251–261.
INTERNATIONAL JOURNAL OF SUSTAINABLE TRANSPORTATION 13

Prati, G., Puchades, V. M., & Pietrantoni, L. (2017). Cyclists as a Belgium: Spatial determinants and ‘re-cycling’ strategies.
minority group? Transportation Research Part F: Traffic Psychology Transportation Research Part A: Policy and Practice, 45, 118–137.
and Behaviour, 47, 34–41. Vandenbulcke, G., Thomas, I., & Int Panis, L. (2014). Predicting cycling
Sagaris, L., & Ortuzar, J. D. (2015). Reflections on citizen-technical accident risk in Brussels: A spatial case–control approach. Accident
dialogue as part of cycling-inclusive planning in Santiago, Chile. Analysis and Prevention, 62, 341–357.
Research in Transportation Economics, 53, 20–30. van Wee, B., & B€ orjesson, M. (2015). How to make CBA more suitable
Schepers, P., Twisk, D., Fishman, E., Fyhri, A., & Jensen, A. (2017). for evaluating cycling policies. Transport Policy, 44, 117–124.
The Dutch road to a high level of cycling safety. Safety Science, 92, Vij, A., & Walker, J. (2016). How, when and why integrated choice and
264–273. latent variable models are latently useful. Transportation Research
Sigurdardottir, S. B., Kaplan, S., & Møller, M. (2014). The Part B: Methodological, 90, 192–217.
motivation underlying adolescents’ intended time-frame for driving Vrtalova, J. (2009). Czechs ride a bike (in Czech). Proceedings of the
licensure and car ownership: a socio-ecological approach. Transport Cyklokonference. Olomouc, Czech Republic.
Policy, 36, 19–25. Walker, J., & Ben-Akiva, M. (2002). Generalized random utility model.
Tapp, A., Davis, A., Nancarrow, C., & Jones, S. (2016). Great Britain Mathematical Social Sciences, 43, 303–343.
adults’ opinions on cycling: Implications for policy. Transportation Wang, Y., Li, L., & Prato, C. G. (2018). The relation between working
Research Part A: Policy and Practice, 89, 14–28. conditions, aberrant driving behaviour and crash propensity among
Titze, S., Merom, D., Rissel, C., & Bauman, A. (2014). Epidemiology of taxi drivers in China. Accident Analysis and Prevention, doi:10.1016/
cycling for exercise, recreation or sport in Australia and its contribu- j.aap.2018.03.028.
tion to health-enhancing physical activity. Journal of Science and Wegman, F., Zhang, F., & Dijkstra, A. (2012). How to make more cycling
Medicine in Sport, 17, 485–490. good for road safety? Accident; Analysis and Prevention, 44, 19–29.
Vandenbulcke, G., Dujardin, C., Thomas, I., de Geus, B., Degraeuwe, Wooldridge, J. M. (2009). Introductory econometrics: A modern
B., Meeusen, R., & Int Panis, L. (2011). Cycle commuting in approach (4th ed.). Mason, OH: South-Western Cengage Learning.

You might also like