ArticlePDF Available

Abstract and Figures

Mobility as a Service (MaaS) is expected to significantly change mobility patterns, yet it is still not clear who will embrace this new mobility paradigm and how MaaS will impact passengers' transportation. In the paper, we identify factors relevant for MaaS adoption based on a survey comprised of over thousand respondents in the Netherlands. We find five clusters in relation to individuals' inclinations to adopt MaaS in the context of urban mobility. We characterize each of the clusters, allowing for the examining of different customer segments regarding MaaS. The cluster with the highest inclination for future MaaS adoption is also the largest cluster (re-presenting one third of respondents). Individuals in this cluster have multimodal weekly mobility patterns. On the contrary, current unimodal car users are the least likely to adopt MaaS. We identify high (mobility) ownership need and low technology adoption (present in three of the five clusters) as the main barriers that can hinder MaaS adoption. Policies that directly address these two barriers can stimulate MaaS adoption.
Content may be subject to copyright.
Alonso-González, M.J., Hoogendoorn-Lanser, S., van Oort, N., Cats, O. & Hoogendoorn, S.P.
(2020) Drivers and barriers in adopting Mobility as a Service (MaaS) A latent class cluster
analysis of attitudes. Transportation Research Part A: Policy and Practice, 132, 378-401.
Drivers and barriers in adopting Mobility as a Service
(MaaS) – A latent class cluster analysis of attitudes
María J. Alonso-González1, Sascha Hoogendoorn-Lanser2, Niels van Oort1, Oded Cats1, Serge
1 Department of Transport and Planning, Delft University of Technology, Delft, The
2 KiM Netherlands Institute for Transport Policy Analysis, The Hague, The Netherlands
Five distinct latent clusters identified regarding attitudes towards MaaS
Methodologies used: exploratory factor analysis and latent class cluster analysis
Largest cluster (32% of the sample): MaaS-ready individuals
Relation found between current mobility patterns and MaaS adoption potential
(Car) ownership need and technological capabilities main barriers for MaaS adoption
Keywords: Mobility as a Service (MaaS); pooled on-demand services; attitudes; latent class cluster
analysis; adoption barriers
Mobility as a Service (MaaS) is expected to significantly change mobility patterns, yet it is still not
clear who will embrace this new mobility paradigm and how MaaS will impact passengers’
transportation. In the paper, we identify factors relevant for MaaS adoption based on a survey
comprised of over thousand respondents in the Netherlands. We find five clusters in relation to
individuals’ inclinations to adopt MaaS in the context of urban mobility. We characterize each of the
clusters, allowing for the examining of different customer segments regarding MaaS. The cluster with
the highest inclination for future MaaS adoption is also the largest cluster (representing one third of
respondents). Individuals in this cluster have multimodal weekly mobility patterns. On the contrary,
current unimodal car users are the least likely to adopt MaaS. We identify high (mobility) ownership
need and low technology adoption (present in three of the five clusters) as the main barriers that can
hinder MaaS adoption. Policies that directly address these two barriers can stimulate MaaS adoption.
1 Introduction
Urban transportation is changing rapidly, with the emergence of a broad spectrum of on-demand
modes such as bike-sharing, car-sharing or ride-sharing appearing in urban areas. Even if these
mobility services have been around since the 20th century, only recently their real-time operation in
large settings has become a reality. They increase the modal choice set of travellers and their
accessibility to different locations, but the wide range of options available also implies some degree
of extra complexity for the user. In order to avoid this extra complexity and to maximise the benefits
that all these options can bring when integrated, Mobility as a Service (MaaS) is emerging.
MaaS is a service offered to the user in a single mobile app platform, which integrates all aspects of
the travel experience, including booking, payment, and information both before and during the trip
(Jittrapirom et al. (2017) and Kamargianni et al. (2016) provide an overview of early MaaS schemes).
In essence, MaaS brings an individual from A to B regardless of the mode. In dense urban settings in
which congestion, liveability and parking space are high on the urban mobility agenda, a robust
public transport system would ideally constitute the core of MaaS, with the new on-demand modes
acting as first/last mile solutions or to complement public transport for trips for which it does not
provide a convenient service (Li and Voege, 2017). The transport integration that has for long been
considered a precondition to reduce car use in favour of public transport (Chowdhury and Ceder,
2016; Givoni and Banister, 2010; Janic, 2001) is therefore provided in MaaS.
Previous research indicates that MaaS has the potential to induce modal shifts towards a more public
transport and less car oriented lifestyle (Karlsson et al., 2017; smile mobility, 2015) while it increases
users’ travel satisfaction (Sochor et al., 2016). As a result, MaaS has recently attracted much
attention, to the extent that it is expected to become the drive of a mobility revolution comparable
with the introduction of the private car in the 20th century (Goodall et al., 2017; Shaheen et al.,
2018). However, there has been a self-selection effect among individuals participating in the
researched early stage MaaS pilots (Strömberg et al., 2016). It is unknown if the general population
will replicate the modal shifts of individuals in these MaaS pilots and whether public transport or
rather on-demand services will play the mayor role in urban MaaS schemes (car users partly explain
their current mode choice decisions by referring to the inflexibility of transit (Clauss and Döppe,
In this study, we contribute to the understanding of who will embrace MaaS and which shifts in
mobility patterns MaaS is likely to occasion. Limited quantitative research has been done so far on
this topic other than the resulting from pilot evaluations, even if MaaS is expected to significantly
change our travel patterns. Our study goes beyond the consideration of early adopters and identifies
not only the characteristics of potential users of MaaS, but also the barriers that may be holding
other individuals from adopting this new mobility paradigm. We also investigate if public transport,
or rather other on-demand services are more attractive to the different traveller groups, which
indicate which changes in mobility patterns can take place as a result of MaaS. In our study, we focus
on urban areas of the Netherlands, and we discuss what the results indicate for other urban settings.
Within the study of the on-demand modes present in MaaS, we pay special attention to pooled on-
demand services. They can add flexibility without compromising on sustainability and efficient use of
mobility resources. By pooled on-demand modes, we refer to the new generation of taxi-like services
(usually booked via an app) that match different travel requests in the same vehicle (usually) in real-
time, without these matched trips needing to start or end at the same location. Examples of these
services are pooled ridesourcing services such as UberPool or UberExpressPOOL, or microtransit
services such as the services offered by Via or Chariot. Tachet et al. (2017) demonstrated that pooled
on-demand services have a high potential in urban settings, given that individual mobility patterns
are highly shareable for very diverse urban networks. Moreover, simulation studies show that
combinations of pooled on-demand services with traditional public transport (Martinez and Viegas,
2017) or with individual on-demand services (such as bike-sharing (Luo et al., 2017)), lead to drastic
reductions in the number of vehicles needed, carbon emissions and congestion, and they improve
passenger trip times and accessibility simultaneously. As a result, their contribution can indeed be
key in future MaaS schemes.
The main contributions of the present study are the following. First of all, we identify user clusters
with respect to their attitudes towards MaaS, identifying which segments of the population are more
likely to engage in MaaS (and whether pooled on-demand services also deem apt in them from an
attitudinal perspective). Second of all, we investigate if there is a relation between current mobility
patterns and the inclination towards MaaS, and interpret what this can mean to future urban
mobility. Third of all, we identify barriers that can hold back users from adopting MaaS. Finally, based
on the presented new insights, we propose a series of recommendations and policy implications
tailored to the different clusters present in the study to support future MaaS adoption.
The remainder of the paper is organised as follows: Section 2 explains the methodology employed in
this research; Section 3 presents the study results; Section 4 gives detailed insights on the individuals
belonging to the clusters presented in Section 3; Section 5 discusses the key results, focusing on
policy implications, and Section 6 provides the final conclusions.
2 Research methodology
In this section, we discuss the overall research approach, including the design of the survey and the
data analysis approach.
2.1 Survey design
We performed a survey in order to identify potential future users of sustainable MaaS schemes in the
Netherlands. Given that the higher densities of urban areas better allow for the economically viable
coexistence of a robust transit system and different on-demand services, we exclusively targeted
individuals living in (sub)urban areas in The Netherlands in our study. The survey included several
attitudinal Likert-scale statements regarding attitudes towards MaaS, with an emphasis on pooled
on-demand services. The included attitudinal indicators are explained in detail in the following
Survey respondents were recruited from the Netherlands Mobility Panel (MobilitieitsPanel
Nederland, MPN), which is an annual household panel designed for the longitudinal study of travel
behaviour in the Netherlands (Hoogendoorn-Lanser et al., 2015). In addition to the annual panel
waves, MPN respondents occasionally take part in specific questionnaires, as is the one designed for
the current piece of research.
2.1.1 Attitudinal indicators
MaaS is still in its first stages. Therefore, the study of transport behaviour in real MaaS settings is still
limited to the small number of MaaS pilots currently available. We add to this knowledge by carefully
designing a series of attitudinal indicators to better understand the mobility changes MaaS will spark.
This methodology is underpinned by previous research which has found a relation between attitudes
and behaviour in mode choice (Molin et al., 2016). Moreover, previous research has shown that
attitudinal approaches that are used as a base for mobility segmentation are advantageous as a
starting point for related policy interventions (Haustein, 2012; Haustein and Hunecke, 2013).
Durand et al. (2018) identified three main aspects relevant when investigating changes in travel
preferences that can take place as a result of MaaS: (i) mobility integration, (ii) shared mobility
modes, and (iii) mobile applications. In our analysis, we add a category focusing on willingness to pay,
to have a notion of the business case for MaaS. Figure 1 shows the key aspects of the attitudinal
indicators included in the survey, which will be described in detail below. The complete formulation
of the attitudinal indicators as well as their source (where applicable) are detailed in Appendix A.
Figure 1: Key aspects of the attitudinal Likert-scale indicators
Mobility integration. Individuals need to be willing to integrate different modes of transport as part
of their travel patterns in order to exploit the benefits provided by MaaS. This willingness to use
different modes can, in turn, be influenced by individuals’ attitudes towards public transport and
private car. Therefore, we include three subcategories:
i) Multimodal mind-set. We understand the multimodal mind-set as the willingness to
integrate different modes of transport into one’s travel patterns. Similarly, we refer to
multimodal individuals as those who include different transport modes in their weekly
mobility. With regard to the multimodal mind-set, we differentiate two aspects, namely
the attitude towards multimodality with the traditional modes, and the openness to
innovate in mobility.
ii) Public transport attitude. Attitudes towards public transport in the Netherlands are
more negative than those towards bike or car (Kennisinstituut voor Mobiliteitsbeleid
(KiM), 2018), yet a positive attitude towards public transport is helpful in order to accept
a MaaS scheme with public transport as its core. We do not include operational aspects
in our statements due to large differences in frequency and reliability between different
available services. Rather, we focus on sharing (common characteristic for all public
transport modes) and common goals of public transport and MaaS: cost and
environmental impact reduction.
iii) Private car attitude. From a utilitarian perspective, MaaS can offer a good alternative to
using a privately-owned car. However, symbolic and affective motives related to car
usage have been found more important than utilitarian ones (Steg, 2005). This would
make it more difficult to shift from the current mobility paradigm towards MaaS.
Therefore, we address these motives in our indicators.
Shared mobility modes. Given the still limited experience of most individuals with these services, we
consider novel shared mobility modes independently, and not merged in the attitudes towards the
more general mobility integration, as suggested by Durand et al. (2018). In this study, we focus on
pooled on-demand services as an example of shared mobility modes. These services do not only
provide the flexibility of on-demand services, but they also offer a collective service, fitting the needs
of congested urban areas.
Pooled on-demand services (referred to as FLEXI when presented to respondents) were described in
detail in the questionnaire. It was introduced as a mobility service which could have a maximum of
six people in a vehicle and was bookable in real time via an app (or via a mobile phone for those not
owning a smartphone). The pick-up point was assumed to be 1-minute walking distance from their
location, and detours could take place to pick up or drop off other passengers. Before being
presented with the related attitudinal indicators, respondents also completed two stated preference
experiments focusing on reliability of these pooled on-demand services. This way, respondents had a
better understanding of both the flexibility (+) and the variability (-) associated with flexible route
and schedule services and could form an opinion towards these services prior to indicating their
attitude towards the envisaged service. In turn, this allowed us to ask respondents about their
intention to use pooled on-demand services.
Within pooled on-demand services, our main interest is to analyse their flexibility trait. Flexibility is
the common characteristic of all on-demand services, and is arguably the fundamental difference
between these services and traditional public transport. Therefore, even if only pooled on-demand
services are explicitly addressed, the outcomes of some of the indicators included can (at least
partially) be transferred to other on-demand services. We cover aspects that address both temporal
and spatial flexibility.
Additionally, we analyse the safety construct. Adequate safety is the (non-performance related) most
basic transportation need (Peek and van Hagen, 2002) and a point of concern of some individuals for
pooled on-demand services specifically (Morales Sarriera et al., 2017). Traffic safety and social safety
are covered in the indicators.
Mobile applications. Given that individuals interact with MaaS services via an app interface, it is
necessary to investigate their willingness to adopt the app. Even in countries where mobile phone
adoption is almost ubiquitous, attitudes and skills can widely differ among individuals. Potential users
need to not only have a smartphone and internet connection to operate the app, they also need to
be willing to install the app, have enough skills to operate it, and have trust in the app. These three
aspects are covered in our study.
Willingness to pay. The added value of MaaS lies in its integration of all modes of transport and travel
stages, and in its real-time information functions, which enable both better services and better
information. Under this category, we want to better understand respondents’ willingness to pay for
improved mobility, as well as their perceived need for improvements. Some studies consider
bundling packages (i.e., having monthly subscriptions instead of paying per individual trip) a key
aspect in MaaS. MaaS, as is considered in this study, does not require bundles. However, we also
include a statement regarding bundling preferences to obtain a first impression on this aspect. We
refer the reader to Ho et al. (2018) and Matyas and Kamargianni (2018) for those looking for studies
regarding MaaS willingness to pay in bundling options.
All attitudinal indicators are presented to respondents as 5-point Likert-scale statements (strongly
disagree / disagree / neutral / agree / strongly agree). Moreover, respondents are also given the ‘Not
applicable’ answer option. Indicators are presented to respondents in blocks of either 4 or 5
statements. The order of the statements is randomised within each block.
2.1.2 Habits and current behaviour
Since habits and current behaviour are important predictors of future transportation behaviour
(Lanzini and Khan, 2017), we complemented the previous attitudinal indicators with questions
related to respondents’ experience with aspects relevant to MaaS.
We inquire respondents’ adoption of mobile technology (needed to operate any MaaS app) and their
usage experience with the predecessor of the MaaS app: the journey planners (multimodal journey
planners are considered Level 1 MaaS apps (Sochor et al., 2017)). Also, we look into individuals’
current mobility patterns. We already had information regarding individuals’ travel patterns from the
2017 wave of the MPN general annual survey. We add to this information by inquiring about
respondents’ familiarity with on-demand services. Additionally, to better understand what drives
respondents while shaping their transport mode choices in their trips, we ask them for their motives
in this decision process.
2.2 Analysis framework
Figure 2 provides a step-wise overview of the main steps of the analysis. First, the data is cleaned.
Respondents who require an unrealistically low time to complete the questionnaire, recurrently
straight line (i.e., do not differentiate in ratings) and repeatedly select the ‘not-applicable’ option are
considered invalid. Following, exploratory factor analysis and latent class cluster analysis are
performed for variable reduction and clustering purposes.
Even if segmentation in transport literature often stems from differences in socioeconomic
characteristics or current travel behaviour, previous research has identified attitudes as important
predictors of travel behaviour (Pronello and Camusso, 2011). For example, Hunecke et al. (2010)
found mobility-related attitude-based segmentation to yield greater differences in mobility
behaviour than those based on socioeconomics, and Redmond (2000) found it to result in the highest
predictive power for travel mode choice. Research into attitude-based segmentation has significantly
increased in recent years. The common methodology, as the one followed in the paper at hand, is
composed of a variable reduction technique and a subsequent cluster analysis. We refer the reader
to Anable (2005), Haustein and Hunecke (2013) or Pronello and Camusso (2011) for further literature
on previous research regarding mobility attitude-based segmentation. The following Sections 2.2.1
and 2.2.2 explain the methodology used in our research in more detail.
Figure 2: Step-wise scheme of the analysis framework
2.2.1 Exploratory Factor Analysis
In this study, we look for relationships among the variables that may be different from the prior
expectations of the categories presented in Figure 1. Therefore, Exploratory Factor Analysis (EFA;
(Williams et al., 2010)) is the variable reduction technique used in this study. EFA accounts for the
common variance among the variables (and is not to be confused with principal component analysis)
(Suhr, 2005).
EFA can be performed exclusively on interval or ratio level variables (Suhr, 2005). Equidistance is
often assumed between the different levels of Likert-scales, which allows us to perform EFA on our
data. To identify if a considerable number of respondents does not feel addressed by some of the
statements, we included the ‘non-applicable’ option. However, this option introduces data that falls
out of the Likert-scale. We remove from the analysis any variable with a considerable number of non-
applicable responses from the posterior analysis (not-applicable responses are not distributed at
random). Low recurrence of non-applicability in a variable (<5.5%) is accepted and this data is treated
as missing at random. This data is imputed using expectation maximisation, which produces the
maximum likelihood estimation of parameters using all observed information (Acock, 2005). We
impute the correlation matrix (using the SPSS add-on module presented in (Weaver and Maxwell,
2014)) instead of imputing the variables themselves, to overcome SPSS shortfall in including standard
errors in the expectation maximisation imputation (von Hippel, 2004).
Factor scores are then calculated using a non-refined method. These methods are more stable across
different samples than refined methods (Distefano et al., 2009). If factor loading differences among
the indicators of the different factors are small, the ‘non-weighted sum score method’ will be used.
Otherwise, the ‘weighted sum score method’ will be preferred. Both methods allow for a direct
interpretation of the factor value in relation to the 5-point Likert scale presented to respondents.
2.2.2 Latent Class Cluster Analysis
We aim at identifying respondents that share similar attitudes on the researched indicators. We
hypothesise that attitudes on these indicators are to some extent related to each other,
encompassed in their attitude towards MaaS. To this end, we perform latent class cluster analysis
(LCCA). LCCA models, also referred to as finite mixture models, group individuals in different classes
according to an unobserved (latent) class variable that explains their responses on a set of observed
indicators (Molin et al., 2016).
Figure 3 shows the conceptual latent class model used in the analysis. The EFA factors are the
indicators of the model that help delve into the latent variable that is behind the differentiation of
the latent classes. The covariates, represented in the lower part of Figure 3, help characterise the
different classes. Covariates on socioeconomic, mobility and technology-related characteristics are
added to the model after a model without covariates with adequate model fit has been identified.
Whenever the covariates do not improve the model, they are only included as passive covariates, to
aid cluster identification.
Figure 3: Scheme of the latent class cluster model with the investigated covariates
The mathematical formulation of the model with the covariates takes the following form (Vermunt
and Magidson, 2016):
) = (|
=1 (1)
Where x is the latent variable with its K categories,
 individual's i set of covariates and 
individual's i response to indicator m (M being the number of indicators). The first factor of the
equation refers to the probability of belonging to a certain latent class given the individual's
covariates, and the second factor is the probability density of given x. This mathematical
formulation holds assuming that the indicator variables are independent of each other conditional on
the latent variable x (Vermunt and Magidson, 2016). A violation of this assumption in our model,
which can be measured by means of the bivariate residuals, would indicate that the model lacks local
fit and that it cannot be trusted (Oberski, 2016). We therefore examine this assumption by studying
the bivariate residuals (BVR). Applied research often considers BVR to be chi-squared distributed, yet
this approach does not always work satisfactorily (Oberski et al., 2013). Instead, Oberski et al. (2013)
suggest to analyse the BVR p-values of the parametric bootstrapping. We follow this procedure in
combination with the study of the bootstrapped L² of the overall model, as done in Oberski (2013).
Some of the variance present in the initial data of the attitudinal indicators is lost by using the
obtained factors as only model indicators in the LCCA. Additionally, this approach treats the EFA
factors as observed variables in the LCCA, ignoring the uncertainty that arises from the measurement
of the factors through its attitudinal indicators. Still, the large number of statements included makes
this double approach (variable reduction and subsequent cluster analysis) the rule in attitude-based
segmentation studies, as previously mentioned in Section 2.2.
3 Results
The analysis and modelling approach detailed in the previous section has been applied to a dataset
representative of the urban Dutch population. Figure 4 indicates the research questions that are
answered in the different sections of the analysis and interpretation of the results. Data collection
and descriptive statistics are first presented (Section 3.1) followed by the Exploratory Factor Analysis
(Section 3.2) and the Latent Class Cluster identification (Section 3.3). These clusters are further
characterised in Section 4.
Figure 4: Research questions answered in the analysis and interpretation of the results
3.1 Data collection and sample description
To test our questionnaire, an on-line pilot was performed on April 2018. No modifications of the
attitudinal indicators used in this study were needed after the pilot. Our final questionnaire
(conducted on-line and in Dutch) was distributed on May 2018. A total of 1077 respondents finished
the questionnaire, of which 1006 (93%) were considered valid respondents after data cleaning. Only
individuals aged 18 and older in possession of a mobile telephone took part in the questionnaire. We
targeted respondents living in areas with more than 1,500 inhabitants/km² (highly urbanised areas
according the ‘urbanity degree’ indicator used in the Netherlands (Centraal Bureau voor de Statistiek
(CBS), 1992)), and all respondents belonged to different households. The socio-economic
characteristics of the sample, as well as the Dutch values (both for (very) high urbanised areas and
for the whole of The Netherlands) are included in Table 1.
Our sample satisfactorily represents the shares between the two levels of urbanisation levels (high
urbanised areas and very high urbanised areas) and gender. Regarding age, middle aged adults are a
bit underrepresented and the elderly population slightly overrepresented. We can observe some
differences between the age shares of the urban areas exclusively and the average of the
Netherlands. As expected, younger adults are more prominent in urban environments. This is well
represented in our sample.
We do not have values regarding education level, working status and household composition for the
urbanised areas, only for the average of the Dutch population. Differences between our sample and
the national average are as expected: a higher proportion of highly educated respondents and
working individuals as well as a higher share of single person households than the national average.
In general, we consider the representativeness of our sample to the shares of the target population
to be adequate.
Dutch (very) high
urbanised areas
Dutch 2018
18* to 39
40 to 64
65 and above
Work status
No working
1 person household
> 1 person household
Urbanisation level
Very high urbanised
(>2500 inhab./km²)
High urbanised (1500-2500
* 18 to 39 for the share sample, but 20 to 39 for the Dutch 2018
Table 1: Comparison between the sample and Dutch population for different socio-economic variables. Sources for the
population data: (Centraal Bureau voor de Statistiek (CBS), 2018a), (Centraal Bureau voor de Statistiek (CBS), 2018b),
(Centraal Bureau voor de Statistiek (CBS), 2018c), (Centraal Bureau voor de Statistiek (CBS), 2018d)
3.2 Exploratory Factor Analysis of MaaS indicators
2.6% of the total data from the attitudinal indicators was marked as ‘not applicable’, a high share of
which is present in three out of the 31 indicators. These three indicators are not considered for the
EFA analysis. The remaining missing values (1.4% of the total data) were imputed using expectation
maximisation. We performed the EFA employing the Principal Axis Factoring extraction method
(unlike other methods it does not require the multivariate normality assumption (Fabrigar et al.,
1999)) with oblimin oblique rotation (which allows for correlation among factors and thus better
replicates human behaviours (Williams et al., 2010)). We investigate the suitability of the data for
EFA with the Kaiser-Meyer-Olkin’s (KMO) measure of sampling adequacy and Bartlett’s test of
sphericity (Field, 2009). We obtain a KMO of 0.835, which shows good sample adequacy (Hutcheson
and Sofroniou, 1999), and the Bartlett’s test of sphericity is <0.001, which indicates sufficient
relations between indicators for the EFA. Since the average communality (i.e., the proportion of the
common variance present in the variables (Field, 2009)) is lower than 0.6, and the sample size is well
over 200, we follow the scree plot criterion (Cattell, 1966) to decide on the number of factors (Field,
2009). This leads us to retain 5 factors in the factor analysis, which explains 44.6% of the variance.
Table 2 shows the factors founds and the factor loadings for the rotated pattern matrix. For
interpretation, we only consider loadings bigger than 0.4, as advised in (Field, 2009). The rest of the
indicators belonging to a factor are depicted in Table 2 in grey without the loading. A subsequent EFA
on the variables that loaded significantly (>0.4) on the previous factors, leads to the exact same
factors and very similar indicators (KMO=0.802, Bartlett’s test of sphericity <0.001, 51.1% variance
explained). Only these loaded statements are considered for the posterior LCCA. The comparable
loadings of the indicators belonging to the different factors indicate that they all contribute to a
similar degree to the factor to which they belong. Therefore, factor scores are calculated using the
“non-weighted sum scores” method. (For the interested reader, the scree plots of both EFA are
included in Appendix B).
Factors and their indicators
Mobility integration factor
It is important to use public transport to preserve the environment
I choose to travel with public transport or to share rides to reduce my trip costs
I do not mind which transport mode I use, as long as it suits my trip needs
I am willing to try new ways to travel
People like me only use their own bike and/or car (reversed)
I often compare different travel options and transport modes before choosing how to
I like the privacy in the car or bike (reversed)
It makes me uncomfortable to ride with strangers on public transport
I would not mind if other travellers get in or off the FLEXI vehicle during my ride
I think the public transport is not so clean or decent
I like travelling always in the same way
It is essential to be able to easily combine different transport modes (such as bus, car,
bike or car-sharing) in order to improve transportation in the Netherlands
FLEXI over Public Transport (PT) factor (shared mobility modes)
FLEXI seems to me more reliable than current public transport
I would feel safer in FLEXI than in a regular bus
I find FLEXI’s flexibility in the departure time more convenient than traditional transit.
The proximity of a driver would make me feel safe in FLEXI
FLEXI would give me the freedom to travel where I need to be when needed
I like that FLEXI does not have a fixed schedule or route
FLEXI concerns factor (shared mobility modes)
FLEXI does not have fix schedules. That would worry me.
I would be worried that FLEXI departs without me
I think that FLEXI drivers do not drive carefully
I would find it annoying that FLEXI does not drive the fastest route (e.g., FLEXI’s route is
18 minutes instead of 15 minutes)
Mobile application factor
I would use a (smartphone) app if it gave me access to all available travel alternatives
I like to pay for my rides via a (smartphone) app
It is easy for me to find FLEXI’s pick-up point if it is displayed on a map in the
(smartphone) app
Willingness to pay factor
I would be ready to pay for precise and reliable travel information
I am willing to pay more to have a more predictable travel time for my journey
I find it difficult to find information of all available travel alternatives
Table 2: Results of the pattern matrix of the exploratory factor analysis
The factors found are well in line with those from the survey design phase (Figure 1). The mobility
integration factor shows how a positive multimodal mind-set aligns with a positive attitude towards
public transport and a low car drive. In line with expectations, those three subcategories belong to an
overarching factor. Interestingly, we found two factors regarding indicators pertaining (pooled)
shared mobility modes, and they do not pertain to the flexibility and safety traits of these services.
Instead, they refer to (i) the comprehensive preference of FLEXI (i.e., pooled on-demand services)
over traditional transit (or vice versa), and (ii) concerns towards the new mobility service. The two
factors being distinct suggests that even if there may be people that prefer public transport over
pooled on-demand services, they may not necessarily have concerns related to using pooled on-
demand services; similarly, individuals that prefer pooled on-demand services over public transport
are still not necessarily very positive towards these services and may have concerns related to their
3.3 Latent Class Cluster identification
We perform the Latent Class Cluster Analysis (LCCA) using the Latent GOLD software (version 5.1)
(Vermunt and Magidson, 2016). We include the five EFA factors as well as the intention to use
pooled on-demand services (binary variable, yes/no) as model indicators. To decide on the number
of classes, we analyse both the BIC and the AIC global goodness-of-fit statistics in models ranging
from one to seven classes. While the lowest BIC is shown for the 3-class model, the AIC keeps
decreasing with the increase in the number of classes. We also examine the classification errors for
all models. Taking all these three things into account, we consider the 5-class model as the most
adequate. In this model, the BVR of the pair ‘FLEXI over PT’ & ‘FLEXI concerns’ is very low. Thus, we
add a direct effect between these two factors, freeing the local dependence between them. All the
attitudinal statements involved in these two factors share attitudes towards (the fictitious) FLEXI.
Similar wording and varying interpretation of this new service from different respondents (as a result
of the service description in the survey) may have led to the association between the two factors in
the LCCA. As a result, we consider adding the direct effect useful (Magidson and Vermunt, 2004).
Given that the overall bootstrapped L² p-value of the model is adequate, we do not add additional
direct effects.
As the next step, we explore the effect of covariates presented in Figure 3 on the latent class
membership. Inclusion of covariates leads to changes in the clusters, but it helps differentiate
individuals in the different clusters further, which, in turn, helps target policy recommendations. The
covariate analysis is first done per type (using the types mentioned in Figure 3). Covariates that prove
significant per type are then put together incrementally one by one, and deleted if they become
insignificant when combining different covariates. We find seven covariates to improve the model:
3G bundle availability, working status, education level, urbanisation level, bike use frequency,
acquaintance with bike-sharing systems and presence of children in the household. The overall
bootstrapped L² p-value of the final model is 0.30 (values > 0.05 provide an adequate fit (Vermunt
and Magidson, 2005)), and the entropy of the model amounts to 0.66.
The profile of the indicators and active covariates of our final model are depicted in Table 3 (for the
interested reader, parameters are included in Appendix C). We name the five clusters, ordered by
their share from the largest to the smallest, as follows:
- Cluster 1 (32% of the sample): “MaaS-FLEXI-ready individuals”. This cluster, which includes
roughly one third of the respondents, has the highest average on all six indicators in
comparison to all other 4 clusters, indicating the highest inclination for MaaS adoption and a
remarkable 99% usage intention towards pooled on-demand services (FLEXI). It has the
highest willingness to pay of all clusters, albeit still lower than the neutral value (2.9) -
suggesting that the urban Dutch society is not willing to pay for improvements in mobility.
- Cluster 2 (25%): “Mobility neutrals”. With a quarter of the sample, this cluster has an average
of neutral in relation to almost all factors. They can be regarded as conservative, undecided
or neutral-minded. Intention to adopt pooled on-demand services is the second highest
among the five clusters, which can indicate that even if they have an overall neutral
approach regarding the analysed attitudes, they are still open towards adopting new mobility
- Cluster 3 (22%): “Technological car-lovers”. This cluster differs from the two previous clusters
in the low value of the ‘mobility integration’ factor, showing a stronger inclination towards
privately owned modes over public transport or other shared modes. Further, adoption
intention towards pooled on-demand services is lower than the average of our sample
despite this group having a neutral attitude towards these services. The further analysis of
the covariates shows that within the owned modes, it is the car which is most dominant in
this cluster. This preference may stem from enthusiasm towards the car or due to perceived
need. Despite the low value of the factor related to mobility integration, this cluster is
associated with a high value in the ‘mobile application’ factor. This underscores the
importance of differentiating mobile-application from mobility-related aspects in the study
of MaaS adoption.
- Cluster 4 (15%): “Multimodal public transport supporters”. With roughly one sixth of the
sample, this is the only cluster that next to Cluster 1 has a higher than neutral average value
for both ‘mobility integration’ and ‘mobile application’ factor. Notwithstanding, this cluster
strongly differs from Cluster 1 in the FLEXI over PT factor: while individuals of Cluster 1 prefer
pooled on-demand services over traditional public transport, this is the other way around for
individuals in this cluster. This difference highlights that having a positive attitude towards
mobility integration does not imply future adherence to the new shared modes. Intention to
use pooled on-demand services is at a level between the one observed among respondents
in Cluster 3 and Cluster 1. Also, respondents in this cluster have a lower average score in the
willingness to pay factor than the previous three clusters for improvements in mobility,
showing higher cost sensitivity.
- Cluster 5 (6%): “Anti new-mobility individuals. The smallest cluster can be described as the
contra of Cluster 1: showing negative attitude towards all presented factors and an
extremely low intention to use pooled on-demand services (12%). This cluster shares with
Cluster 3 the low value for the ‘mobility integration’ factor. However, unlike Cluster 3,
respondents in this cluster also show limited technology affinity and a very negative attitude
towards pooled on-demand services. Respondents in this cluster are therefore very unlikely
to adopt any new mobility solution that is presented to them.
gical car-
Anti new-
Cluster Size 32% 25% 22% 15% 6%
Mobility integration factor
FLEXI over PT factor
FLEXI concerns factor
Mobile application factor
Willingness to pay factor
Intention to use FLEXI for free-
time purposes
Active covariates
Working (voluntary work
Highest education
Child under 12 years old in
Urbanisation level
Highly urbanised
Very highly urbanised
Reported bike use frequency
(almost) never
less than 1 day/month
1 to 3 days a month
1 to 3 days a week
4 or more days a week
Heard about bike sharing
systems (no Dutch OV-bicycle)
3G bundle available on
smartphone or tablet
Table 3: Profile of the final LCCA model for both indicators and active covariates. For the active covariates, we highlight in
bold the class with the highest share for each characteristic.
We graphically present the scores of the five EFA factors for the different clusters in Figure 5. We
further investigate the average values of the individual attitudinal indicators for the different clusters
in Figure 6. Indicators excluded from the LCCA (share of ‘non-applicable’ values >5% or EFA factor
loadings <0.4) are also included in the radar graphs for a more comprehensive overview of all studied
aspects. To ease the interpretation, indicators negative to MaaS and/or to their pooled on-demand
services have been reversed. The radar graphs confirm that the general trends represented by the
factors are also present for the individual attitudinal indicators, with ‘MaaS-FLEXI-ready individuals’
scoring, in general, highest and anti new-mobility individuals’ scoring lowest. The strong positive
attitude of the ‘multimodal public transport supporters’ towards public transport is also clear from
the public transport related statements in the graphs. The graphs also provide deeper insights into
the extent to which MaaS related indicators differ. Privacy stands out for its importance while
willingness to pay for travel information is distinct for its low scores. Regarding the mobile
application factor, the app payment acceptance indicator scores significantly lower than the other
related indicators.
Figure 5: Average score of the five EFA factors for the different clusters.
Figure 6: Average score for the attitudinal indicators related to: (a) mobility integration; (b) (pooled) shared mobility modes, and; (c) mobile applications and willingness to pay
4 Detailed characterisation of the clusters
The five clusters of our final model are further profiled with the information of covariates (active and
passive). We differentiate three aspects (as indicated in Figure 3): (a) socioeconomic characteristics,
(b) mobility characteristics, and (c) technology related characteristics. They are discussed in the
following sub-sections 4.1-4.3, respectively.
4.1 Socioeconomic characteristics’ analysis of the latent clusters
Different socioeconomic covariates are included in the model to better understand the five clusters.
Working status, education level, urbanisation level of the residence location, and the existence of
children under 12 years old improve the membership function of the model (active covariates,
depicted in Table 3). We also add age, gender, household composition and income as passive
covariates in the model (see Table 4).
‘MaaS-FLEXI-ready individuals’ tend to be highly educated, young, have slightly higher average
incomes and reside in the highest urbanised areas. These characteristics go in line with the
characteristics that have been attributed to adopters of shared-mobility services (Alemi, 2018;
Clewlow, 2016; Shaheen et al., 2012). ‘Multimodal PT supporters’ have a similar socioeconomic
profile, only differing from the first cluster in their slightly lower average income.
‘Mobility neutrals’ are associated with a high percentage of 65+ age old respondents (57%). Most of
the individuals in this cluster (79%) do not work, arguably due to the large number of retired people
in this cluster, and have a lower average income.
The ‘technological car-lovers’ and the ‘anti new-mobility individuals’, the classes less inclined
towards mobility integration, share the (slight) over-representativeness of males. This may be
explained by the higher modal share of car among men than women in the Netherlands (Molin et al.,
2016). These two classes strongly differ in relation to other socioeconomic characteristics, though.
‘Technological car-lovers’ are distinct from all others for the higher percentage of households with
children (37%), many of them including children aged 12 or younger (23%). In line with this result,
Md Oakil et al. (2016) found a higher car dependency among those becoming parents. ‘Anti new-
mobility individuals’ have the most balanced age composition, representing roughly even shares of
all age segments, and this cluster is associated with relatively lower income individuals, higher only
those of the ‘mobility neutrals.
l car-lovers
Anti new-
18-34 years
35-49 years
50-64 years
65+ years old
Couple (or
single parent)
+ children
Personal net
No personal
< 2000 Eur
2000 - 3000
>3000 Eur
Missing value
Table 4: Socioeconomic inactive covariates for individuals of the five clusters. For each cluster, we highlight in bold the
class with the highest share.
4.2 Mobility characteristics’ analysis of the latent clusters
This subsection presents a detailed analysis of the mobility characteristics of the five clusters. We
first examine respondents’ travel patterns and then their main drives when choosing a transport
4.2.1 Travel patterns
Reported bike frequency use and bike sharing awareness are active covariates of the model. Other
variables related to mobility are also added to the model as inactive covariates (presented in Table
5), namely: car ownership, public transport card possession, car use (stated frequency), public
transport use (stated frequency), weekly mobility patterns (stated), usage of new modes and main
reasons to choose a transport mode.
‘MaaS-FLEXI-ready individuals’ and ‘multimodal PT supporters’ are the classes with the highest share
of individuals in possession of a public transport smartcard (over 90%) while the ‘technological car-
lovers’ and the ‘anti new-mobility individuals’ have the highest household car ownership shares
(roughly 90%). The shares of the ‘mobility neutrals’ in these two aspects are in the middle of the five
groups, in line with their intermediate position towards mobility. Usage of car and public transport
resemble the trends in car ownership and smartcard possession. Interestingly, bike usage follows the
same pattern as public transport, with ‘MaaS-FLEXI-ready individuals’ and ‘multimodal PT supporters’
biking the most often and ‘technological car-lovers’ and ‘anti new-mobility individuals’ the least
l car-lovers
Anti new-
Car in household
Public transport card
Reported car frequency
(almost) never
Less than 1 per
1 to 3 days per
1 to 3 days per week
4 or more days per
Reported train frequency
(almost) never
Less than 1 per
1 to 3 days per
1 to 3 days per week
4 or more days per
Reported BTM
(almost) never
Less than 1 per
1 to 3 days per
1 to 3 days per week
4 or more days per
OV-bicycle ever used
(specific bike sharing
Bike sharing ever used
(different from OV-bicycle)
Uber ever used
Car sharing used (in the past
12 months, question from
annual 2017 MPN wave)
Table 5: Mobility inactive covariates for individuals of the five clusters. For each characteristic, we highlight in bold the
class with the highest share.
We visualise the weekly mobility patterns of the individuals in Figure 7 (considering car, public
transport and (e-)bike). 40% of ‘technological car-lovers’ and ‘anti new-mobility individuals’ have an
unimodal car behaviour, while this percentage drops to around 10% for ‘MaaS-FLEXI-ready
individuals’ and ‘multimodal PT supporters’ (the two most multimodal clusters). Nonetheless, car
usage is still more recurrent than public transport usage in all five clusters. Around 40% of
‘multimodal PT supporters’ and around 30% of ‘MaaS-FLEXI-ready individuals’ use some sort of
public transport on a weekly basis. This percentage drops to less than 10% for ‘technological car-
lovers’ and ‘anti new-mobility individuals’. Moreover, a large share of individuals in these last two
clusters report that they never use public transport. These results further show the alignment
between attitudes and behaviour regarding mobility. Current unimodal car users are the least likely
to be attracted by MaaS and the shared flexible transport modes offered by it.
Figure 7: Current weekly mobility patterns of respondents of het different latent classes (train and BTM have been
merged in the public transport (PT) category)
We also analyse both individuals’ awareness and usage of new shared mobility modes. Uber and OV-
bikes (station-based bike-sharing of the Dutch train operator NS) are familiar to the large majority of
respondents. Respondents are less familiar with bike-sharing schemes other than the OV-bikes (now
proliferating in the Netherlands) (Table 3), with the two clusters with higher multimodal mind-sets
(‘MaaS-FLEXI-ready individuals’ and ‘multimodal PT supporters’) being more aware of their existence
than the non-multimodal-minded clusters (‘technological car-lovers’ and ‘anti new-mobility
individuals’). When examining usage of new modes of transport, we observe that the share of people
who have used these modes varies depending on the mode, but is always highest for ‘MaaS-FLEXI-
ready individuals’ and ‘multimodal PT supporters’ (OV-bike 25-32%; other bike-sharing systems 4-6%;
Uber 12-19%; car-sharing 4-6%) than for the other groups (for which values under 5% are the rule).
‘Mobility neutrals’ resemble more ‘technological car-lovers’ and ‘anti new-mobility individuals’ with
respect to new mobility modes. Presumably, the higher age range of these respondents (and
somewhat lower technology capabilities) may be a hindrance in the usage of new modes of
transport, even if they might be more willing to be multimodal.
4.2.2 Drives in mode choice
Next, we analyse the main drives for choosing a mode of transport for the individuals in each of the
clusters. Among 15 different possibilities (comfort, relax, time, safety, flexibility, joy, status,
reliability, price, environment, directness, ownership, health, carrying space, and other) respondents
were asked to choose the three that are most relevant for them in deciding which mode of transport
to use. These drives are depicted in Figure 8, ordered from most to least chosen. Three
characteristics set the two more car-driven clusters (‘technological car lovers’ and ‘anti new-mobility
individuals’) apart from the other three.
The first one is ownership. Despite it not being a strict reason to choose a mode of transport but
rather a precondition state, it is the most often mentioned reason among respondents from these
two clusters (50-60% chose this factor in contrast to around 30% of respondents in the other three
clusters). Therefore, mode ownership may indeed be one of the reasons behind their lesser interest
in MaaS. The second is price relevance. There seems to be a link between multimodal-minded
individuals and price consciousness, the more unimodal car individuals being less driven by economic
reasons in their mobility decisions. And the third is environmental friendliness. The two more car-
driven clusters are less environmentally friendly than the other three (even if this is not a major drive
for any of the clusters). When asked directly whether respondents took into account the
environment in their travel behaviour, less than 25% from the two more car-driven clusters did so, in
contrast to around 40% of respondents in the other three clusters.
It is also worth noticing the low number of respondents that chose safety as driving force in their
mode decisions. This is likely not due to them granting safety a low importance. Rather, they
presumably consider safety a precondition present in all modes from which they make their mode
Figure 8: Share of respondents of the different latent clusters for whom each of the presented statements were among
the three most important reasons to choose a mode of transport
4.3 Technology related characteristics’ analysis of the latent clusters
For a user to make use of MaaS and their on-demand services, he needs to have a smartphone and
internet connection. However, 29% of the ‘mobility neutralsand 22% of the ‘anti new-mobility
individuals’ do not currently own a smartphone, and a much higher percentage (79% and 43%
respectively), are not subscribed to a 3G bundle, necessary for ubiquitous internet connection. As a
result, these two groups are in a disadvantageous situation to use new mobility solutions.
‘Multimodal public transport supporters’ also lie a bit behind the top tier technology classes (MaaS-
FLEXI-ready individualsand ‘technological car lovers’), with 11% of respondents lacking 3G bundles
(see Table 3).
Additionally to the MaaS-related attitudinal statements included in our analysis, respondents were
faced with five Likert-scale statements regarding their innovativeness attitude (see Appendix A for
the statements’ description). A ‘general innovativeness factor’ is calculated from these using the
“non-weighted sum scores” method (after checking that all five statements load together
satisfactorily). “MaaS-FLEXI ready individuals” are the most positive towards innovativeness (3.4),
followed by “Mobility neutrals” (3.2) and “Technological car-lovers” (3.1). The somewhat lower score
of “Multimodal public transport supporters” (2.8) highlights that their lower openness to innovation
encompasses other aspects beyond new on-demand mobility services. Finally, as could be expected,
“Anti new-mobility individuals” have the lowest average value (2.4).
We also analyse journey planner usage (see Table 6). Technology adoption and attitude towards
integrated mobility seem to explain the encountered differences among the clusters. The vast
majority of individuals in the two pro-integrated-mobility clusters (‘MaaS-FLEXI-ready individuals’
and ‘multimodal public transport supporters’) use travel information via their smartphone or tablet
(over 50% of them on a weekly basis), whereas over one third of respondents in the less
technological clusters (‘mobility neutrals’ and ‘anti new-mobility individuals’) never do so. Motives to
look for travel information also vary widely among classes. While one third of respondents from the
pro-integrated-mobility clusters use travel information to help them decide the most adequate mode
for a given trip, only 7% - 14% of respondents in the other three clusters do so. ‘Technological car-
lovers’ have the highest percentage of individuals using car-related travel information, but their rates
using travel information for public transport or active mode trips are half than those for the pro-
integrated-mobility clusters.
LC 1
LC 2
LC 3
LC 4
LC 5
al car-lovers
al public
Anti new-
How often do you look for
travel information via your
smartphone and/or tablet?
less than once a month
1-3 days a month
1-3 days a week
4 or more days a week
I use travel or route
information …
…to decide which mode
of transport I use
[car or motorcycle]
find information about
my travel time,
congestion or accidents
[car or motorcycle]…to
decide which route to
[public transport] … to
get information about
schedules, travel time
and delays
[public transport] … to
decide which route to
[bicycle, moped or on
foot] … to decide which
route to take
I do not use any online
travel and route
Table 6: Journey planners’ usage for individuals of the five clusters
5 Discussion
In this section, we discuss the key findings and provide some policy recommendations specific for
each of the clusters.
5.1 Key findings
From the mobility point of view, MaaS integrates the available mobility alternatives. Results of this
study show that there is an underlying mobility integration factor, in which a positive multimodal
mind-set is aligned with a favourable attitude towards public transport and a low car drive. Results
also show that these attitudes are aligned with current mobility patterns. As a result, individuals with
more unimodal car behaviours seem less inclined to adopt MaaS. This is in line with earlier research;
e.g. Ho et al. (2018) also identified very frequent car users as less likely to adopt MaaS.
Our two clusters with a most favourable attitude towards mobility integration are also the two most
multimodal clusters. Individuals in these two clusters tend to be young, highly educated people who
live in more dense urban areas and have no children. These socioeconomic characteristics have also
been found among early adopters of shared modes (Alemi, 2018; Clewlow, 2016; Dias et al., 2017;
Shaheen et al., 2012), as well as among the more general multimodal individuals (both in Europe
(Molin et al., 2016) and in the USA (Buehler and Hamre, 2015)). We also found that it is more
common among individuals belonging to these two clusters to rely on travel information for their
transport mode choices instead of solely considering their preferred or habitual mode of transport.
Indeed, multimodal individuals are known to have more complex strategies to choose transport
mode and exercise weaker travel habits (Verplanken et al., 1997). This, in turn, facilitates the
introduction of new mobility solutions such as MaaS.
We found, however, a strong difference between these two more multimodal clusters. While ‘MaaS-
FLEXI-ready individuals’ (32% of the sample) have a very positive attitude towards pooled on-demand
services, ‘multimodal public transport supporters’ (15% of the sample) strongly prefer traditional
transit over other new modes. Previous research has highlighted that public transport users are less
likely to shift from fixed public transport usage to pooled on-demand services (Al-Ayyash et al., 2016)
or to adopt MaaS (Ho et al., 2018), in line with our observations for the ‘multimodal public transport
supporters’. This can be due to the (in general) higher usage of public transport by lower income
individuals (Hensher, 1998; Ryley et al., 2014), for which the on-demand modes of transport included
in MaaS may be perceived as a premium and potentially expensive service. In fact, while ‘MaaS-
FLEXI-ready individuals’ show the highest average score regarding willingness to pay, ‘multimodal
public transport supporters’ have the second lowest willingness to pay among the five found clusters.
In the Dutch setting, having less income inequalities (Gini coefficient of 0.28 (World Bank, n.d.)) than
in the two countries of the abovementioned studies (Lebanon 0.32 and Australia 0.35 (World Bank,
n.d.)), and relatively high fares of public transport (eurostat, 2018), some current public transport
adepts (the ‘MaaS-FLEXI-ready individuals’) are arguably more open towards accepting alternative
on-demand services. This reasoning seems consistent with Hall et al. (2018), who suggest that it is
current public transport users with the higher incomes that are more open to complementing their
public transport usage with on-demand services, which is what makes adopting MaaS an attractive
alternative. In addition, the share of public transport users in the Dutch population is currently quite
low, with most individuals not using public transport on a weekly basis, even in the more public
transport minded clusters. As a result, there is potential for these users to incur a modal shift from
their car trips to MaaS. In contexts different than the Netherlands, we expect clusters that resemble
the same characteristics as the ones found in this study. The higher the percentage of public
transport users, the higher their technological capabilities and interest, and the lower their cost
sensitivity; the higher the adoption potential for MaaS will be in that setting.
This research has also shown that pooled on-demand services are more appealing than transit for
‘mobility neutralsand ‘technological car-lovers’. Pooled on-demand services can thus attract
individuals from these clusters to more sustainable mobility patterns. Similarly, pooled on-demand
services can facilitate a switch from the private car and into MaaS for areas characterised by poor
public transport, as suggested by Lavieri and Bhat (2018) for the American context.
We identify two main barriers for potential MaaS adoption: (a) high (car) ownership need as a
determinant of mode choice (for ‘technological car-lovers’ and anti new-mobility individuals’), and
(b) low technology adoption (for ‘mobility neutralsand anti new-mobility individuals’). Additionally,
clusters more inclined to keep their unimodal car behaviour showed lower environmental and
financial sensitivity. Strong sense of ownership, as well as low environmental and financial sensitivity
have also been found in literature as important variables that deter individuals from moving away
from a car-centric behaviour and into adopting new mobility solutions (Burkhardt and Millard-Ball,
2006; Efthymiou et al., 2013; Lane, 2005; Paundra et al., 2017; Zheng et al., 2009). Additionally,
Lavieri and Bhat, (2018) also found technology adoption as a relevant barrier for MaaS adoption in
the USA context. To this end, some policy recommendations tailored to each of the five latent classes
found in our analysis are described in the following subsection.
5.2 User cluster specific recommendations
Based on the results of this study, we highlight some relevant policy recommendations that can
increase the adoption of (sustainable) MaaS schemes in relation to the five clusters:
1. TheMaaS-FLEXI-ready individuals (32% of the sample) are most inclined to adopt MaaS
schemes and use pooled on-demand services thereby. These individuals are therefore more
likely to reduce their car usage in favour of other modes. Simultaneously, they can also be
expected to (slightly) lower their public transport usage by switching to on-demand services
such as pooled on-demand services, given their attitudinal preference towards the new
mode against traditional transit. Travel awareness campaigns can support the modal shift of
this cluster away from private car usage by focusing on concrete functional benefits that
MaaS can bring them (time and price benefits) while avoid a major shift from traditional
transit usage by appealing to their environmental sensibility.
2. The “Mobility neutrals (25%) are mainly composed of individuals aged 65 and older. The
analysis of technology related covariates showed how their lower technological adequacy
can prevent them from profiting from new mobility trends. Providing hybrid systems that do
not only rely on a mobile app but also include a smartcard ticket version can address this
barrier. Allowing for SMS correspondence or having a call centre for ordering purposes (even
if implemented at a small fee for the customer) can also allow that individuals with no
smartphones or internet connection can profit from on-demand services or real time
3. The “Technological car-lovers” (22%) have a car-centred attitude and behaviour, as well as a
below average environmental friendliness or cost sensitivity, making it difficult to trigger a
behavioural shift. Previous research suggests promoting new mobility modes to these
individuals solely as an alternative for the occasions in which their car is unavailable instead
of suggesting to replace it altogether (Paundra et al., 2017). This can help them experience
the new system and its novelty, which may appeal to their high technological affinity.
Attention should also be given to providing mobility alternatives that suit the needs of
families with children, more prevalent in this cluster. Additionally, measures to avoid that
young families shift towards unimodal car usage with the birth of their first child as well as
measures to facilitate a mode shift away from car-based patterns once these children grow
older can help reduce the size of this cluster.
4. The “Multimodal public transport supporters” (15%) have positive attitudes and behaviour
towards public transport usage. These individuals do not exclude new shared modes yet
(strongly) prefer scheduled public transport. Still, only around 40% of their individuals use
public transport weekly, less than the percentage that use car on a weekly basis. The
introduction of new modes can help these individuals reach destinations for which arguably
they currently need the car. As a result, their multimodal mind-set with positive attitudes
towards public transport and lower car drive can become more aligned with their future
travel patterns. Given their above average positive attitude towards transit, they can become
the most sustainable MaaS users, considering public transport as main mode and other on-
demand services as mere complements to transit when necessary. Compared to ‘MaaS-
FLEXI-ready individuals’ and ‘technological car lovers’, this cluster has a somewhat lower
technology affinity. Easy to use MaaS apps offered by trusted public transport operators can
provide a familiar and reliable environment for these individuals in their MaaS adoption
process. Their public transport card/subscription could be extended to give them access to
additional shared mobility services and enable them to try these for free. This measure could
help them overcome their resistance to innovation and does not require them to pay via an
app (which they would rather not do).
5. The cluster Anti new-mobility individuals” (6%) represents the individuals least inclined to
adopt MaaS, since they show both high psychological car ownership and low technology
adoption. Strategies applied to ‘mobility neutralsand ‘technological car lovers’ can also be
of relevance to individuals in this cluster. Still, these individuals are unlikely to adopt MaaS or
on-demand services such as pooled on-demand services in the short term. This cluster likely
represents the laggards of mobility innovations (Rogers, 1983).
6 Conclusions
The present study has identified five different clusters in relation to individuals’ inclination to adopt
MaaS based on attitudinal indicators. Special focus was given in this research to pooled on-demand
services, which exemplify the flexibility characteristics of on-demand services while accounting for
the collective mobility services, needed to meet the objectives of urban mobility (reduce congestion,
reduce parking space, increase liveability, etc.).
To this end, we first identified relevant factors regarding MaaS and designed a series of attitudinal
indicators addressing them. We presented these aspects to a representative sample of urban Dutch
population, having a valid sample size of over thousand respondents. We then performed an
exploratory factor analysis and latent class cluster analysis on the data as data reduction and
clustering techniques so as to identify homogeneous clusters. To provide a comprehensive picture of
the individuals belonging to the different clusters, we enriched our model with a series of covariates
that covered socioeconomic, mobility and technology-related characteristics.
Two of the identified clusters (‘MaaS-FLEXI-ready individuals’ and ‘multimodal public transport
supporters’, which represent 47% of the respondents) have positive inclinations towards two main
aspects of MaaS (mobility-integration aspects and mobile-application aspects). However, the
somewhat lower (despite positive) app inclination of individuals in the latter cluster, their below
average willingness to pay and their strong preference of traditional public transport over (pooled)
on-demand services by individuals belonging to this cluster, may prevent individuals of this cluster to
adopt MaaS at a first instance. Even if these two clusters are the ones with the highest shares of
public transport usage, their average car usage is still higher, indicating potential for shifts from
private car. The MaaS adoption potential on settings different from the one in this study will likely
also depend on the share of public transport users, with urban areas with higher shares of public
transport users having more individuals that are ready to adopt MaaS.
Nonetheless, before any modal shifts are materialised, enough availability of on-demand services
needs to be granted, so that these individuals can find the anticipated mobility benefits that MaaS
promises them. Also, their willingness to pay showed to be average to low. This should be taken into
account when designing the offered services. Individuals belonging to the other three clusters
presented high (car) ownership needs and/or low technology adoption, which have been identified in
this study as main barriers towards MaaS adoption and as a starting point for policy
recommendations to increase MaaS adoption by these individuals. Policy makers, public transport
operators, MaaS providers and companies entering the shared mobility landscape can use findings in
this research to evaluate the possible changes that urban settings can undergo as a result of MaaS
and provide targeted strategies to different customer segments of the population.
Even if behaviour and attitudes are closely linked, our research (pertaining to attitudes) does not
allow us to conclude to what extent those attitudes will culminate in a behavioural change or if
habitual behaviour will emerge. This is best tested in real life pilots or full launch MaaS schemes.
Also, the obtained results are dependent on the attitudinal statements presented to respondents in
the study. While we tried to cover a wide range of attitudes, some aspects such as autonomy /
perceived behavioural control, which have been previously found to be predictors of PT usage
(Anable, 2005; Hunecke et al., 2010), have not been included in the present study. In our study, we
adopted the main MaaS aspects identified in Durand et al. (2018) in defining the indicators. Further
research could consider a theoretical basis such as the Theory of Planned Behaviour (Ajzen, 1991) or
the Technology Acceptance Model (Davis, 1989) as basis for deriving the single indicators.
Future MaaS pilots could consider involving a representative sample of the population among their
participants, so as to assess mobility shifts and characteristics beyond those for early adopters. This
would enable the comparison between the expectations derived from attitudes and behavioural
intentions to actual behaviour, and could additionally help analyse the impact of MaaS for different
trip types. Given the novelty of the research topic, and to avoid overloading respondents, this
research only considered pooled on-demand services explicitly. Further research could also consider
other on-demand services different from pooled on-demand services.
This work was funded by NWO (The Organisation for Scientific Research from the Netherlands), as
part of the SCRIPTS (Smart Cities Responsive Intelligent Public Transport Services) research project.
The authors thank Anne Durand and Lucas Harms from the Kennisinstituut voor Mobiliteitsbeleid
(KiM) for their comments pertaining the design of the attitudinal indicators. We also thank Maarten
Kroesen from TU Delft for an interesting discussion regarding the set-up of the methodology of this
study. Finally, we want to thank the three anonymous reviewers for their valuable comments which
helped improve the paper.
Appendix A. Attitudinal indicators
Figure 1
Full statement in English
Source (where
Multimodal mind-set
I do not mind which transport mode I
use, as long as it suits my trip needs
I often compare different travel
options and transport modes before
choosing how to travel
Modified from
(Atasoy et al.,
Mode integration
It is essential to be able to easily
combine different transport modes
(such as bus, car, bike or car-sharing)
in order to improve transportation in
the Netherlands
Way of travel
I am willing to try new ways to travel
I like travelling always in the same
Public transport attitude
Uneasiness of
It makes me uncomfortable to ride
with strangers on public transport
Modified from
(Rubin, 2011)
PT cleanliness
I think the public transport is not so
clean or decent
It is important to use public transport
to preserve the environment
Cost saving
I choose to travel with public
transport or to share rides to reduce
my trip costs
Private car attitude
Ownership need
I would like to have the convenience
of a car without owning one myself
Modified from
(Kamargianni et al.,
Privacy need
I like the privacy in the car or bike
Modified from
(Spears et al.,
People like me only use their own bike
and/or car
Car usage vs cost
I would use the car less if there would
be a cheaper alternative
Flexibility trait FLEXI
I like that FLEXI does not have a fixed
schedule or route
FLEXI would give me the freedom to
travel where I need to be when
Reliability (FLEXI
vs PT)
FLEXI seems to me more reliable than
current public transport
I find FLEXI’s flexibility in the
departure time more convenient than
traditional transit.
FLEXI does not have fix schedules.
That would worry me.
Missed pick-up
I would be worried that FLEXI departs
without me
Modified from
(Khattak and Yim,
In-vehicle time
I would find it annoying that FLEXI
does not drive the fastest route (e.g.,
FLEXI’s route is 18 minutes instead of
15 minutes)
Modified from (Al-
Ayyash et al., 2016)
Number of stops
I would not mind if other travellers
get in or off the FLEXI vehicle during
my ride
Modified from (Al-
Ayyash et al., 2016)
Safety trait FLEXI
Safety (FLEXI vs
I would feel safer in FLEXI than in a
regular bus
Driving skills
I think that FLEXI drivers do not drive
In-vehicle safety
The proximity of a driver would make
me feel safe in FLEXI
App adoption
I would use a (smartphone) app if it
gave me access to all available travel
App literacy
It is easy for me to find FLEXI’s pick-up
point if it is displayed on a map in the
(smartphone) app
In-app payments
I like to pay for my rides via a
(smartphone) app
Willingness to
pay for
I would be ready to pay for precise
and reliable travel information
Willingness to
pay for reliable
I am willing to pay more to have a
more predictable travel time for my
Modified from
(Shiftan et al.,
Information need
I find it difficult to find information of
all available travel alternatives
Price bundling
I would prefer a monthly subscription
instead than paying individually for
each trip that I make in a month
(exclusively used for the technology
related characteristics in Section 4.3)
I try new services, such as Netflix or
Uber, before my friends and family
Modified from
(Roehrich, 2004)
I try new products, such as a fitbit or
the newest smartphone, before my
friends and family
Modified from
(Roehrich, 2004)
I often purchase new products, even
though they are expensive
Modified from
(Jensen et al.,
My family and friends usually come to
me for advice about new products
and services
Modified from
(Caiati, 2018)
I am enthusiastic about the
possibilities offered by new
Modified from
(Ewing and
Sarigöllü, 2000)
Table A. 1: Attitudinal indicators used, including their sources (where applicable) and their relation to the keywords used
in Figure 1
Appendix B. Scree plot of the Exploratory Factor Analyses
Figure A. 1: Scree plot of the EFA with (a) all mentioned indicators, and (b) only indicators loading significantly (>0.4) in
the first EFA.
Appendix C. Parameters of the final LCCA model
Mobility integration factor
Mobile application factor
FLEXI over PT factor
FLEXI intention to use
Willingness to pay factor
FLEXI concerns factor
Table A. 2: Parameters of the model indicators
Table A. 3: Parameters of direct effects
Working (voluntary work
Highest education
Exists child under 12 years
Urbanisation level
Highly urbanised
Very highly urbanised
Bike usage frequency
Bike sharing systems heard
3G bundle available
Table A. 4: Parameters of the active covariates
Acock, A.C., 2005. Working With Missing Values. J. Marriage Fam. 67, 10121028.
Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50, 179.
Al-Ayyash, Z., Abou-Zeid, M., Kaysi, I., 2016. Modeling the demand for a shared-ride taxi service: An
application to an organization-based context. Transp. Policy 48, 169182.
Alemi, F., 2018. What Makes Travelers Use Ridehailing? Exploring the Latent Constructs behind the
Adoption and Frequency of Use of Ridehailing Services, and Their Impacts on the Use of Other
Travel Modes. University of California Davis.
Anable, J., 2005. “Complacent Car Addicts”; or “Aspiring Environmentalists”? Identifying travel
behaviour segments using attitude theory. Transp. Policy 12, 6578.
Atasoy, B., Glerum, A., Hurtubia, R., Bierlaire, M., 2010. Demand for Public Transport Services:
Integrating Qualitative and Quantitative Methods, in: 10th Swiss Transport Research
Buehler, R., Hamre, A., 2015. The multimodal majority? Driving, walking, cycling, and public
transportation use among American adults. Transportation (Amst). 42, 10811101.
Burkhardt, J.E., Millard-Ball, A., 2006. Who is Attracted to Carsharing? Transp. Res. Rec. J. Transp.
Res. Board 1986, 98105.
Caiati, V., 2018. Innovation adoption and diffusion in transportation: modelling the long-term
demand for new mobility services based on platform technology, in: 97th Annual Meeting of
the Transportation Research Board. Washington D.C.
Cattell, R.B., 1966. The Scree Test For The Number Of Factors. Multivariate Behav. Res. 1, 245276.
Centraal Bureau voor de Statistiek (CBS), 2018a. Bevolking; kerncijfers (in Dutch) [WWW Document].
(accessed 11.5.18).
Centraal Bureau voor de Statistiek (CBS), 2018b. Bevolking; hoogstbehaald onderwijsniveau en
onderwijsrichting (in Dutch) [WWW Document]. URL
(accessed 11.5.18).
Centraal Bureau voor de Statistiek (CBS), 2018c. Arbeidsdeelname; kerncijfers (in Dutch) [WWW
Document]. URL
(accessed 11.5.18).
Centraal Bureau voor de Statistiek (CBS), 2018d. Regionale kerncijfers Nederland; Bevolking;
Stedelijkheidsklasse (in Dutch) [WWW Document]. URL
(accessed 11.5.18).
Centraal Bureau voor de Statistiek (CBS), 1992. Sociale kerncijfers - Nieuwe maatstaf van
stedelijkheid. Mens en Maatsch. (in Dutch).
Chowdhury, S., Ceder, A. (Avi), 2016. Users’ willingness to ride an integrated public-transport service:
A literature review. Transp. Policy 48, 183195.
Clauss, T., Döppe, S., 2016. Why do urban travelers select multimodal travel options: A repertory grid
analysis. Transp. Res. Part A Policy Pract. 93, 93116.
Clewlow, R.R., 2016. Shared-Use Mobility in the United States: Current Adoption and Potential
Impacts on Travel Behaviour. Present. 95th Annial Meet. Transp. Res. Board.
Davis, F.D., 1989. Perceived usefulness, perceived ease of use, and user acceptance of information
technology. MIS Q. 13, 319340.
Dias, F.F., Lavieri, P.S., Garikapati, V.M., Astroza, S., Pendyala, R.M., Bhat, C.R., 2017. A behavioral
choice model of the use of car-sharing and ride-sourcing services. Transportation (Amst). 44, 1
Distefano, C., Zhu, M., Mîndrilă, D., 2009. Understanding and Using Factor Scores: Considerations for
the Applied Researcher. Pract. Assessment, Res. Eval. 14, 111.
Durand, A., Harms, L., Hoogendoorn-Lanser, S., Zijlstra, T., 2018. Mobility-as-a-Service and changes in
travel preferences and travel behaviour: a literature review.
Efthymiou, D., Antoniou, C., Waddell, P., 2013. Factors affecting the adoption of vehicle sharing
systems by young drivers. Transp. Policy 29, 6473.
eurostat, 2018. Price levels for personal transport equipment, transport services, communication,
restaurants and hotels [WWW Document]. Comp. price levels Consum. goods Serv. URL
s_and_hotels (accessed 1.8.19).
Ewing, G., Sarigöllü, E., 2000. Assessing Consumer Preferences For Clean-Fuel Vehicles: A Discrete
Choice Experiment. J. Public Policy Mark. 19, 106118.
Fabrigar, L.R., Wegener, D.T., MacCallum, R.C., Strahan, E.J., 1999. Evaluating the use of exploratory
factor analysis in psychological research. Psychol. Methods 4, 272299.
Field, A., 2009. Discovering statistics using IBM SPSS statistics, 3rd editio. ed. SAGE Publications Ltd,
Givoni, M., Banister, D., 2010. Integrated Transport: from policy to practice. Routledge.
Goodall, B.W., Dovey, T., Bornstein, J., Bonthron, B., 2017. The rise of mobility as a service. Deloitte
Hall, J.D., Palsson, C., Price, J., 2018. Is Uber a substitute or complement for public transit? J. Urban
Econ. 108, 3650.
Haustein, S., 2012. Mobility behavior of the elderly: An attitude-based segmentation approach for a
heterogeneous target group. Transportation (Amst). 39, 10791103.
Haustein, S., Hunecke, M., 2013. Identifying target groups for environmentally sustainable transport:
Assessment of different segmentation approaches. Curr. Opin. Environ. Sustain.
Hensher, D.A., 1998. The imbalance between car and public transport use in urban Australia: why
does it exist? Transp. Policy 5, 193204.
Ho, C.Q., Hensher, D.A., Mulley, C., Wong, Y.Z., 2018. Potential uptake and willingness-to-pay for
Mobility as a Service (MaaS): A stated choice study. Transp. Res. Part A Policy Pract. 117, 302
Hoogendoorn-Lanser, S., Schaap, N.T.W., Olde Kalter, M.J., 2015. The netherlands mobility panel: An
innovative design approach for web-based longitudinal travel data collection. Transp. Res.
Procedia 11, 311329.
Hunecke, M., Haustein, S., Böhler, S., Grischkat, S., 2010. Attitude-based target groups to reduce the
ecological impact of daily mobility behavior. Environ. Behav. 42, 343.
Hutcheson, G.D., Sofroniou, N., 1999. The multivariate social scientist: Introductory statistics using
generalized linear models. Sage.
Janic, M., 2001. Integrated transport systems in the European Union: an overview of some recent
developments. Transp. Rev. 21, 469497.
Jensen, A.F., Cherchi, E., de Dios Ortúzar, J., 2014. A long panel survey to elicit variation in
preferences and attitudes in the choice of electric vehicles. Transportation (Amst). 41, 973–993.
Jittrapirom, P., Caiati, V., Feneri, A.-M., Ebrahimigharehbaghi, S., Alonso-González, M.J., Narayan, J.,
2017. Mobility as a Service: A Critical Review of Definitions, Assessments of Schemes, and Key
Challenges. Urban Plan. 2, 13.
Kamargianni, M., Li, W., Matyas, M., Schäfer, A., 2016. A Critical Review of New Mobility Services for
Urban Transport. Transp. Res. Procedia 14, 32943303.
Kamargianni, M., Matyas, M., Li, W., 2017. Londoners’ attitudes towards car-ownership and Mobility-
as-a-Service: Impact assessment and opportunities that lie ahead. UCL Energy Institute’s
MaaSLab Rep. Prep. Transp. London 52.
Karlsson, M., Sochor, J., Aapaoja, A., Eckhardt, J., König, D., 2017. Deliverable 4: Impact Assessment.
MAASiFiE project funded by CEDR.
Kennisinstituut voor Mobiliteitsbeleid (KiM), 2018. Stabiele beelden: Trends in beleving en
beeldvorming van mobiliteit (Trends in perception and mobility imaging).
Khattak, A.J., Yim, Y., 2004. Traveler Response to Innovative Personalized Demand-Responsive Transit
in the San Francisco Bay Area. J. Urban Plan. Dev. 130, 4255.
Lane, C., 2005. PhillyCarShare: First-Year Social and Mobility Impacts of Carsharing in Philadelphia,
Pennsylvania. Transp. Res. Rec. J. Transp. Res. Board 1927, 158166.
Lanzini, P., Khan, S.A., 2017. Shedding light on the psychological and behavioral determinants of
travel mode choice: A meta-analysis. Transp. Res. Part F Traffic Psychol. Behav. 48, 1327.
Lavieri, P.S., Bhat, C.R., 2018. MaaS in Car-Dominated Cities: Modeling the adoption, frequency, and
characteristics of ride-hailing trips in Dallas, TX. Tech. Pap. Dep. Civil, Archit. Environ. Eng. Univ.
Texas Austin,.
Li, Y., Voege, T., 2017. Mobility as a Service (MaaS): Challenges of Implementation and Policy
Required. J. Transp. Technol. 7, 95106.
Luo, Q., Dou, X., Di, X., Cornelius Hampshire, R., 2017. Multimodal Connections between Bikesharing
and Ride-Hailing: An Empirical Study in New York City.
Magidson, J.;, Vermunt, J., 2004. Latent class models 175198.
Martinez, L.M., Viegas, J.M., 2017. Assessing the impacts of deploying a shared self-driving urban
mobility system: An agent-based model applied to the city of Lisbon, Portugal. Int. J. Transp. Sci.
Technol. 6, 1327.
Matyas, M., Kamargianni, M., 2018. Survey design for exploring demand for Mobility as a Service
plans. Transportation (Amst).
Md Oakil, A.T., Manting, D., Nijland, H., 2016. Dynamics in car ownership: The role of entry into
parenthood. Eur. J. Transp. Infrastruct. Res. 16, 661673.
Molin, E., Mokhtarian, P., Kroesen, M., 2016. Multimodal travel groups and attitudes: A latent class
cluster analysis of Dutch travelers. Transp. Res. Part A Policy Pract. 83, 1429.
Morales Sarriera, J., Escovar Álvarez, G., Blynn, K., Alesbury, A., Scully, T., Zhao, J., 2017. To share or
not to share: investigating the social aspects of dynamic ridesharing, in: Transportation
Research Board 96th Annual Meeting. pp. 109117.
Oberski, D.L., 2016. Beyond the number of classes: separating substantive from non-substantive
dependence in latent class analysis. Adv. Data Anal. Classif. 10, 171182.
Oberski, D.L., 2013. Local dependence in latent class models : application to voting in elections, in:
Advances in Latent Variables [Sis 2013 Conference Proceedings]. Milan, Italy: Vita e Pensiero,
pp. 110.
Oberski, D.L., van Kollenburg, G.H., Vermunt, J.K., 2013. A Monte Carlo evaluation of three methods
to detect local dependence in binary data latent class models. Adv. Data Anal. Classif. 7, 267
Paundra, J., Rook, L., van Dalen, J., Ketter, W., 2017. Preferences for car sharing services: Effects of
instrumental attributes and psychological ownership. J. Environ. Psychol. 53, 121130.
Peek, G.-J., van Hagen, M., 2002. Creating Synergy In and Around Stations: Three Strategies for
Adding Value. Transp. Res. Rec. J. Transp. Res. Board 1793.
Pronello, C., Camusso, C., 2011. Travellers’ profiles definition using statistical multivariate analysis of
attitudinal variables. J. Transp. Geogr. 19, 12941308.
Redmond, L., 2000. Identifying and Analyzing Travel-Related Attitudinal, Personality, and Lifestyle
Clusters in the San Francisco Bay Area. University of California Davis.
Roehrich, G., 2004. Consumer innovativeness - Concepts and measurements. J. Bus. Res.
Rogers, E.M., 1983. Diffusion of Innovations, 3rd Editio. ed. The Free Press.
Rubin, J., 2011. Choosing Transit: The Influence of Past Travel Behavior, Attitudes and Habits on
Present Choices. University of California, Berkeley.
Ryley, T.J., Stanley, P.A., Enoch, M.P., Zanni, A.M., Quddus, M.A., 2014. Investigating the contribution
of Demand Responsive Transport to a sustainable local public transport system. Res. Transp.
Econ. 48, 364372.
Shaheen, S.A., Martin, E.W., Cohen, A.P., Finson, R.S., Shaheen, S.A., 2012. Public Bikesharing in
North America : Early Operator and User Understanding. Mineta Transp. Inst.
Shaheen, S., Totte, H., Stocker, A., 2018. Future of Mobility White Paper. UC Berkeley.
Shiftan, Y., Outwater, M.L., Zhou, Y., 2008. Transit market research using structural equation
modeling and attitudinal market segmentation. Transp. Policy 15, 186195.
smile mobility, 2015. SMILE einfach mobil - Pilot operation and survey [WWW Document]. URL (accessed 1.11.19).
Sochor, J., Arby, H., Arlsson, M.K., Sarasini, S., 2017. A topological approach to Mobility as a Service :
A proposed tool for understanding requirements and effects , and for aiding the integration of
societal goals, in: ICoMaaS 2017 Proceedings. pp. 187208.
Sochor, J., Karlsson, M., Strömberg, H., 2016. Trying out Mobility as a Service. Experiences from a
Field Trial and Implications for Understanding Demand. Transp. Res. Rec. J. Transp. Res. Board
Spears, S., Houston, D., Boarnet, M.G., 2013. Illuminating the unseen in transit use: A framework for
examining the effect of attitudes and perceptions on travel behavior. Transp. Res. Part A Policy
Pract. 58, 4053.
Steg, L., 2005. Car use: Lust and must. Instrumental, symbolic and affective motives for car use.
Transp. Res. Part A Policy Pract. 39, 147162.
Strömberg, H., Rexfelt, O., Karlsson, I.C.M.A., Sochor, J., 2016. Trying on change - Trialability as a
change moderator for sustainable travel behaviour. Travel Behav. Soc. 4, 6068.
Suhr, D.D., 2005. Principal Component Analysis vs. Exploratory Factor Analysis, in: SUGI 30
Tachet, R., Sagarra, O., Santi, P., Resta, G., Szell, M., Strogatz, S.H., Ratti, C., 2017. Scaling Law of
Urban Ride Sharing. Sci. Rep. 7, 42868.
Vermunt, J.K., Magidson, J., 2016. Technical Guide for Latent GOLD 5 . 1: Basic, Advanced, and
Syntax, Statistical Innovations Inc.
Vermunt, J.K., Magidson, J., 2005. LatentGOLD 4.0 User’s Guide. Statistical Innovations.
Verplanken, B., Aarts, H., Van Knippenberg, A., 1997. Habit, information acquisition, and the process
of making travel mode choices. Eur. J. Soc. Psychol. 27, 539560.<539::AID-EJSP831>3.0.CO;2-A
von Hippel, P.T., 2004. Biases in SPSS 12.0 Missing Value Analysis. Am. Stat. 58, 160164.
Weaver, B., Maxwell, H., 2014. Exploratory factor analysis and reliability analysis with missing data: A
simple method for SPSS users. Quant. Methods Psychol. 10, 143152.
Williams, B., Onsman, A., Brown, T., 2010. Exploratory factor analysis: A five-step guide for novices.
Australas. J. Paramed. 8, 113.
World Bank, n.d. GINI index (World Bank estimate) | Data [WWW Document]. URL (accessed 1.8.19).
Zheng, J., Scott, M., Rodriguez, M., Sierzchula, W., Platz, D., Guo, J., Adams, T., 2009. Carsharing in a
University Community. Transp. Res. Rec. J. Transp. Res. Board 2110, 1826.
... Data were collected through a smartphone-based stated preference survey. Alonso-González et al. [22] identified five groups of travelers in relation to their individual inclinations to adopt MaaS in the context of urban mobility. The group with the highest inclination to adopt future MaaS schemes (which is the largest) have multimodal weekly mobility patterns, while the group with the least availability is mainly comprised by unimodal car users. ...
... Travel Choice Survey Approach Parameters City/Nation [14] Bundle RP/SP M (behavioural) X London (UK) [19] Bundle RP/SP M (behavioural) X London (UK) [20] Bundle RP/SP M (behavioural) X London (UK) [21] Mode-service RP/SP M (behavioural) X Cambridge (USA) [22] Mode SP S (cluster) -The Netherland [23] Bundle/mode SP M (behavioural) X The Netherland [24] Bundle -S -- [25] Bundle/Mode SP/RP M (behavioural) X Australia [26] Mode SP/RP S (sample) -London (UK) [30,31] Bundle From the literature review, it emerges that several scientific studies about travel demand models have been proposed. However, there is the need to further develop them in order to support the quantitative estimations of MaaS on users' choices. ...
... The demand for MaaS was lower among older individuals and among individuals who lived in rich car-dependent suburban areas. According to local public transport, taxis and long-distance public transport were the most popular transport services, followed by car rentals and rides-hare services, inside the tested MaaS schemes [22]. ...
Full-text available
Urban mobility is evolving today towards the concept of Mobility as a Service (MaaS). MaaS allows passengers to use different transport services as a single option, by using a digital platform. Therefore, according to the MaaS concept, the mobility needs of passengers are the central element of the transport service. The objective of this paper is to build an updated state-of-the-art of the main disaggregated and aggregated variables connected to travel demand in presence of MaaS. According to the above objective, this paper deals with methods and case studies to analyze passengers’ behaviour in the presence of MaaS. The methods described rely on the Transportation System Models (TSMs), in particular with the travel demand modelling component. The travel demand may be estimated by means of disaggregated, or sample, surveys (e.g., individual choices) and of aggregate surveys (e.g., characteristics of the area, traffic flows). The surveys are generally supported by Information Communication System (ICT) tools, such as: smartphones; smartcards; Global Position Systems (GPS); points of interest. The analysis of case studies allows to aggregate the existing scientific literature according to some criteria: the choice dimension of users (e.g., mode, bundle and path, or a combination of them); the characteristics of the survey (e.g., revealed preferences or stated preferences); the presence of behavioural theoretical background and of calibrated choice model(s).
... Several other studies have arrived at similar conclusions (e.g., Matyas and Kamargianni, 2021;Sochor and Sarasini 2017;Tsouros et al., 2021;Zijlstra et al., 2020). Additionally, multi-modal behavior has been pinpointed as a driver for interest in MaaS (e.g., Alonso-González et al., 2020;Polydoropoulou et al., 2020). In contrast, the findings on how car ownership influences MaaS adoption diverge (see Caiati et al. 2020;Matyas and Kamargianni, 2021). ...
... For instance, Zijlstra et al. (2020) found that "early adopters are likely to be highly mobile, have a high socioeconomic status, high levels of education and high personal incomes" (p.197). Zijlstra et al. (2020) also establised that younger adults are more eager to adopt MaaS than older adults (see also Alonso-González et al., 2020;Sochor et al., 2018;Vij et al., 2020), and concluded that overall, the socio-demographic characteristics of likely early adopters of MaaS resemble the general characteristics of early adopters of new technologies (cf. Rogers, 1995). ...
Full-text available
Despite widespread interest, empirical research on how end-users perceive and use Mobility-as-a-Service (MaaS) is scarce. To address this knowledge gap, this article analyzes the end-user process of adopting a MaaS service entitled EC2B, which was launched in Gothenburg, Sweden, in the spring of 2019. The contribution to the MaaS literature is three-fold. Firstly, the article provides insight into potential end-users by describing the characteristics and motives of the studied group of adopters. Secondly, it improves the understanding of the potential effects of MaaS by outlining how the EC2B service was used and how it influenced travel behavior. Thirdly, the article informs strategies for facilitating MaaS adoption by outlining what types of drivers and barriers the end-users faced during different stages of the adoption process. The reported findings underscore previous assertions that MaaS is much more than just an app and a subscription plan and highlight a mutually reinforcing relationship between the introduction of MaaS and the implementation of policies aimed at reducing car use.
... Furthermore, the family situation was found to play a major role, as families with children used MaaS significantly less frequently and used cars more frequently [11,60,90]. In addition, higher education, regular use of car sharing, a regular income, and an ecological attitude, as well as the desire to reduce car use, influenced the intention to use MaaS [26,67,116]. Furthermore, if experience had already been gained with multimodal platforms, the likelihood that users would use MaaS increased [79] and it decreased in the absence of experience [52]. ...
Full-text available
The number of publications about mobility as a service (MaaS) has increased rapidly in the past years, spreading into various fields. In this paper, a total of 127 scientific publications about MaaS are reviewed and categorized into an overarching thematic framework in order to highlight key areas of research and further potential. Three research questions are highlighted in this review: (1) which topics are discussed in the existing MaaS literature? (2) what are the main results in the specific fields? and (3) where are gaps for further research? Publications have focused substantially on the topics of the market, users, data and technology, and the impact on the transportation system. The review shows that, regarding the concept, multi-level approaches have been established for the organization and cooperation of the actors involved, showing different levels of integration of public and private actors on a regional and supra-regional level. Various groups have already been identified as potential users, but the relatively low willingness to pay and the requirements regarding the individualization of mobility services pose problems that have not yet been solved. There is also a constant and unsolved challenge regarding the sensitive motion data that must be stored and processed. Significant research is still needed, including assessments of the impact of MaaS and what effects the service will have on the established use of transport modes, as well as how packages need to be designed and priced in order to optimally reach users.
... The ambition is that the integration of public and shared mobility services can make both services more accessible and useful to a larger user group. However, literature published on this topic notes that this can only be achieved if the barriers of making an effective scheme of shared mobility are accounted for during the integration process ( Alonso-González, Hoogendoorn-Lanser, van Oort, Cats, & Hoogendoorn, 2020 ). Some of the important factors in addressing these barriers are inclusiveness, accessibility, equity in terms of fair distribution of cost, and a citizen-oriented approach where the users' needs are central ( Machado, De Salles Hue, Berssaneti, & Quintanilha, 2018 ). ...
Full-text available
To create an integrated transport system that can compete with and reduce private car usage, we need a better understanding of the transport and user characteristics that relate to people's intentions to use shared and public transport at a mobility hub. For this purpose, this paper describes the results of a survey surrounding the case study of Leyenburg, The Hague in which a scenario of integrating shared mobility at an existing public transport stop is proposed. This study investigates the intention to use shared modes and public transport in a multimodal transport network and the factors and user characteristics that affect this intention. As digital technologies become important in the integration of modalities by offering digital planning and payment options, concerns regarding digital exclusion in transport services are growing. In this paper we developed a digital skills measure to reflects one's ability to perform tasks that are inherent to the digital services seen in the transport sector. Using an ordinal logistics regression analysis, the study has found that the intention to use shared transport is higher for people who are younger, have a high level of education and a high level of digital skills. In addition, having prior experience with shared transport in the past year and currently using multiple means of transportation during the trip are positively affecting the intention to use shared transport. The intention to combine shared transport with the bus or tram during a trip is similar to the intention to use shared transport and is related to similar characteristics, except for education. The intention to use the bus or tram is found to be mainly related to current transport usage and trip-specific factors and not to other user characteristics. For transport providers, the results provide evidence that offering shared motor scooters and bicycles would be an attractive option for young and highly-educated users who intend to combine the use of shared and public transport.
... Since the emergence of the platform economy in the mobility context, empirical evidence has been collected and analysed, revealing new phenomena that call for further, model-based analyses. Prime examples of which include the comparison of services offered by public transport and ride-hailing platforms [6], driver acceptance behaviour [7], drivers' labour choices [8], Mobility as a Service (MaaS) adoption [9], mode choice [10] and spatio-temporal demand patterns [11]. ...
Full-text available
Two-sided mobility platforms, such as Uber and Lyft, widely emerged in the urban mobility landscape. Distributed supply of individual drivers, matched with travellers via intermediate platform yields a new class of phenomena not present in urban mobility before. Such disruptive changes to transportation systems call for a simulation framework where researchers from various and across disciplines may introduce models aimed at representing the complex dynamics of platform-driven urban mobility. In this work, we present MaaSSim, a lightweight agent-based simulator reproducing the transport system used by two kinds of agents: (i) travellers, requesting to travel from their origin to destination at a given time, and (ii) drivers supplying their travel needs by offering them rides. An intermediate agent, the platform, matches demand with supply. Agents are individual decision-makers. Specifically, travellers may decide which mode they use or reject an incoming offer; drivers may opt-out from the system or reject incoming requests. All of the above behaviours are modelled through user-defined modules, allowing to represent agents’ taste variations (heterogeneity), their previous experiences (learning) and available information (system control). MaaSSim is a flexible open-source python library capable of realistically reproducing complex interactions between agents of a two-sided mobility platform. MaaSSim is available from a public repository, along with a set of tutorials and reproducible use-case scenarios, as demonstrated with a series of illustrative examples and a comprehensive case study.
... The recent trends in integrated mobility service as an outcome of increased TNCs and stable public transport infrastructure have made it possible to attract more travellers' segments towards using public transport. Particularly, those who were reluctant in using public transport due to a lack of door-to-door service are increasingly interested in the modern concept of mobility-as-as service (Maas)-an integrated mobility solution including various public and private transport service providers linked through technology powered by apps (Alonso-González et al., 2020;Ho et al., 2018;Loubser et al., 2020). However, MaaS is only in its nascent stage of development, and there are many unanswered questions about the development, execution, and success of MaaS, particularly in developing countries. ...
The purpose of this study is to develop a typology of cycle parking type preferences among a sample who have access to a cycle and travel within or to the electoral county of Dún Laoghaire-Rathdown, Ireland. Valid online survey questionnaire data were gathered from 574 respondents. First, using Principal Components Analysis, we reduced 11 cycle parking type preference variables into three cycle parking type components: Open, Locked and Guarded. Second, analysing cycle parking type preference data for each component, we implemented both a hierarchical and k-means cluster analysis to generate a five cluster solution comprising of Informal, Open, Any, Accessible and Secure cycle parking preference clusters. Third and last, we profiled our five cluster solution, examining the demographic composition, current and hypothetical mobility/cycling practices, and perceptions of cycle parking in Dún Laoghaire-Rathdown at present for each of the clusters. We conclude that our cycle parking type preference typology could be used to inform local and national cycle parking policy and planning efforts in several ways: i) targeting clusters that may yield the greatest increase in aggregate cycle ridership, ii) catering for clusters on the basis of enhancing cycle equity for those demographically underrepresented in cycle ridership and/or potentially marginalised in cycle infrastructure planning efforts, and iii) providing for clusters on the basis of strategically promoting particular cycle-activities (Cass and Faulconbridge, 2016), such as cycle-shopping, and end use practices (Spurling, 2020) that may require unique forms of cycle parking.
Mobility emerges as one of the axes on which cities base their response to the challenges of sustainability. It is a complex phenomenon where elements located at different levels interact, meaning that it is increasingly being studied from a multi-level perspective. There are numerous smart mobility initiatives around the world, although there are doubts about their generalisation. The purpose of this article is to stablish the configurations of elements that determine the degree to which a city is making changes to implement sustainable mobility solution, and to study if those are generalisable across continents. Based on the data of 60 cities from different continets provided by the Urban Mobility Readiness Index, the configurations of elements that explain both the transition towards the implementation of sustainable mobility solutions and the denial of this phenomenon are established, three in each case. QCA was applied to a model that use the multilevel perspective.The main contribution is that infrastructures maintain the pivotal role, although the joint presence of other elements is also required, including certain characteristics of the city, such as its innovative character, as well as a high population density. Individual experiences can be extrapolated between cities on different continents.
Full-text available
On-demand mobility services are promising to revolutionise urban travel, but preliminary studies are showing they may actually increase total vehicle miles travelled, worsening road congestion in cities. In this study, we assess the demand for on-demand mobility services in urban areas, using a stated preference survey, to understand the potential impact of introducing on-demand services on the current modal split. The survey was carried out in the Netherlands and offered respondents a choice between bike, car, public transport and on-demand services. 1,063 valid responses are analysed with a multinomial logit and a latent class choice model. By means of the latter, we uncover four distinctive groups of travellers based on the observed choice behaviour. The majority of the sample, the Sharing-ready cyclists (55%), are avid cyclists and do not see on-demand mobility as an alternative for making urban trips. Two classes, Tech-ready individuals (27%) and Flex-ready individuals (9%) would potentially use on-demand services: the former is fairly time-sensitive and would thus use on-demand service if they were sufficiently fast. The latter is highly cost-sensitive, and would therefore use the service primarily if it is cheap. The fourth class, Flex-sceptic individuals (9%) shows very limited potential for using on-demand services.
Full-text available
Mobility-as-a-Service (MaaS), a transport concept integrating various mobility services into one single digital platform, elicits high expectations as a means of providing customised door-to-door transport solutions. To date, the frequent claims about the positive contributions MaaS will make towards achieving sustainability goals rely on a scattering of limited yet insightful research findings. Many research questions remain unanswered, however. Are people willing to accept MaaS as a new transport service (on a daily basis)? The KiM Netherlands Institute for Transport Policy Analysis looked for answers by means of an extensive research program. In the initial exploratory phase of the research, KiM conducted an extensive literature review. The findings are presented in the report, ‘Mobility-as-a-Service and changes in travel preferences and travel behaviour: a literature review’.
Full-text available
As the Mobility as a Service (MaaS) concept attracts more interest, there is increased demand for understanding more about MaaS subscription plans. There is a gap in knowledge on how the plans should be created and what transport modes and features they should include in order to cater for the heterogeneous mobility preferences of all the socio-demographic user groups. This paper presents the design of a survey including a stated preference (SP) experiment that captures the complex decision-making process of purchasing MaaS products. Respondents are presented with repeated choices between four hypothetical MaaS plans out of which three are fixed plans and one is a menu option. This approach allows for testing people’s preferences and willingness to pay for flexibility. The attributes of the plans include transport modes and amounts, mode specific features (e.g. 10-min taxi guarantee), transferability (meaning how much of left over mode-attributes can be transferred to the next month), special bonuses (e.g. a free dinner for two) and the price of the plan. The SP is tested with a number of focus groups. Insights on two data collection applications are detailed, first as part of web-based survey, then incorporated into a smartphone-based prompted recall travel survey. The design presented in this paper can be adapted to other areas and provide valuable insights for MaaS products design and pricing.
Full-text available
Mobility as a Service (MaaS), which uses a digital platform to bring all modes of travel into a single on-demand service, has received great attention and research interest. Different business models have emerged in which travellers can either pre-pay for their mobility services bundled into a MaaS plan, or pay-as-they-go using a smart app linked to the service. This study aims to understand how large the potential market of MaaS would be if travellers are offered this one-stop access to a range of mobility services, and how much potential users might value each item included in a MaaS plan. A stated choice survey of 252 individuals administered via a face-to-face method is conducted in Sydney, Australia and a state of the art preference model is estimated to address the research questions. Results indicate that almost half of the sampled respondents would take MaaS offerings, and the potential uptake levels vary significantly across population segments, with infrequent car users being the most likely adopters, and car non-users the least. On average, Sydney travellers are willing to pay $6.40 for an hour of access to car-share, with one-way car-share valued more than station-based car-share. Estimated willingness-to-pay for unlimited use of public transport is $5.90 per day which is much lower than the current daily cap. These findings suggest a careful segmentation of the market and a cross-subsidy strategy is likely to be required by MaaS suppliers to obtain a commercially viable uptake level.
Conference Paper
Full-text available
The purpose of this paper is to shed light on the concept of MaaS and what characterizes a ‘MaaS service’, as well as to propose a topology of MaaS as a tool for facilitating the discussion of MaaS, enabling the ‘comparison of’ different services, understanding MaaS’ potential effects, and aiding the integration of societal goals into MaaS services. Based on a literature review analyzing existing definitions, and an expert workshop identifying key aspects and ascertaining service differentiations accordingly, the resulting proposed topology consists of MaaS Levels 0 to 4 as characterized by different types of integration: 0 no integration; 1 integration of information; 2 integration of booking and payment; 3 integration of the service offer, including contracts and responsibilities; 4 integration of societal goals. The levels are described in terms of their added value and further discussed regarding implications for business, society, users, and technical requirements. Then, a deeper discussion also delves into the potential in expanding upon Level 4 and ways by which services and societal goals can become more fully integrated. The proposed topology adds clarity to the discussion of such a trending topic and enables the positioning of services along the MaaS spectrum. It also deepens the understanding of why MaaS can take time to establish, and can help support the development of action plans in terms of what needs to be done depending on what type of MaaS one wants to develop. Further analysis is desirable regarding the possibilities and problems linked with the different levels of MaaS. Such an analysis is key to understanding which effects can be achieved via the implementation of different levels of MaaS services in terms of e.g. social, economic and ecological sustainability, and business potential.
Full-text available
Transportation network companies (TNCs) have introduced shared-ride versions of their ordinary services, such as UberPool or Lyft Line. The concept is simple: passengers pay less in fares for an incremental increase in time spent picking up and dropping off other riders. This paper focuses on the social and behavioral considerations of shared rides, which have not been explored as thoroughly as time and cost trade-offs in transportation. A survey of TNC users conducted through Mechanical Turk in June and July of 2016, which had 997 respondents across the United States, found that (a) users of dynamic ridesharing services reported that social interactions were relevant to mode choice, although not as much as traditional factors such as time and cost; (b) overall, the possibility of having a negative social interaction was more of a deterrent to use of dynamic ridesharing than the potential of having a positive social interaction was an incentive; (c) there was evidence that a substantial number of riders harbored feelings of prejudice toward passengers of different social class and race, and these passengers were much more likely to prefer having more information about potential future passengers; (d) most dynamic ridesharing users were motivated by ease and speed, compared with walking and public transportation; and (e) safety in dynamic ridesharing was an important issue, especially for women, many of whom reported feeling unsafe and preferred to be matched with passengers of the same sex.
Full-text available
There are a number of disruptive mobility services that are increasingly finding their way into the marketplace. Two key examples of such services are car-sharing services and ride-sourcing services. In an effort to better understand the influence of various exogenous socio-economic and demographic variables on the frequency of use of ride-sourcing and car-sharing services, this paper presents a bivariate ordered probit model estimated on a survey data set derived from the 2014–2015 Puget Sound Regional Travel Study. Model estimation results show that users of these services tend to be young, well-educated, higher-income, working individuals residing in higher-density areas. There are significant interaction effects reflecting the influence of children and the built environment on disruptive mobility service usage. The model developed in this paper provides key insights into factors affecting market penetration of these services, and can be integrated in larger travel forecasting model systems to better predict the adoption and use of mobility-on-demand services.
Full-text available
Car sharing services gain momentum as a potential alternative to various modes of transportation, including privately owned cars. This trend goes hand in hand with a renewed interest in the sharing economy, which has as essential premise that product ownership is of minor relevance. Using an online experiment, this study investigates if individual differences in psychological ownership influence the effects of well-known instrumental car attributes (price, parking convenience, and car type) on people's intention to select a shared car. Results confirmed that instrumental attributes generally impact preferences for car sharing services, and that a low psychological ownership may lead to a higher preference for a shared car under specific circumstances. This suggests that not only instrumental car attributes, but also psychological disposition, specifically psychological ownership, of potential customers need to be taken into consideration when developing measures to stimulate car sharing services in society.
How Uber affects public transit ridership is a relevant policy question facing cities worldwide. Theoretically, Uber's effect on transit is ambiguous: while Uber is an alternative mode of travel, it can also increase the reach and flexibility of public transit's fixed-route, fixed-schedule service. We estimate the effect of Uber on public transit ridership using a difference-in-differences design that exploits variation across U.S. metropolitan areas in both the intensity of Uber penetration and the timing of Uber entry. We find that Uber is a complement for the average transit agency, increasing ridership by five percent after two years. This average effect masks considerable heterogeneity, with Uber increasing ridership more in larger cities and for smaller transit agencies.
Carsharing offers access to cars and other vehicles without ownership of those vehicles. This transportation option is growing rapidly in the United States and Canada. In appropriate community settings, carsharing can increase mobility, reduce vehicle travel, and complement other transportation modes. In a TCRP project that provided a wide-ranging analysis of carsharing in North America, direct contacts with carsharing members through focus groups and a web-based survey were used to determine demographic characteristics of users, their travel patterns, and their attitudes about carsharing. Special attention was paid to why members joined carsharing organizations, how they used the services, and what they liked and disliked about carsharing. With descriptive statistics from the Internet survey and qualitative analyses of focus group results (both checked against previous literature), it was determined that carsharing appeals to individuals who can be considered to be social activists, environmental protectors, innovators, economizers, or practical travelers. Carsharing companies and their partners could conceivably increase their membership by targeting such individuals and others with certain demographic characteristics.