Introduction

Providing dense and high frequency public transport service in scattered suburban areas has always been challenging, which aggravates car dependency in metropolitan outskirts. In this context Park & Ride (P&R) facilities are introduced in low density peripheries to ease the access to public transport for car users, simultaneously relieving congestion in central areas by intercepting the vehicle flow (Tang et al.1). For P&R to attract users, the relative advantages of P&R use must exceed the convenience of driving directly to the destination.

P&R facilities first started to appear in the 1930s (Noel2) and are commonly defined as parking lots typically located on the urban fringe that provide direct access to public transport service. Nowadays, technological advancements like Mobility-as-a-Service (MaaS) can greatly improve the attractiveness and efficiency of P&R. While there is no formal definition of MaaS, the term is generally understood as a mobility service that integrates a variety of transport options into a single platform where the end user can comprehensively assess all available means of transportation, pay for the trip, receive notifications about the route, etc. Given the widespread use of mobile phones, MaaS is usually provided to end users as a mobile travel app. At the moment readily available travel apps offer route planners, where the user can see transport modes available for a selected route, and parking apps, allowing to book and pay for selected parking space. In an ideal MaaS scenario these functionalities would be integrated into a single app, where a person can book a parking space, pay for it and for the public transport ticket, know public transport schedule, etc. This would enable smooth use of public transport and private vehicle and minimize the uncertainties (lack of available parking space, broken ticket vending machine, etc) from combining different modes. However, technology acceptance and adoption has been proven to be a complex process challenged by a multitude of factors (individual’s experience with technology, perceived usefulness, social influences, etc) and their interactions (Straub3). Hence, the effect of integrating P&R into MaaS depends on the end user´s willingness to adopt MaaS.

Since in this case two services (MaaS and Park&Ride) are intertwined to mutually reinforce each other, more integrated and comprehensive approach addressing both is necessary. For example, MaaS could ease and increase the use of Park&Ride facilities while having P&R option integrated in a travel app could attract frequent car users to the platform (Mingardo et al.4; Yang et al.5; Kalašová et al.6). However, to the best of our knowledge, there are no studies that would address both topics, focusing specifically on regular car drivers from suburbs who are responsible for considerable traffic flow that P&R policies and MaaS solutions have been struggling to attenuate. Acceptability and use of associated travel apps by end users (Peng7; Han and Kim8) are key success factors for MaaS and embedded in it intelligent parking management. Therefore, it is essential to provide a better understanding of individual motivations of car users to adopt digital travel planning tools. It has been previously reported (Ye et al.9; Van’t Veer et al.10) that adoption intention among car owners is not homogeneous and can be affected by existing travel habits, so a more detailed analysis is needed.

In this paper, we aim to explore the determinants behind the intention to adopt a travel planning tool among suburban residents who combine car and public transport in their daily trips (P&R users) and those who never/almost never use P&R and drive for the entire length of their trip (car-only users). Hence, the objectives of this paper are:

  • provide insights to better understand trip characteristics and latent constructs (technophilia, gain, hedonic, and normative motives) of P&R and car-only users;

  • gauge the potential for travel app adoption in the two groups and explore whether they are motivated by different determinants;

  • guide the development and promotion of a travel app, providing recommendations that increase the overall satisfaction with user-centred travel apps aimed at promoting sustainable transport modes and improving the efficiency of mobility management policies.

To reach these objectives, this study relies on a survey conducted in Madrid Metropolitan Region that explores travel app adoption intention in peripheral areas.

The paper is structured as follows: the next section provides a review of literature related to two main topics covered in this paper—P&R use and travel app adoption. Following section (‘Methods’) provides is a brief description of Madrid Metropolitan Region as well as information on data and methodology, detailing the processes of groups’ definition, exploratory factor analysis and variable selection. Section “Results” provides the results of ordinal logit models, explaining the motivation behind the adoption of travel apps among daily P&R users and car-only users. Finally, sections “Discussion” and “Conclusions” discuss the results and provide some concluding remarks.

Literature review

The present section addresses two main issues covered in this paper: the issue of P&R use in a metropolitan context and the existing evidence on the adoption of travel apps.

In the context of ongoing urbanisation and suburban growth P&R facilities are meant to support a more efficient and sustainable mode split among suburban residents who regularly travel to the core central areas. P&R aims to reduce private vehicle use between the peripheries and central areas while increasing public transport patronage on this segment.

P&R can potentially provide several advantages for its users and society in general. User benefits include cost savings (fuel economy, reduced vehicle depreciation and vehicle maintenance costs, reduced parking fees), potential travel time savings and improved comfort levels (peak-hour driving in congested areas is avoided). On the other hand, using P&R implies transfer between modes, making inevitable additional walking, waiting, and transfer penalty (Jara-Diaz et al.11).

Societal benefits include lower congestion and pollution levels, lower parking demand in central areas, increased accessibility, and increased transit ridership (Parkhurst12; Carlson and Owen13; Duncan and Cao14; Mei et al.15). On the other hand, the shortcomings of P&R are related to the decline in station area attractiveness for other uses and possible adverse effects in terms of travel habits, namely, transfer from sustainable modes to car or induced demand in car trips as less congestion makes travelling to core area by car more attractive (Noel2; Duncan16; Meek et al.17; Parkhurst18).

Overall, maximizing P&R efficiency means supporting public transport by attracting potential users from car drivers without facilitating car use by providing (free) parking. In practice, this balance has been quite challenging to achieve, and the results of P&R introduction are often mixed. For example, Walton and Sunseri19 and Kimpton et al.20 revealed that parking availability in station proximity encouraged driving to the station among those who would otherwise access the service by foot or bicycle. Parkhurst12 stated that 70% and 56% of P&R users in York and Oxford, respectively, were former car drivers, but the remaining users diverted to P&R from bus. In Rotterdam, a stated preference survey among P&R users showed that in the absence of P&R, 23.4% of respondents would drive to the final destination, while 30.6% would use public transport. In the Hague, these figures were 19% and 37%, respectively. Additionally, it was reported that some P&R facilities were used as simple parking lots, where 6–81% of respondents would leave their car and then walk to a closely located final destination (this would typically occur on sites with employment facilities nearby) (Mingardo21).

To cope with the possible adverse effects, P&R provision and management requires innovative approaches to maximise the intended effects and minimise the undesirable ones. In this context, several authors (Seik22; Noel2; Mingardo21; Parkhurst et al.23) have advocated for a comprehensive, region-wide management of P&R facilities that would enable more efficient location choice for P&R, harmonisation of parking policies on a regional scale, and mechanisms for monitoring P&R use. All three aspects are closely related: P&R located in remote areas would intercept car drivers at an early stage of their trip but might be underused if strict parking limitations in the central area are absent (i.e., parking policies are not harmonised at a regional/metropolitan level). In contrast, the P&R facilities at the entrance to the central area would have high demand but underperform in terms of VKT (vehicle km travelled) and/or emission reduction, as trip can be mostly made by car. Regular monitoring of P&R use would allow understanding of users’ needs/expectations and means to improve the overall functioning of this complex system. In this regard, travel apps can be helpful for both policymakers and end-users. For the first, they can facilitate mobility data collection, provide insights about users’ travel behaviour, and eventually enable more efficient transport demand management. For the latter, travel apps can suggest in real-time optimal routes adjusted to specific needs of every user (Polydoropoulou et al.24). Nevertheless, this potential can only be fully exploited if travel app use becomes widespread in a given region.

Considering travel app adoption, although their advent and worldwide spread are rather recent, much research is already available on the matter (Casquero et al.25). This research is motivated by expectations that travel app use can contribute to sustainable travel choices by displaying to an app user a range of alternatives (public transport, shared modes, walking) to driving. While the potential of travel apps in reducing car use and promoting sustainable modes can be rather substantial (Gabrielli and Maimone26; Smith et al.27), it can only be exploited if an individual decides to use an app.

The goal-frame theory suggests that individuals’ motivation to use an app is defined by the need to satisfy three main “goals”: gain, hedonic and normative. An individual pursuing gain goals would expect direct benefits from app use (cost and/or travel time savings). Someone going after hedonic goals would prefer an app with entertainment features/social networking functions that can improve users’ mood. Finally, people following normative goals (or motives) would look for an app that allows them to comply with social norms and/or act morally (Dastjerdi et al.28). For example, someone driven by normative motives may cycle assuming that it is the right thing to do. Oftentimes, the goal-frame theory is complemented by other potentially influential latent factors such as technophilia (defined as personal affinity, interest and openness towards technology), her/his eco-friendly mindset, pro-car attitudes, etc. (Velazquez et al.29; Dastjerdi et al.30; Dastjerdi et al.28; Jamal and Habib31).

Considering directly observed individual characteristics, existing findings on the intention to use travel apps suggest that younger respondents, students, public transport pass holders, people living in dense urban areas, and people making frequent non-work-related trips are more prone to use trip planners (Dastjerdi et al.30; Lopez-Carreiro et al.32; Jamal and Habib31; Lopez-Carreiro et al.33). Instead, unimodal travellers are less likely to use travel apps (Lopez-Carreiro et al.33; Lopez-Carreiro et al.34).

Regarding latent factors, technophilia and gain motives are often cited as one of the strongest determinants of app use intention (Velazquez et al.29; Dastjerdi et al.30; Dastjerdi et al.28; Lopez-Carreiro et al.34). Besides, having pro-green (pro-environmental) attitudes was also positively related to app adoption intention (Dastjerdi et al.28; Jamal and Habib31).

Frequently, directly observed characteristics interact with latent factors; for example, being young is associated with greater technophilia levels (Velazquez et al.29), which in turn explains the intention to use an app (Velazquez et al.29; Dastjerdi et al.30). Dastjerdi et al.30) reported that gain motives are stronger among females and youngsters but weaker among people taking short trips and/or working in rural/suburban areas. Hedonic motives were also found to be stronger for females but weaker for frequent public transport users and people with higher levels of education. Normative motives were stronger for respondents with kids and people taking longer trips (more than 90 min) in public transport. Frequent app use was also associated with higher levels of technophilia (Velazquez et al.29; Dastjerdi et al.28).

Evidence on MaaS adoption among car users appears controversial as some studies (Fiorenze et al.35; Bahamonde-Birke et al.36; Van’t Veer et al.10) report that MaaS adoption intention is comparatively low among regular car users, but other studies contradict this finding. Ho et al.37 highlighted that generally car users had higher chances of MaaS adoption than non-car users. Among car users, people driving daily were less likely to adopt MaaS than those driving one/four days a week. Alonso-González et al.38 noted that although MaaS adoption intention and experience with travel apps was high among car owners/regular car users, they showed little interest in public transport, shared modes or pooled on-demand services. This resonates with the findings from Alyavina et al.39, where regular car users showed interest in MaaS services that imply car use (car sharing, ride hailing, routing options for driving), being reluctant to give up the car. In other words, these individuals were interested in MaaS as a tool that would ease car travelling.

Following these findings, this paper aims to expand existing knowledge on travel app adoption intention, focusing specifically on suburban residents who use cars for their daily commute. These commuters can be divided into two groups: those who use car for the whole length of their trip (car-only group) and those who use car for a part of the trip and cover the rest by public transport (P&R group). Diverting car-only users to P&R is beneficial because in this case at least part of a trip is made using sustainable modes. The analysis is centred around car-only group and P&R group for three main reasons. First, trips from peripheries represent a large part of daily commutes in many metropolitan areas, yet evidence is lacking about travel app adoption intention among commuters from suburban areas. Given that transportation options in the suburbs are limited and trips are generally longer, suburban residents may have different motivations/expectations from app use. Second, for travel apps to successfully promote a transfer to public transport among car drivers, we need to know what motivates drivers to use an app in the first place. Third, the potential of many smart mobility solutions, including intelligent parking management, depends on users’ willingness to use a service-related smartphone app. Hence, this issue should be addressed in further detail.

Methods

Madrid Metropolitan Region

Madrid Metropolitan Region is located in central Spain and has a total of 6.8 million residents. Of them, 49% live in Madrid, capital of the country and Spain´s largest city. The region can be roughly divided into one CBD and three rings radiating from it: first, residential areas inside Madrid city, then the Metropolitan ring, where some large cities neighbour low-density areas, and, last, the outer Regional ring (sparsely populated and primarily rural area). The connection between Madrid City and the rest of the region is provided by suburban rail, highways, and intercity buses (Fig. 1). On a working day, 15.8 million trips are made on average in the region (CRTM40).

Fig. 1: Transport network in Madrid Metropolitan Region.
figure 1

The region is provided with suburban rail system, metro system, LRT system, and a dense highway network.

Full size image

The average car ownership rate in the region is 575 cars/1000 inhabitants (452/1000 inhabitants for Madrid city) (OMM, 202341). According to the latest mobility survey (CRTM, 201840), around 45,1% of all trips for work/study in the region are made by car/moto, 32,3% by public transport, and 22% by walking/cycling. However, these figures vary depending on the resident´s origin and destination. Inside Madrid city, only 39% of trips for work/study are made by car, while for the trips between Madrid and city’s periphery this figure reaches 68%. Given this substantial difference and considerable share of car commute from the peripheries to the centre, Madrid Region represents a challenging case study for the analysis of P&R and travel app potential.

To attenuate daily traffic pressure from the peripheries on the CBD, the Aparca+T plan is currently being implemented in the region. Aparca+T consists of a region-wide system of P&R facilities near suburban rail stations that will be part of the regional MaaS app and public transport ticketing system. In the future, the app will allow users to calculate different routing options, book a parking space, receive real-time information on facilities’ occupancy rates and waiting time until the next public transport service. Integrated ticketing provides reimbursement of parking charges upon public transport pass validation for users parking for 5–16 h from 5 am to 4 pm. Currently (November 2023), eight Apaca+T facilities are operational. The system is expected to gradually absorb the remaining Park&Rides in the region by 2029, eventually consolidating 46,140 lots.

Meanwhile, several travel apps are operating in the region. Popular route planners such as Google Maps and Moovit calculate routes by car, public transport, walking and bike, but lack additional functionalities (like payment options). Also, there are service-specific apps for Metro, urban buses, and suburban trains. These provide reliable (mostly real-time) information about schedules and often enable payment options but do not include other transport modes in the planner. Finally, shared mobility operators have their apps (Romero et al.42).

Survey design

This study relies on a survey conducted in Madrid Region to examine travel app adoption intention in urban peripheries and assess the attractiveness of different app features and functionalities for the public. The survey was elaborated in cooperation with Madrid Public Transport Authority and covered major topics commonly addressed in the existing literature on MaaS adoption as well as several specific aspects identified during 42 semi-structured interviews with different involved actors (for details, please consult Lopez-Carreiro et al.43).

At the beginning of the survey, the respondents were informed that they are taking part in a research project aiming to develop a travel app for Madrid Region. Several images of a hypothetical app were presented to the respondents throughout the survey (Fig. 2) and accompanied by a description of its future functionalities. The set functionalities resulted from the literature review and stakeholders’ inputs during the interviews. The functionalities included real-time information, variety of payment options, different routing options, carbon footprint approximation, calculations of health benefits from using sustainable modes, and sharing routes on social media. The respondents were asked to evaluate these functionalities in terms of usefulness and convenience and assess the perceived ease of app use.

Fig. 2: Illustration of an app prototype in the survey.
figure 2

Illustrations like these were included in the survey to help the respondents visualize future app interface and functionalities.

Full size image

Data collection

In spring in 2022, the survey was disseminated using the mailing list of Madrid Public Transport Authority (CRTM), distribution of flyers with a link to the survey in different parts of the region, on-street face-to-face interviews, and snowballing method. The average response rate was 5%. In total, 6350 complete surveys were received.

Based on the information provided by the respondents, it was possible to geolocate the origin and destination of their most frequent trips. This allowed to expand the dataset by accommodating location-specific variables that reflect built environment and transport supply characteristics at both trip ends, as well as connectivity between them. Open-source data (OpenStreetMap and open data of the Regional Transport Authority) was used to calculate location-specific variables. TravelTime API calculated trip time driving for each OD pair on a typical working day.

Variables

The survey collected information about three main topics: socio-economic characteristics, latent constructs, and travel behaviour. Additionally, location-specific characteristics were included in the analysis. Altogether, these variables were considered as determinants of travel app adoption intention (Fig. 3).

Fig. 3: Research framework.
figure 3

It is hypothesized that travel app adoption intention is influenced by four large groups of variables: socio-economic characteristics, location-specific characteristics, current travel behaviour/trip characteristics, and latent constructs.

Full size image

Survey section collecting socio-economic characteristics covered standard questions about respondent’s age, gender, income level, education level, and employment status.

Latent constructs included in the survey were defined in accordance with the goal-frame theory (previously described in the Literature review). The survey contained in total 25 items which measured on a 7-point Likert scale individual’s gain, hedonic, and normative motivations (total of 13, 7, and 5 items, respectively). Additionally, this section measured on a 7-point Likert scale individual’s technophilia level (4 items). Exploratory factor analysis (EFA) was used to transform these sets of items into individual factor scores that can be subsequently introduced in the model as explanatory variables.

The four latent constructs (gain motives, hedonic motives, normative motives, and technophilia) were hypothesized as drivers for travel app adoption intention. The latter was detected in the survey by the question “What is the probability that you will use the travel app once it becomes available?” and was aggregated to a three-point scale (unlikely, neutral, likely).

The section about current travel behaviour included questions about respondent’s mode choice, car ownership, and hold of a public transport pass.

Finally, location-specific characteristics reflecting the built environment and transport supply at both trip ends were included in the analysis as they might also affect the travel app adoption intention. For example, in more dense and central locations public transport offer is usually quite diversified and, seeing these potential options, an individual might be willing to use a travel app to check whether these services are convenient for his/her trip.

For more details about the variables employed consult Supplementary Note 1.

Methodology

A multi-step approach was developed to comprehensively address the research objectives.

Since existing evidence suggests that MaaS adoption intention varies depending on individual’s travel habits (i.e., whether a person is frequent car users or multimodal traveller or public transport user), at the beginning the respondents were divided based on their current travel behaviour into two groups: daily P&R users and car-only users.

As a second step, propensity score matching (PSM) was employed to ensure that P&R and car-only groups were comparable. PSM is a statistical technique that, provided a set of characteristics in a given group (ex., socio-economic characteristics of daily P&R users), finds observations that match these characteristics from a distinct group (ex., car-only users) to form a proper comparison group. Therefore, the second group consists of car-only users who are similar to daily P&R users in terms of socioeconomic Fig. 4. Subsampling procedure characteristics and commuting patterns. Figure 4 illustrates the subsampling procedure.

Fig. 4: Subsampling procedure.
figure 4

Two groups of individuals are defined for the analysis: daily Park&Ride users and car-only users.

Full size image

Following, the differences between the groups in terms of latent constructs, location-specific characteristics at origin and destination and trip characteristics are analysed to better understand travel choices of both groups. Then, ordinal logit models are used to explain the adoption intention in both groups, highlighting their potential differences in terms of app use motivation. A summary of the methodological steps is provided in Fig. 5.

Fig. 5: Methodological steps.
figure 5

The methodology includes three main steps: first, similarities and differences between the groups are analysed using descriptive statistics, then EFA is employed to explore the latent constructs in both groups, and, finally, ordinal logit models are employed to assess the determinants of travel app adoption intention.

Full size image

In the sample collected by the survey, 231 respondents residing outside Madrid used P&R facilities daily. These respondents were defined as the P&R group in this study. Following, respondents who live outside Madrid, drive a car for their most frequent trips and never/almost never (maximum 3 times per month) use P&R were selected (\(N\) = 855). Next, propensity score matching (PSM) was employed to extract from them respondents whose socioeconomic characteristics match those of the P&R group. Eliminating differences between the groups in socioeconomic characteristics guarantees that differences in the app adoption intention between P&R users and car-only users are not due to inherently different composition of the groups, as both groups are very similar.

PSM was made using 1:1 nearest neighbour propensity score matching without replacement using logistic regression to estimate the score. The covariates used for PSM (nearest neighbour matching) were income level, gender (being female), age, education (having a university degree), occupation (being a student, unemployed or retired), and origin-destination of the most frequent trip (whether the trip is from the periphery to Madrid or within the periphery). Changes in standardized mean differences for all the covariates and maximum Empirical CDF Statistics (eCDF) are provided in Table 1. Since after PSM all standardized mean differences and maximum eCDF statistics were close to zero, the balance between P&R group and car-only users was considered adequate. A set of 231 respondents resulting from the PSM was defined as a car-only group.

Table 1 Balance statistics for P&R group and car-only users before and after propensity-score matching
Full size table

Next, exploratory factor analysis (EFA) was performed to explore the latent constructs in both groups. Both samples had 22 items per 231 observations. EFA procedure was the same for both groups: using minres as a factor extraction method with oblique rotation to allow for correlations between the factors. Items loading less than 0.4, cross-loading items, and factors with fewer than three items were dropped. The resulting factor scores for both samples were saved to be used as covariates representing respondents’ gain, hedonic and/or normative motives and technophilia levels in ordinal logit models. EFA was performed using psych package in R (Revelle44).

Finally, ordered logit models were run separately for both samples to explore the determinants of travel app adoption intention in both groups. Ordered logit models were developed specifically to accommodate dependent variables measured on an ordinal scale, suggesting that the values of the underlying variable (adoption intention) can be summarized in ordered categories (in our case, unlikely/neutral/likely use of a future app). Complementary log-log link functions were used in both models as higher categories were more likely for both groups.

Regarding the covariates, stepwise forward variable selection was applied to both models to ensure the consistency of the results. The variables found significant in explaining adoption intention while ensuring the best model fit are presented in Table 2.

Table 2 Explanatory variables in ordinal logit models
Full size table

Results

The present section is dedicated to analysing the results obtained from P&R and car-only groups and their comparison. First, a descriptive comparative analysis of both groups is provided to understand their differences and similarities. Second, the results of EFA are presented. Third, the results of ordinal logit models estimating travel app adoption intention are provided and discussed for both groups.

Park&Ride group and car-only group: descriptive analysis

Since individuals were assigned to P&R or car-only group based on their mode choice, their remaining characteristics were unknown. This section aims to provide a thorough description of P&R and car-only groups to understand their composition and ease the interpretation of the modelling results.

Since PSM allowed the car-only group to be as similar as possible to the P&R group in socioeconomic characteristics, there are hardly any differences between them in this regard (Table 3). Chi-square tests were used to confirm no statistically significant differences between the groups in these variables (for more information about test results, please see Supplementary Note 2).

Table 3 Socioeconomic characteristics of Park&Ride and car-only groups
Full size table

Nevertheless, it must be highlighted that several statistically significant differences exist between the initial car-only group (i.e., before PSM) and the final car-only group (after PSM) that is used in the subsequent analysis. Initial car-only group demonstrates a higher income level with the majority (38%) of respondents gaining 40,001–80,000€/year, higher percentage (76%) of university degree holders, and higher share (7%) of retired individuals. This means that the defined car-only group (post-PSM), while comparable to Park&Ride group in socioeconomic characteristics, is not a proper reflection of all car-only drivers. Also, car-only users that travel between the suburbs are underrepresented in the post-PSM subsample. This is an inevitable shortcoming when selecting observations that resemble Park&Ride users since they travel mostly from suburbs to Madrid and not to other suburbs.

As seen from the table, a typical P&R daily user in Madrid Metropolitan Region is 42 years old, most likely a working person who lives in the periphery and works in Madrid city. His/her annual income is slightly lower than the regional average (45,093€/year), as the majority (57%) of people in this group earn less than 40,000€/year.

Although P&R and car-only groups are very similar in socioeconomic characteristics, chi-square test reveals statistically significant differences between the groups in app adoption intention. Table 4 illustrates the distribution of adoption intention in both groups, highlighting that P&R users show greater intention to adopt a travel app than individuals from the car-only group.

Table 4 Travel app adoption intention of Park&Ride and car-only groups
Full size table

Figure 6 illustrates the origins and the destinations of the two groups, P&R users (a) and car-only users (b). Spatial distribution of the origins is quite similar in both groups: all origins are concentrated along the main highways A-2 and A-6. However, destinations of car-only users are much more spatially dispersed than P&R users. For P&R users, 65% of trips end in Madrid CBD, while for car-only users this figure is only 34%. Rather high parking fees, limited parking availability and the overall stress of navigating in the city centre might discourage P&R users from driving to their destination. These factors, however, do not seem to matter for a significant share of car-only users (34%) who drive to CBD.

Fig. 6: Origins and destinations of the sample.
figure 6

a Origins and destionations of individuals belonging to the Park&Ride group. b Origins and destionations of car-only users.

Full size image

Having street-level origin/destination data for all respondents allowed to test for possible statistically significant differences between the groups considering the built environment and transport supply at both trip ends. Mann–Whitney (for non-normally distributed continuous variables) tests were performed to detect statistically significant differences between the groups in this aspect (Table 5).

Table 5 Differences between Park&Ride and car-only groups in built environment characteristics
Full size table

Contrary to what could be expected, respondents from the car-only group live in more urbanized areas with denser street networks, higher building density and greater number of local amenities than the P&R group. However, considering destinations, P&R users travel to more urban locations than the car-only group (the number of street intersections, buildings, and local amenities within an 800-m buffer from destinations for the P&R group is higher than for the car-only group). In the context of Madrid city, this means that the destinations of daily P&R users are likely to be in areas with high parking fees, which might explain their decision to park and ride.

Furthermore, Table 6 highlights the differences between the groups regarding transport supply at both trip ends.

Table 6 Differences between Park&Ride and car-only groups in transport infrastructure supply
Full size table

Although respondents from the P&R group rarely live in the vicinity of a suburban rail station, access to a suburban rail station at their destinations is considerably better. Considering public transport coverage in respondents’ TAZs (traffic analysis zone—a geographic unit of analysis employed in transport planning and management), for P&R group only 14% of origin TAZ lies within an 800-m buffer from the station, but 45% of destination TAZ lies within an 800-m buffer from the station. Also, the distance to public transport facilities (metro station, suburban rail station, or intercity bus station serving more than 10 routes) at their destinations is shorter than the same indicator for the car-only group. Instead, the destinations of the car-only respondents are located closer to highways.

Statistically significant differences in trip characteristics between the groups are summarized in Table 7.

Table 7 Differences between Park&Ride and car-only groups in trip-related characteristics
Full size table

While aerial trip length and trip time driving, although different, are still relatively similar for both groups, there are surprisingly significant differences in average trip times reported by the respondents (32 min for the P&R group against 57 min for the car-only group). Possibly, the P&R group saves time because they manage to avoid peak-hour congestion, do not need to search for a parking lot in busy central areas, and/or they may have a more optimistic estimate of their travel time as they can use it for other purposes (reading, web browsing, etc).

Exploratory factor analysis

Following the goal-frame theory, it was hypothesised that for both groups the intention to adopt a travel app is driven by gain, hedonic and normative motives, and personal affinity to technology (technophilia). Tables 8 and 9 summarise EFA results for P&R and car-only groups, respectively.

Table 8 Rotated factor matrix for Park&Ride group
Full size table
Table 9 Rotated factor matrix for car-only group
Full size table

Although the EFA procedure was the same for both groups, the results were slightly different. The normative factor was not detected in the P&R group, probably because these individuals already practice socially approved travel behaviour (i.e., they use public transport), so normative motives are not particularly relevant for them. Instead, in the car-only group, hedonic and normative motives merged into a single factor. Respondents in this group were welcoming an app that would allow them to estimate and reduce trip-related CO2 emissions. Although respondents in this group are car-only users, they are seemingly concerned about environmental impacts of vehicle driving. Interestingly, for this group, environmental concerns (normative items) were linked with a willingness to share with others one´s achievements in emissions reductions (hedonic item) via a travel app. Surprisingly, car-only users also appreciated real-time information about public transport occupancy levels in a travel app. This may indicate that car-only users might consider other travel options if fully informed.

Travel app adoption intention

This section provides the results of ordinal logit models estimating the travel app adoption intention among daily P&R and car-only users. Tables 10 and 11 provide model results for both groups with similar explanatory power.

Table 10 Model results for travel app adoption intention—Park&Ride group
Full size table
Table 11 Model results for travel app adoption intention—car-only group
Full size table

For the P&R group, technophilia was the strongest driver behind the app adoption intention, meaning that computer-literate individuals among daily P&R users are more inclined to resort to trip planners. Quite unexpectedly, hedonic motives were inversely related to travel app use. Apparently, this group does not seek entertainment/social networking features in a trip planner. Instead, these people are motivated to use a travel app that would proportionate direct benefits (“gain_motives”), such as real-time information about occupancy levels in public transport, alerts about possible unexpected events on the route, etc. Also, among daily P&R users, women were more susceptible to adopting a travel app. Although gender gap between men and women in mode choice has narrowed in the past thirty years, women are still more inclined than men to use public transport. In the latest Madrid mobility survey, 22% of men in the region used public transport, and 45.1% used a car, while for women, these figures were 27.4% and 35.2%, respectively (CRTM, 201840). Perhaps women in the P&R group have higher adoption intention because they want to use more public transport; for that, travel apps can be helpful.

Lastly, travel app adoption intention for this group increases with the distance to Madrid (“dist_CBD”), meaning that people from distant areas are more interested in travel app use. There might be several explanations: first, longer trip distances imply greater uncertainty on the way (accidents, congestion, etc), so people from faraway areas feel a greater need for a travel app to make their trips more predictable. Second, public transport in the periphery is generally limited to intercity buses and trains, which have lower frequencies in these areas, so missing a train or bus is highly inconvenient. In these circumstances, accurate (ideally real-time) information about public transport arrival times is essential for daily P&R users living further away. Third, existing trip planners that operate in the peripheries have limited functionalities and are not fully integrated with apps operating in Madrid. As many residents of the peripheries regularly travel to Madrid, they might have higher travel app adoption intention, thinking that it would be a single MaaS integrated at a regional level, which is currently absent.

Similarly to the P&R group, gain motives are strong determinants of travel app use for the car-only group. The second strongest determinant turned out to be trip time driving (“triptime_drive”), meaning that the longer the travel time by car (for the respondent’s most frequent trip), the higher his/her travel app adoption intention. This also resonates with the model output for the P&R group, yet in this case, travel time driving is a more significant predictor than the distance to Madrid as respondents are drivers. Besides, for the car-only group factor, combining hedonic and normative motives (“hed_norm”) is positively related to travel app adoption intention. This can signify that, although people in this group are car-only users, they still have environmental concerns. Their motivation to use travel apps is partially shaped by their willingness to practice more socially acceptable (sustainable) travel behaviour. Simultaneously, they also welcome entertainment functionalities in an app. Furthermore, for the car-only group, being a student was positively related to travel app adoption, which might be related to the decentralized location of different university campuses in the region. University campuses in suburban areas are relatively well-connected by public transport to Madrid CBD. Still, their connection to suburban/peripheral areas is poor, so students driving from the outskirts (as in our sample) might experience difficulties in accessing these campuses and expect that a travel app can facilitate their trips. For the P&R model, being a student was not determining the adoption intention probably because students in this group happen to study in more centrally located campuses that are better connected to the regional public transport network.

In addition to that, built environment characteristics at the destination turned out to be significant determinants of adoption intention. Car-only users that travel to more dense and urbanized locations are more willing to adopt a travel app, although the estimate for this determinant (“buildings_dest”) is comparatively low. Instead, proximity of a destination to suburban rail station (“subrail_dest”) was rather strongly and positively related to adoption intention. This means that among car-only users, those whose destinations are near transit stations are more likely to start using a travel app, probably because they see alternative travel options and think that a travel app could assist in the modal shift. However, it should be noted that proximity to bus services was not found significant in any group. It might be because currently the combination of a car and a bus (even for bus routes operating on dedicated bus lanes) is not supported by existing P&R facilities: these are mostly located in proximity to suburban rail stations, not bus terminals/stops. As a result, the insignificance of bus proximity might be due to the absence of a service offer as such. Alternatively, it is also possible that car users are not interested in combining car and bus use because the relative advantages of a bus compared to a car are not so obvious.

Discussion

The analysis focused on comparing travel app adoption intention in two groups with similar socioeconomic characteristics and commuting patterns (most trips are from the periphery to the centre of Madrid) but different travel habits: one group gathers daily P&R users and the other car-only users.

The results suggest that the potential for travel app adoption in the Madrid Region is rather high, as 90% of the respondents in the P&R group and 81% in the car-only group were likely to adopt it. This potential is especially valuable as it stems from regular drivers living in suburban areas where the use of private vehicles is widespread. As our respondents are mostly adult working people repeatedly commuting to central areas, one could doubt their openness to new services like travel apps. Also, several previous findings revealed low willingness to use MaaS among regular car users (Fiorenze et al.35; Bahamonde-Birke et al.36; Van’t Veer et al.10). However, in our case both groups are enthusiastic about it. This can be a good sign, as some previous studies have suggested that high adoption intention can potentially be converted into more sustainable travel habits. For example, a study in Manila (Philippines) showed that about 75% of those willing to adopt MaaS expressed intention to increase public transport use (Hasselwander et al.45). Similarly, Feneri et al.46 reported that, provided with MaaS, car users were more inclined to change their transport mode than walkers or cyclists.

In line with existing evidence (Fiorenze et al.35; Alonso-González et al.38; Lopez-Carreiro et al.32; Hasselwander et al.45; Ho47), our results also suggest that multimodal travellers and public transport users are more inclined to adopt MaaS, since adoption intention was greater in P&R group than in car-only group.

Driven by gain motives, respondents in both groups welcome a travel app offering them direct benefits. As suggested by the EFA, items loading to the factor “gain motives” were essentially the same for both groups: respondents highly valued an app that would make their trips more predictable by correctly estimating travel times and alerting them about possible accidents on the route. Compared to these functionalities, monetary savings were less important, especially for P&R users. Apparently, this group considers existing travel expenses acceptable as they are already reduced thanks to public transport use. Instead, car-only are more motivated by the potential monetary savings of app use. Thus, an app offering cheaper travel options (i.e., made partially or fully by public transport) has the potential to draw regular car users by public transport.

Both models indicate significant demand for a travel app in peripheral areas. This aligns with previous findings reported by Lopez-Carreiro et al.32 indicating that longer travel times are positively related to travel app use. However, it should also be noted that in a study by Zijlstra et al.48 potential for MaaS adoption was higher in more urbanised areas. In contrast, in this study, readiness to adopt a travel app was greater among those living further away from Madrid and making longer trips. These people are likely to reside in low-density rural areas. Local authorities should exploit this potential when developing mobility policy measures, such as region-wide P&R strategies. For residents living far away from the metropolitan core, the cost of commuting by car is noticeable, which explains why car-only respondents were interested in alternative (possibly cheaper) travel options and wanted to know public transport occupancy rates. This finding highlights the potential of travel apps to attract users from remote areas. Also, it supports the location of P&R facilities in the periphery: residents from the outskirts are interested in peripheric P&R as they provide monetary savings by minimizing trip segment made by private vehicle. In remote and sparsely populated areas P&R has greater chances to provide the expected results without compromising other sustainable transport modes: there, the distances are too large to be covered by walk, cycling infrastructure is currently lacking, and densities are too low to provide frequent bus service. Instead, locating P&R facilities in consolidated urban areas or in relatively central areas is risky due to potential adverse effects: people may park but not use public transport or drive to the facility instead of walking/cycling.

At the same time, residents from the most urbanised area (Madrid city) were not included in the analysis as this paper focused on suburban residents. In general, built environment characteristics at trip origin were not found to be significant in our case, although recently it has been reported that living near a transit station was positively correlated with adoption intention (Bahamonde-Birke et al.36). Instead, connectivity between origin and destination and destination characteristics (for the car-only group) appeared to be significant predictors of travel app adoption. Compared to car-only users, P&R users live in less dense settings further away from transport facilities, but their destinations are more urbanized and better served by public transport. Thus, it is possible that built environment/public transport availability at the destination promote more sustainable travel choices and make people drive less. This highlights the importance of pursuing transit-oriented development (TOD) policies in locations that concentrate employment opportunities. As parking limitations in urbanized central areas make them inconvenient for private vehicle use, people will search for alternative modes of access (including P&R). In this moment of uncertainty (when an individual is hesitant about car as the most optimal/convenient travel choice) MaaS can play a crucial role in assisting with switching to other transport modes. In this sense, high MaaS adoption intention of MaaS might reflect individual’s skepticism about the practicality of his/her current travel habits and willingness to search for more efficient travel alternatives, which can be provided via MaaS. Although P&R scheme is not as sustainable as making the whole trip by public transport, it is still better than driving for the whole length of a trip and can be more feasible alternative for suburban residents living in low-density environments with poor public transport supply.

The remaining covariates explaining travel app adoption differ for P&R and car-only groups. As such, personal affinity to technology was positively related to travel app adoption in the P&R group but was not a significant predictor for the car-only group. Also, hedonic motives were inversely related to adoption intention for daily P&R users, but for car-only users, hedonic-normative motives significantly increased the probability of app adoption. This could be a sign of underlying discomfort that car-only users are experiencing about driving every day, and their hope that a travel app could suggest them more sustainable and entertaining travel alternatives (similarly to the results reported by Dastjerdi et al.30; Alyavina et al.39).

Overall, in many metropolitan areas MaaS deployment is still in early stages, and little is known about the actual effect of travel apps on travel behaviour, especially in the long-term (Feneri et al.46; Ho47; Alyavina et al.49). In this context of uncertainty, our paper seeks to understand the motivations behind travel app adoption intention in two different groups of car users. This understanding can help in developing user-oriented travel apps, taking into consideration the motivations of suburban residents who use cars for their commute. If MaaS meets their needs and expectations and proves to be a useful tool for their everyday individual mobility, possibly it would also be used to explore other travel options besides the car or at least in combination with it.

Conclusions

The complexity of modern mobility systems (multitude of travel options, large distances between origins and destinations, different user profiles, etc) makes their efficient management rather complex. Different policies are being implemented to improve system efficiency while simultaneously pursuing sustainability goals and supporting public transport patronage. Some of these policies have rather high complementarity potential and could reinforce each other, like MaaS and P&R. However, there is still much uncertainty about the interactions of different policies and users’ responsiveness to a variety of suggested alternatives.

The present paper analyses travel app adoption intention among daily P&R users and car-only users (i.e., those who drive a car for the whole length of their trip) to explore the determinants of app use in each case. The study focused on Madrid Metropolitan Region, more precisely, on the residents of suburban and peripheral areas that are more likely to generate large volumes of car trips to the urban centre given the limited public transport supply in the metropolitan outskirts.

A multi-step methodology was developed to achieve the study objectives. First, a car-only group was defined mirroring the socioeconomic characteristics of daily P&R users among individuals that never/almost never use P&R and use the car as a single mode for their most frequent trips. This was made to ensure that both groups are comparable, i.e., there are no significant differences between them in income, age, etc. Second, EFA was performed to explore latent variables in each group. Third, ordinal logit models explaining travel app adoption intention were developed.

Significant differences in the travel app adoption intention exist between the groups, with daily P&R users being more inclined to use a travel app than car-only users. This finding confirms that regular public transport users, including those who access the service by car, are more inclined to adopt MaaS. Hence, there are potential synergies between policies aimed at MaaS deployment and those aimed at improving public transport service.

Expectations of direct benefits from app use (gain motives) and variables reflecting longer trips were common for both groups and increased the probability of travel app adoption. Nevertheless, most drivers of adoption intention are different for daily P&R users and car-only users. Still, the adoption intention is rather high in both groups, so potentially accepting a new trip-planning tool is likely (if it satisfies users’ needs).

A limitation of our study is that not all car-only users were included in the analysis but only those matching the socioeconomic characteristics of daily P&R users. This enabled us to focus on individuals who, having similar socioeconomic characteristics and travel options, make different mode choices. Future studies could modify our selection procedure to further explore the determinants of travel app adoption in different groups (e.g., car-only users who travel within the suburbs) or employ different modelling approaches (like interaction effects, structural equation modelling, etc).

Overall, our findings suggest that, first, there are certain commonalities between suburban car drivers on which general app promotion strategy can be based, such as improvements in trip predictability, campaigns in peripheral areas, etc. Second, there are also group-specific drivers for the adoption of travel apps. These can be used in targeted awareness campaigns to maximise app’s attractiveness among current P&R users or car-only users. For example, since technophilia was a significant determinant of adoption intention for the P&R group, the new apps can be promoted via existing apps or other digital platforms.