Electric vehicle rental and electric vehicle adoption

Electric vehicle rental and electric vehicle adoption

Research in Transportation Economics 73 (2019) 72–82 Contents lists available at ScienceDirect Research in Transportation Economics journal homepage...

1MB Sizes 0 Downloads 154 Views

Research in Transportation Economics 73 (2019) 72–82

Contents lists available at ScienceDirect

Research in Transportation Economics journal homepage: www.elsevier.com/locate/retrec

Research paper

Electric vehicle rental and electric vehicle adoption a,b,d,∗




Joram H.M. Langbroek , Matej Cebecauer , Jon Malmsten , Joel P. Franklin , Yusak O. Susiloa, Peter Georénd



Department of Urban Planning and Environment, School of Architecture and Built Environment, KTH Royal Institute of Technology, Teknikringen 10, 10044, Stockholm, Sweden Transportation Research Institute, School for Transportation Sciences, Hasselt University, Wetenschapspark 5 bus 6, 3590, Diepenbeek, Belgium c Department of Civil and Architectural Engineering, School of Architecture and Built Environment, KTH Royal Institute of Technology, Teknikringen 10, 10044, Stockholm, Sweden d Integrated Transport Research Lab, School of Industrial Engineering and Management, KTH Royal Institute of Technology, Drottning Kristinas väg 40, 11428, Stockholm, Sweden e Solkompaniet Sverige AB, Västbergavägen 4, 126 30, Hägersten, Sweden b



JEL classification: O33 R49 L91

This case study describes the project Elbilslandet (The Electric Vehicle Country) in Gotland, Sweden, where the island Gotland is made “ready for electric vehicles” by providing a network of charging infrastructure and electric vehicle rental during several summer seasons. The influence of the electric vehicle (EV) rental scheme on the process towards electric vehicle adoption is investigated using the Protection Motivation Theory (PMT) and the Transtheoretical Model of Change (TTM). Moreover, the travel patterns of electric rental cars are compared with those of conventional rental cars. The main results of this study are the following: Firstly, people renting an EV are on average closer to electric vehicle adoption than people renting a conventional vehicle. Secondly, people who rent an EV are at the time of rental associated with more positive attitudes towards EVs, have more knowledge about EVs and would feel more secure driving an EV. Thirdly, EV-rental does not seem to have a large additional effect on the stage-of-change towards EV-adoption of the participants. Lastly, the driving patterns of EVs do not seem to indicate serious limitations regarding driving distance, parking time and the destinations that have been visited, as compared to the driving patterns of conventional rental cars.

Keywords: Electric vehicle adoption Car rental Transtheoretical model of change Protection motivation theory Driving patterns

1. Introduction Despite the fact that large-scale deployment of electric vehicles (EVs) has societal benefits such as improved air quality (Razeghi et al., 2016) and benefits related to the reduction of CO2-emissions (e.g. Choma & Ugaya, 2017), adoption rates in Sweden and many other countries are still low (Elbilstatistik, 2017; Rezvani, Jansson, & Bodin, 2015). Some important reasons hampering electric vehicle adoption are limited driving range (e.g. Lieven, Mühlmeier, Henkel, & Waller, 2011) and upfront purchasing cost (e.g. Carley, Krause, Lane, & Graham, 2013). Moreover, many people are not familiar with EVs, and also the fact that most car dealers exclusively sell conventional vehicles might contribute to EVs not to have broken through. Egbue and Long (2012) stated that there is a resistance among consumers against adopting new technologies that are far apart from the current norm, which might be

another contributing factor. On the Swedish island of Gotland, the project Elbilslandet (2014–2017) was launched to provide charging infrastructure to make the island “ready for the EV” and to stimulate more rental companies on the island to offer EVs for rent, besides the internal combustion engine vehicles (ICEVs) that are generally offered. Over the course of the project, an increasing number of charging facilities have been installed throughout the island, providing a network of charging points. Nowadays, a few fast chargers (50 kW) and a relatively dense network of destination chargers (< 23 kW) are available. The aim of this case study is to explore the role of electric car rental during summer vacations as a potential contributor to consumer readiness to adopt electric vehicles for everyday use. For this study, constructs from the socio-psychological theories Transtheoretical Model of Change (Prochaska, 1991), as well as the Protection Motivation

Corresponding author. Department of Urban Planning and Environment, School of Architecture and Built Environment, KTH Royal Institute of Technology, Teknikringen 10, 10044, Stockholm, Sweden. E-mail addresses: [email protected] (J.H.M. Langbroek), [email protected] (M. Cebecauer), [email protected] (J. Malmsten), [email protected] (J.P. Franklin), [email protected] (Y.O. Susilo), [email protected] (P. Georén). ∗

https://doi.org/10.1016/j.retrec.2019.02.002 Received 21 September 2017; Received in revised form 2 August 2018; Accepted 8 October 2018 Available online 14 February 2019 0739-8859/ © 2019 Elsevier Ltd. All rights reserved.

Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

Theory (Rogers, 1975), have been used. In the following section, more information about these theories is provided. Besides, the Diffusion of Innovations theory (Rogers, 2003) provided a general rationale for introducing EV rental in Gotland, as trialability of an innovation is considered to be a crucial factor. Both the perspective of the car rental company and the perspective of the end user have been taken into consideration. The Transtheoretical Model of Change (TTM; Prochaska, 1991) provides a conceptual framework for examining consumer readiness, by establishing a series of psychological stages before and after a behavioural change is implemented (stages-of-change). This allows progress towards change to be detected in survey studies, even for those who have not implemented the change of interest. By assessing consumer readiness for making a change, future demand can better be anticipated in light of a variety of explanatory factors. The Protection Motivation Theory (PMT; Rogers, 1975) defines motivation for behavioural change to depend on Threat Appraisal (assessment of the threats related to the current behaviour) and Coping Appraisal (assessment of the possibility to cope with the new behaviour). The results of these appraisals influence the motivation to change behaviour according to the PMT. The following research questions have been formulated:

Turrentine, & Sperling, 1994), whereas hedonic and symbolic characteristics got more attention in a later stage (e.g. Noppers, Keizer, Bockarjova, & Steg, 2015; Schuitema, Anable, Skippon, & Kinnear, 2013). Hedonic characteristics are these attributes that contribute to how pleasant or exiting it is to drive an electric vehicle. Symbolic characteristics are used to express self-identity and social identity and are related to the social context. Driving an electric car, for example, has a certain image and can be used to show that you belong to a group of environmentally aware people. In parallel to economic studies of electric vehicle adoption, that sometimes consider the willingness-to-pay for certain characteristics of EVs, psychological studies have concentrated on other aspects of electric vehicle adoption that have been proven to be of interest (e.g. Egbue & Long, 2012; Lane & Potter, 2007; Moons & Pelsmacker, 2012; Rezvani et al., 2015). Some of these studies focus on environmental awareness, whereas other studies focus on social norms. Social norms are perceptions of how a group or society you belong to expects you to behave. However, a similar aspect that can be found in many of these studies (e.g. reviewed by Rezvani et al., 2015) is the focus on motivation to change. When there is a motivation to adopt electric vehicles, people will be likely to adopt electric vehicles. This statement can be questionable in several ways. Firstly, electric vehicle adoption goes much slower than is to be expected following the motivation to change to electric vehicles that has been elicited from many motivational studies. Lane and Potter (2007) investigated the gap between attitudes and action in case of electric vehicle adoption and concluded that informing and educating potential car buyers is needed to increase the intention to adopt an EV. Secondly, electric vehicle adoption cannot be compared to the adoption of any conventional type of vehicle, due to the fact that range is limited and there is a need to charge EVs, which can be very time extensive. Also the development of charging infrastructure that is relatively slow due to the slow deployment of EVs (e.g. van Bree, Verbong, & Kramer, 2010) is a major issue that can refrain people from adopting an electric vehicle, even though they are motivated. Due to the fact that people are not used to have a limited range and need to solve these charging issues, people need to have insight in their own mobility patterns, as well as in the characteristics of electric vehicles and available charging infrastructure before they can make an informed decision about whether it is feasible to adopt an electric vehicle. Therefore, in this study, the transition from driving a conventional, internal combustion engine vehicle (ICEV) towards driving an EV is assumed to be a process rather than an event. In order for this transition to happen, there needs to be a motivation to change as well as a mechanism leading to moving forward in this process of behavioural change. In this study, the Protection Motivation Theory has been used for studying the motivation to change, whereas the Transtheoretical Model of Change has been used for studying the process of behavioural change. These socio-psychological theories will be described in the following subsections.

1. In which ways do people that have rented an EV differ from people that have rented an ICEV with regard to their stage-of-change towards regular electric vehicle use, their knowledge about EVs, their threat appraisal of the current transport network, and their attitudes towards EVs? 2. To which degree does one's stage-of-change towards EV use evolve over time, and does EV-rental influence this process of behavioural change? 3. Do the driving patterns of the electric rental cars in the island of Gotland differ from those of the petrol rental cars? The main aim of electric vehicle rental is its potential contribution to increased electric vehicle adoption. In this study, the process towards EV adoption among people that rented a car in Gotland is studied using the Transtheoretical Model of Change. Until now, this model has only been used in this context to describe the process towards electric vehicle adoption (Langbroek, Franklin, & Susilo, 2017) and the interplay between stage-of-change, socio-cognitive variables and monetary and non-monetary policy incentives (Langbroek, Franklin, & Susilo, 2016), in the absence of any intervention. EV rental can be considered as an intervention that might influence people's stage-of-change. The rest of the paper is structured as follows: in Section 2, a short overview of electric vehicle adoption research is given, followed by a description of the socio-psychological theories that have been used in this study. In Section 3, a short description of the case study area (Gotland) is given. In Section 4, the Methodology is described, followed by the Results of the different analyses in Section 5. In Section 6, there is a Discussion of the results, followed by the Conclusions and policy recommendations in Section 7.

2.2. Protection motivation theory The Protection Motivation Theory has originally been applied to health-related behaviour (Rogers, 1975), but more recently, it has been applied to transport-related issues as well (e.g. Bockarjova & Steg, 2014; Langbroek et al., 2017). According to this theory, motivation to behavioural change is considered as a driving factor for actual behavioural change and motivation is likely to be influenced by Threat Appraisal and Coping Appraisal. Threat Appraisal is an evaluation of potential negative consequences of current behaviour (an appraisal of the risk and consequences). Coping Appraisal is an evaluation of the beneficial effects of the new behaviour in reducing the threat, which is called response efficacy, and of the confidence of one being able to perform the new behaviour, which is called self-efficacy. If the threats connected to the current behaviour (driving conventional cars) are considered severe, if the person considers the electric vehicle to

2. Background 2.1. Prior research on electric vehicle adoption Electric vehicle adoption has been studied by a number of different research methods (e.g. Liao, Molin, & Van Wee, 2017; Rezvani et al., 2015). For example, many studies have focussed on the choice to buy an electric vehicle using the framework of utility maximization (e.g. Carley et al., 2013; Jensen, Cherchi, & Mabit, 2013; Lieven et al., 2011; Rasouli and Timmermans, 2016), stating that the selection of a certain vehicle depends on the utility that the attributes of this vehicle have to the decision maker. In earlier years, most focus is on instrumental aspects such as price, range and fuel consumption (e.g. Kurani, 73

Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

significantly decrease the negative effects connected to the current transport system and if one has confidence in him or herself being able to use electric vehicles for ones travel behaviour, motivation is likely to be high and behavioural change is relatively likely to occur, according to this theory. For example, a person thinks that driving an ICEV contributes to local air pollution causing an increased risk of serious respiratory problems (threat appraisal). The person also thinks that electric vehicles, because of the absence of exhaust gas emissions, do not contribute to local air pollution in the same degree (response efficacy). Because of the fact that this person has gained some knowledge about the range of electric vehicles in his price class, and is aware of his travel patterns as well as the available charging infrastructure, he has confidence to be able using an EV instead of an ICEV (self-efficacy).

In this study, five stages have been used rather than six. A period of 6 months has been selected for the Action and Maintenance stages because this was the period also used by Prochaska (1991). For the Preparation stage, Prochaska (1991) used a shorter time period, but because of the fact that Preparation for EV-adoption might take more time than health related process of change because of the major investment to be done, a period of 6 months has been selected there as well. 2.4. Field studies connected to electric vehicle adoption Throughout the years, some field studies have been done in order to learn about the effects of trying electric vehicles for a specific amount of time. The main rationale for using these trials rather than investigating changing attitudes and behaviour for people actually adopting an EV is the fact that, especially in the early 2010's, the group of EV-users can be considered as innovators or early adopters. This group is thought to be significantly different from main stream car users, so in order to predict how mainstream car users react on starting to use an EV, field trials with a broad scope of users should be organized (Rezvani et al., 2015). Cocron et al. (2011) performed a field trial and their finding was that despite their limited range, EVs fulfil in most of the daily mobility needs. Franke and Krems (2013) investigated range preferences using a three-month field trial and concluded that range preferences decreased after having gained three months of experience with using an EV. Before using an EV, people's range preferences are much higher than after using EVs, which indicates that people overestimate the range they think they need for their daily mobility. Based on this study, one might conclude that getting real-life experience with an electric vehicle has a positive effect on self-efficacy. This result is in accordance with other field studies in Denmark (Jensen et al., 2013) and in the United Kingdom (Graham-Rowe et al., 2012). Jensen et al. (2013) investigated the preferences for different characteristics of EVs before and after a three month trial. Significant differences were observed regarding preferences about driving range, top speed, battery life and charging. Also differences related to characteristics of ICEVs were observed, meaning that using EV does not only lead to an evaluation of EVs, but also to a re-evaluation of ICEVs. Graham-Rowe et al. (2012) collected the reflections of mainstream car drivers that used an EV during a field trial, in order to investigate the topics that people started to reflect about. Besides worries about the costs of EVs, they also had worries about the speed limitations and range limitations of the EVs that they used. Burgess, King, Harris, and Lewis (2013) concluded that trying electric vehicles is crucial for EVs to get a more positive image. Even though these field studies provide people with much more experience than renting EVs for in most cases one day, the advantage of EV-rental is that it is happening during a vacation period, implying that the passengers are in another context than in their normal daily life. This might facilitate behavioural change, as the normal driving context, which reinforces the possible habit of driving conventional vehicles, is interrupted. Generally, repetitive behaviour in a stable context creates habits which are hard to break. For example, Valeri and Cherchi (2016) found that habitual car drivers have a lower chance to change to EVs. On the other hand, habits are more likely to be changed with changing contexts (Verplanken & Wood, 2006). A second advantage of the use of rental cars is that with a limited number of vehicles, a rather large number of people can get in contact with EVs.

2.3. Transtheoretical model of change The Transtheoretical Model of Change is a socio-psychological model (Prochaska, 1991) that considers behavioural change as a process rather than as an event. This process consists of five stages-of change. Certain triggers or processes of change can push one forward to the next stage-of-change. It is also possible to go back in stage (Prochaska, 1991). In some studies, a notion is made of a sixth stage called Termination, where the new behaviour is completely embraced and there is no risk for going back. According to Prochaska and Velicer (1997), this stage does not necessarily exist for all people and has been given relatively little attention. TTM has also mainly been applied to health psychology to describe behavioural changes such as stopping smoking or starting to be more physically active (e.g. Carlson, Taenzer, Koopmans, & Casebeer, 2003; Haas & Nigg, 2009), but more recently, it has also been applied to other fields of research such as transportation science (e.g. Forward, 2014; Friman, Huck, & Olsson, 2017). For electric vehicle use, the gap between Contemplation and Preparation/Action is relatively high, because most people starting to use an electric vehicle have made a major investment in a new car. The advantage of using stage models rather than other theories is that these stage-models such as TTM are processoriented: they acknowledge the fact that behavioural change is a complex phenomenon taking place in different steps. For each step, there are different prerequisites for being able to move forward to the next step. Making use of TTM besides a motivational model such as PMT can add insights to both behavioural change as such as the process leading to this behavioural change. In this study, the possibility was offered to rent an electric vehicle, which can be considered as an intervention and a natural experiment, which implies the possibility to investigate the effects of trying out an EV for a short time on the process of behavioural change towards EV adoption. By mapping the attitudes, and perceptions of the pros, cons and self-efficacy regarding electric vehicle use to people that are in different stages-of-change towards electric vehicle adoption, more insight can be gained in how people's attitudes and perceptions develop when moving to more advanced stages of change. Based on these insights we can then formulate different plausible intervention/marketing policy to influence one's further learning process to the direction that we may want to see. In this study of electric vehicle adoption, the stages-of-change have been defined as follows: 1. Pre-contemplation: not having considered to start using an electric vehicle 2. Contemplation: considering to start using an electric vehicle 3. Preparation: planning to start using an electric vehicle within the coming 6 months 4. Action: having started to use an electric vehicle, but not for longer than 6 months 5. Maintenance: having used an electric vehicle for longer than 6 months

3. Case study area Gotland is the largest island of Sweden and is a popular vacation destination for both Swedes and foreign tourists. Approximately 58,000 inhabitants live in Gotland all year round, but in the summer season, the number of people staying in Gotland increases to around 600,000 people. There is one major town in Gotland, which is Visby. Visby harbour receives ferries from mainland Sweden, and the Gotland's only 74

Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

commercial airport is located a few kilometres from Visby. The other tourist attractions are spread out over the island, and there is also a small island, Fårö, that is accessible by ferry from the northern side of Gotland (Fårösund). Most hotels are located in Visby, but other types of accommodation (camping sites and vacation resorts) can be found throughout the island. The distance between Visby and Fårö, as well as between Visby and the southern part of Gotland, is approximately 80 km (Destination Gotland, 2017). The car rental companies that are part of this study are all located in and around Visby. Most respondents rented a car through Visby Gästhamn (Guest harbour), and a small group rented through an international rental company at the airport of Visby. The average rental period of Visby Gästhamn is generally shorter than the average rental period of Europcar. Many people renting a car from Visby Gästhamn return the car within 24 h. Because of the fact that Gotland is an island, there is an important rental market, as it is rather expensive to transport a motor vehicle from mainland Sweden. Moreover, the trips made by rental cars are relatively similar, as there are a limited number of destinations that people are likely to visit by rental car. The fact that the distances in Gotland are relatively small facilitates the use of electric vehicles with a range of around 150 km. In combination with a charging network consisting of fast chargers and destination chargers, Gotland is considered to be a feasible living lab to test electric vehicles. On the other hand, the island is large enough to make it necessary for many rental guests to charge the vehicle during a car rental. Elbilslandet can be considered as an intervention that aims to stimulate behavioural change using structural strategies as described in Steg and Vlek (2009). By providing an environment with rental cars in combination with a network of charging infrastructure and information about this network, testing an electric vehicle has been facilitated.

are any observable changes. 69 respondents participated in this followup questionnaire (44%). Despite sending three reminders, 56% of the respondents of the first wave were not found willing to participate in the second wave. Besides the questionnaires, GPS-trackers were installed in some of the cars that were included in the project. These GPS-trackers allow investigating travel patterns and providing an image about potential differences between driving patterns by ICEVs and driving patterns by EVs. GPS data was collected from 17 different vehicles, 8 out of them are electric vehicles and 9 out of them are gasoline or diesel vehicles. The data collection took place during the summer season of 2016. In total, data has been collected from 275 driving days (meaning days when a vehicle was rented and where data was collected). GPS signals were saved on average every 24th second. Combinations of GPS probes constituted specific trips based on a set of heuristics. There are two main measurement methods for extraction of the travel distance from GPS probes. The first, commonly used for low-frequency GPS probes (30 s and more) involves map-matching and path inference methods (Jenelius & Koutsopoulos, 2013; Miwa, Kiuchi, Yamamoto, & Morikawa, 2012). The second, using the Euclidean distance between two adjacent probes has been used in the literature for GPS probes with a frequency less than or equal to 30 s (Tang, Liu, Wang, & Wang, 2015). In this study with GPS data with an average probe frequency of 24 s, the Euclidean distance is used. The travel time of a trip has been operationalized as the difference between the timestamps of the first and last probe of this trip. Parking time has been operationalized as the time gap between two adjacent trips, in case more than one activity was carried out. The trip analysis reveals that there were some errors with longer waiting times (e.g. at a signalized intersection or a crossroad, queues in front of a ferry etc.) causing trips to be split into two trips. There is still no clear definition of what a “trip” is (Bricka, Sen, Paleti, & Bhat, 2012) and this definition varies from study to study. The threshold beyond which it is assumed that an activity took place varies in the literature between 45 and 300 s (Schuessler & Axhausen, 2009). Also a more extreme value of 900 s was used in Schuessler and Axhausen (2009). Too small values can result in too many wrongly detected activities and shorter trips, whereas too large values can lead to missing activities. After considering the quality and frequency of the data, combinations of trips with a departure time less than 100 s after the arrival time of the previous trip have been considered as one trip, as that time is too short for most trip purposes or activities. In addition, for the road where queues for the ferry to the island of Fårö in the north of Gotland builds up, the 900 s rule for trips aggregation was applied. Besides this adjustment, only trips longer than 100 m have been considered which is a reasonable threshold with respect to the GPS error that can be larger than 30 m (Schuessler & Axhausen, 2009). This trip length is too short for most trip purposes and is considered to be a reasonable lower limit. Using the processed GPS data, different variables have been compiled such as travel distance and travel time for each trip, as well as parking time between trips. Afterwards, the total travel distance, total travel time and total parking time for the day have been computed for each driving day. The variables concerning parking time have only considered driving days where more than one activity was carried out.

4. Methodology 4.1. Data collection For this study, a two-wave survey was designed. The survey questions are included in the Appendix. The data was collected using CAWI (Computer Assisted Web Interviewing). All people renting a car got an invitation letter with a link to get access to the first wave of the questionnaire. The first wave was connected to the car rental and consisted of questions about the experiences with the car rental. Besides, sociocognitive questions related to constructions from the Transtheoretical Model of Change and the Protection Motivation Theory, as well as behavioural questions and socio-economic questions were included. In the second wave the socio-cognitive questions of the first wave were repeated in order to investigate whether any changes had occurred. The participants of the first wave were invited to participate in the second wave by sending invitations to their e-mail addresses. A maximum of three reminders was sent. The participants were recruited during two summer holidays, creating two cohorts. For the first cohort, the second wave took place 12 months after the car rental. For the second cohort, the second wave took place 6 months after the car rental.

• Summer 2015: Cohort 1, car rental and wave 1 of the survey • Summer 2016: Cohort 2, car rental and wave 1 of the survey • September 2016: Cohort 1, wave 2 of the survey • February 2017: Cohort 2, wave 2 of the survey

4.2. Sample description In Table 1, a description of the sample is provided. In total, 158 respondents participated in the survey connected to the car rental in Gotland. Among the respondents, there are relatively few women and relatively many persons with a university degree (45%), but as there is no reference data about the socio-demographic characteristics of car rental guests, it is difficult to say whether the sample is representative for the population of rental guests in Gotland. On the other hand, the gender distribution is very similar between the respondents who rented an EV and the respondents who rented an ICEV. Furthermore, the age

Right after the car rental in Gotland, the respondents (N = 158 for 2015 and 2016) were asked to participate in a survey in which they answered to socio-cognitive, behavioural and socio-economic questions. This survey was available in Swedish and in English. After 12 months (for the first cohort)or 6 months (for the second cohort), they were asked to participate in a follow-up questionnaire where the sociocognitive questions were repeated in order to investigate whether there 75

Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

Table 1 Sample description. Variable


Values all respondents

Values respondents (renting EV)

Values respondents (renting ICEV)

Gender Number of cars in the household

women # (%) 0-% 1-% 2-% 3 or more - % % of respondents average # of persons % with university degree % of respondents > 800,000 SEK % of respondents working average (in years)

24% 15% 31% 39% 15% 52% 3.08 72% 45% 87% 46 years 158

24% 13% 32% 37% 14% 56% 2.99 70% 44% 88% 46 years 114

26% 21% 28% 42% 9% 44% 3.30 74% 47% 84% 46 years 44

Living in single family house Household size Higher education Household income Occupation Age Number of repondents

distribution is rather broad with respondents between 18 and 78 years old, with an average age of 46 years. The largest part of the respondents is working or has an own company (87%). Most respondents are part of families with children, with 60% of the households consisting of more than 2 persons. The respondents having rented an electric vehicle have very similar socio-economic characteristics compared to the respondents having rented a conventional vehicle. However, relatively many respondents having rented an EV live in a single family home (56% for EV rental guests versus 44% for ICEV rental guests), even though this difference is not statistically significant (Chi-square test, p-value = 0.207).

experience with the EV was better than it would have been with a comparable ICEV. Thirty per cent thinks the driving experience was equal and 11% it was worse. Nevertheless, 37% of the respondents experienced range limitations having an influence on the perceived mobility during the EV-rental. During the rental, 20% of the respondents did not experience any range anxiety, and 57% experienced a minor level of range anxiety. As some of the rental cars that were used in 2016 had a larger battery capacity (30 kW instead of 24 kW), a higher range is available to the rental guests renting these vehicles. The experience among this group seems to be more positive. Only considering this group (N = 41), 70% thinks the driving experience is better than it would have been with an ICEV and 28% thinks it is equal. As the range of EVs is an important factor for positive experience, the expected increasing range of new EVs the coming years might positively affect customer satisfaction with EV rental. In a few open questions, the respondents got the possibility to specify why they rated their experience as they did. The most often mentioned point was the range of the EV being a challenge, which requires more planning and in some cases sacrificing some prior plans. As positive elements were mentioned that the EV is quiet, easy to drive (automatic transmission) and that it makes the car rental a little bit more adventurous. Also symbolic elements such as feeling to drive in a cleaner vehicle were mentioned. Continuous monitoring is important and this increased the insight of respondents in their own travel patterns, but also insight in their driving style. Some participants also mentioned that they discovered that driving more smoothly increases the available range.

4.3. Data analysis After a descriptive analysis of the data, the data about the sociocognitive characteristics has been analysed using hypothesis tests investigating the connection between several variables. For investigating whether the mean value differs between two groups, t-tests have been used for ratio variables, whereas for ordinal variables (such as stage-ofchange), the non-parametric Mann-Whitney U test has been used. These tests are used in order to investigate whether two groups score significantly differently on a certain variable. For the socio-cognitive constructs Threat Appraisal, Response efficacy, Self-efficacy, Pros, Cons and knowledge about EVs, a standardized (averaged) summated rating scale has been used (Spector, 1992). Even though this scale is based on ordinal variables (7-point Likert scale questions), the number of possible values is so large that these composed variables are considered to be ratio variables with a minimum of 1 and a maximum of 7. Also for the analysis of GPS-data, independent t-tests have been used to investigate differences between the average driving patterns of EVs and ICEVs.

5.2. Who rents EVs The participants of this study have either rented an EV or an ICEV on the island of Gotland during the summer months of 2015 (49 respondents) or 2016 (109 respondents). After the rental, they participated in a survey where they answered questions about their attitudes towards electric vehicles and about their stated stage-of-change according to the stages of the Transtheoretical Model of Changed described in Section 2.3. Table 2 shows the division of the different respondents over the different stages-of-change. Table 2 shows that those who have rented an electric vehicle are more likely to be in a more advanced stage in the process of behavioural change and this difference is statistically significant (Mann-Whitney U

5. Results 5.1. General experiences with EV-rental The participants of this study that rented an electric vehicle were asked about their overall impression of the rental, as well as some specific aspects regarding the range of the EV, experienced range anxiety and the quality of charging facilities. Generally, people were satisfied with the EV-rental. Combining the results of 2015 and 2016, 59% of the respondents think that the driving Table 2 Stage-of-change and type of rental vehicle.

EV rented ICEV rented Total







13 (11.4%) 12 (27.3%) 25 (15.8%)

78 (68.4%) 29 (65.9%) 107 (67.9%)

17 (14.9%) 2 (4.5%) 19 (12.0%)

4 (3.5%) 0 (0%) 4 (2.5%)

2 (1.8%) 1 (2.3%) 3 (1.9%)

114 44 158


Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

5.3. Driving patterns of EVs vs ICEVs

Table 3 Socio-cognitive constructs and choice of rental car.

Threat appraisal Response efficacy Self-efficacy Pros EVs Cons EVs Knowledge

Mean values (EV)

Mean values (ICEV)


5.144 6.085 2.944 4.789 3.739 3.291

5.193 6.080 2.826 4.535 4.384 2.462

0.805 0.806 0.594 0.227 0.016 0.001

In the evaluation of whether electric vehicles are suitable as a rental car, it is important to make sure that these vehicles can be used for similar trips, in order to visit the different tourist destinations on the island. Moreover, it is important to make sure that the charging time of EVs does not form a major limitation for the flexibility of the user. In order to investigate these questions, the travel patterns of both electric rental cars and fossil fuel rental cars have been analysed. Generally, it can be concluded that there are no major differences in the destinations visited by electric vehicle rental guests and the destinations visited by rental guests renting ICEVs. Both groups have made trips across the whole island, and the EVs have also been used for trips where the car has to be charged on its way. Fig. 1 shows the destinations of the GPS-tracked trips having been made in Gotland. Because all people having rented a car started their first trip in Visby, we can conclude that both EV-trips and ICEV-trips have been made across the whole island, even in the far northern and southern parts of the island. Fig. 2 shows the chosen routes. Even though the differences are not very large, it seems like the electric vehicles have driven somewhat straighter stretches, using more often the main roads rather than making detours. This could be due to the fact that more people using EVs want to reach their destination using the shortest path in order to make sure that the driving range is sufficient. In order to test whether the travel patterns of people using electric vehicles and conventional vehicles are the same, hypothesis tests have been used. The variables that have been tested are the number of trips per day, total travel distance, maximum travel distance (the longest trip per day), total travel time, maximum travel time, total parking time and maximum parking time. Table 5 and Table 6 show the mean daily values of the EVs and the mean daily values of the ICEVs, as well as the pvalues for the independent t-tests that have been carried out. In Table 6, a subset of the trips is analysed only containing total daily distances longer than 30 km, because of the fact that some renters made very short trips. One of the potential reasons for this is the fact that they only return the vehicle at the end of the rental period. The people having rented electric vehicles have made fewer trips than the people having rented conventional vehicles, but the average distance is slightly higher. Moreover, those who have rented an electric vehicle have on average a lower parking time, which implies that they use the car more intensively. In Table 7, the mean values for detour factors are shown. The detour factor is defined as the deviation (in %) between the selected route between an origin and a destination and the shortest route over the network. Generally, there is no significant difference of the detour factors, neither for all trips nor for trips longer than 50 km. However, for chained trips between Visby and Fårösund/Fårö (see Fig. 3 for a map of the northeaster side of the island), ICEVs have a significantly higher detour factor than EVs. Of the trips by EV, 55% have made no or a negligible detour (less than 5 per cent compared to the shortest distance over the network). Of the trips by ICEV, only 37% has made no or a negligible detour. The trips by EV with a detour of maximum 30% account for 78% of the trips. For the ICEV, this is only 69%. A possible explanation for the difference in detour on this stretch is the fact that the distance between Visby and Fårö is relatively long (approximately 80 km from Visby town to the northern end of Fårö).

test, p-value = 0.004). Relatively few respondents having rented an EV are in the Pre-contemplation stage, and relatively many respondents are in the Preparation stage. This implies that there are self-selection effects: those people that already consider changing to an EV are more likely to take the step to rent an EV on the island of Gotland. Therefore, this project might be less well suited as a first encounter with an electric vehicle. Other measures are needed to be more successful to reach people that are in the Pre-contemplation stage. Langbroek et al. (2017) found that people that are in more advanced stages-of-change have higher self-efficacy to start using EVs and they are more positive towards electric vehicles. The question is whether people that rented an EV, on average being in a more advanced stageof-change, also have different values on these socio-cognitive attributes than people that rented an ICEV. In Table 3, a comparison is made of attitudinal variables between people that rented an EV in Gotland and people who rented an ICEV in Gotland. These variables have been measured using Likert-scale questions that were combined using an averaged summated rating scale (as in Spector, 1992). For these comparisons, independent t-tests have been used. Following Table 3, it appears that the respondents having rented an EV are less negative towards the perceived disadvantages of EVs than the respondents having rented an ICEV. In the survey, it was asked how people perceive the additional cost of use and the range of EVs. The group having rented an EV on the island of Gotland perceived on average that EVs are costing less and that the range is less of a problem. The people having rented an EV also have higher levels of knowledge about EVs. This is both about perceived knowledge (indicating how much the respondents think they know about EVs) and the number of electric vehicle brands they know. Regarding the other socio-cognitive variables (Threat appraisal, Response efficacy, Self-efficacy and Pros EVs), no significant difference has been found between the group having rented an EV and the group having rented an ICEV. This might be due to the relatively small sample size, but also the absolute differences between the mean values are relatively low. When investigating the answers to specific Likert-scale questions (see Table 4), there are indicators for people that have made the decision to rent an electric vehicle rather than an ICEV to be more positive towards EVs and to have more knowledge about EVs. For these comparisons, again independent t-tests have been used. The people having rent an electric vehicle in Gotland also feel more save to use EVs, which might be related to the fact that they have more knowledge about the available policy incentives and the available charging infrastructure.

Table 4 Indicators at question level. Indicator

Value EV renter

Value ICEV renter


I know much about EVs [1–7] It is more expensive to drive an EV than to drive an ICEV [1–7] I would feel secure to drive an EV [1–7] I know about which benefits that are available if I would want to switch to an EV [1–7] I have no idea where I could find EV charging infrastructure [1–7]

3.781 2.790 5.620 3.425 3.982

3.023 3.660 4.909 2.386 4.864

0.021 0.005 0.020 0.001 0.028


Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

Fig. 1. Destinations EV trips and ICEV trips.

Fig. 2. Chosen routes ICEVs and EVs.


Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

5.5. EV rental and the process of change towards electric vehicles

Table 5 Comparison travel patterns EV-rental and ICEV-rental. Variable

Mean daily values EVs

Mean daily values ICEVs


Number of trips Total distance Average distance Maximum trip distance Total parking time Maximum parking time

6.62 trips 135.2 km 21.1 km 51.9 km 4h27 m 2h14 m

6.76 trips 115.2 km 15.8 km 41.7 km 4h58 m 2h34 m

0.707 0.016 0.000 0.000 0.070 0.080

It has been hypothesized that electric vehicle rental contributes to the process of behavioural change towards electric vehicle use, because real-life experience with EVs provides more insights into the specific characteristics of these vehicles and special use requirements concerning charging and driving range. In order to measure to which degree the process of behavioural change is influenced by EV-rental a follow-up questionnaire was used. For the cohort that rented a vehicle in 2015, the follow-up questionnaire was sent one year after the EVrental (September 2016), whereas for the cohort that rented a vehicle in 2016, the follow-up questionnaire was sent six months after the EVrental (February 2017). In 2015, 49 respondents participated in the first survey. Forty of them rented an electric vehicle, whereas nine of them rented a conventional, internal combustion engine car. In 2016, 125 respondents participated in the first survey. 85 of them rented an EV, 26 an ICEV and four respondents did not reply to the question about the type of vehicle rented. The follow-up questionnaire has been answered by 27 respondents from the 2015 cohort, out of them 24 having rented an EV. From the 2016 cohort, 39 respondents participated in the follow-up questionnaire, out of which 29 had rented an EV in Gotland during 2016. Table 9 shows the development of stage-of-change of the respondents that participated in both the first survey and the follow-up questionnaire. Out of 53 participants, 31 did not change stage-ofchange. 10 respondents went backwards in the process of behavioural change, whereas 12 respondents made a progress. The number of regular EV-users among the participants of this study renting an EV increased from four respondents directly after the car rental in Gotland to six respondents after a period of 12 or 6 months after the rental. Only thirteen respondents who rented an ICEV participated in the follow-up questionnaire. Ten out of them did not progress their stageof-change. Three respondents went forwards: two respondents started using an EV and one respondent progressed from the Pre-contemplation towards the Contemplation stage. Besides the question about the stage-of-change, the follow-up questionnaire also contained socio-cognitive questions in order to measure to which degree people have changed their attitudes towards EVs (Pros and Cons) and whether people have changed their opinions about the problems of the current transport system (Threat appraisal) and the possibility of EVs to solve this problem (response efficacy), or ones self-efficacy or confidence in oneself to cope with the characteristics of the electric vehicle. For the statistical analyses, paired sample ttests have been used. In Table 10, the results of the different paired sample t-tests are shown. The results show the differences in the mean values for the different aspects between the measurement during the car rental and the measurement 6–12 months after the car rental. If the mean value is positive, the score has increased in the after measurement. If the value is negative, the score has decreased. Table 10 shows that there are no statistically significant differences for the socio-cognitive variables. Despite the fact that the sample was small, which makes it hard to draw conclusions, it is clear that no big differences have been registered between the situation during the car rental and after the car rental. The knowledge of people has not significantly increased after renting an EV and neither has the self-efficacy increased. Therefore, the experience of renting an EV on Gotland does not seem to have a major impact, despite the fact that there were some movements in the stage-of-change and a slight trend moving forward. There are even indications for a backward trend: the knowledge level has decreased in the after measurement compared to the measurement during EV-rental.

Table 6 Comparison travel patterns EV-rental and ICEV-rental (total distance > 30 km). Variable

Mean daily values EVs

Mean daily values ICEVs


Number of trips Total distance Average distance Maximum trip distance Total parking time Maximum parking time

7.03 148.8 km 22.9 km 56.9 km 4h24 m 2h6m

7.96 147.0 km 19.3 km 52.6 km 5h9m 2h32 m

0.014 0.811 0.001 0.084 0.006 0.024

Table 7 Detour factors. Variable

Mean value EV

Mean value ICEV


Detour factor (all trips) Detour factor (trips > 50 km) Detour factor (chained trips VisbyFårö/Fårösund)

34.9% 17.5% 17.9%

35.0% 20.4% 25.6%

0.966 0.411 0.003

From Tables 5 and 6, there are no indications that these people who have rented an electric vehicle use more time to charge their vehicle at tourist destinations. This result is in coherence with a survey question concerning where the participant has charged his or her electric vehicle (see Table 8). 62 respondents (79%) have indicated in the survey that they have charged their vehicle during the rental period. The most popular way of charging was using the fast charging stations (77% of charging events), and after that tourist destinations were the most popular charging location. Some people having rented the EV for several days have also charged their car at the hotel or campsite (see Table 8). Some EV-users charged their EV at different locations during the rental period. 5.4. Range anxiety and driving distance In the survey, some questions were included about the experiences of the participants with the electric vehicle rental. The answers to the survey questions have been linked to the GPS data in order to investigate the links between the experiences of the participants on the one hand and the more quantitative variables such as total travel distance and parking time on the other hand. For 201 driving days, data from surveys could be linked to the available GPS-data. Most respondents rented a car during one day, but there are some respondents that rented a car during up to 6 days. There seems to be a positive relation between range anxiety and total driving distance. Participants, who felt that the battery of the EV was not sufficient for the trips they wanted to make, travelled over a significantly longer distance (ANOVA, p-value 0.011). Range anxiety seems to be connected to long travel distances, which implies that there are no indications that people with range anxiety drive less.

6. Discussion This case study shows that electric vehicles seem to be rather 79

Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

Fig. 3. Map of northern Gotland and Fårö.

vehicle in Gotland, it can be noticed that there are relatively few persons that are in the Pre-contemplation stage and many in the Preparation stage. This implies a self-selection effect: those who already consider changing to EV-use are more likely to rent an EV. The other side of the coin is the finding that there is not such a clear pattern of people having rented an EV to be more likely to move forward in the process of behavioural change. It is believed that the results of this study regarding the socio-cognitive factors can be transferred to different contexts. However, as Gotland is an island that has a limited number of destinations that tourists are likely to visit, it is relatively easy to build a system in order to provide e-mobility in this specific context. If you rent EVs in different locations, for example in mainland Sweden, there would be more potential destinations. For rental companies, it would be harder to give information about whether an EV could be a feasible option for somebody who wants to rent a car. This study has been carried out based on people that rent a car from specific car rental companies in Gotland. Some contextual variables could not be controlled for. For example, it has not been registered what time the vehicle was collected and what time it was returned. As cars rented from the airport of Visby are often being rented for a longer

Table 8 Charging behaviour EV-renters. Answer category

# Events



Did not charge EV Charged EV Charged: used fast charging Charged: at tourist destination Charged: hotel/camping

16 62 48 18 8

21% 79% 77% 29% 13%

% % % % %

of of of of of

respondents renting EV respondents renting EV people that charged EV people that charged EV people that charged EV

feasible as rental cars on the island of Gotland. The overall assessment by car renters is rather positive and the GPS-tracks show that the driving patterns of electric vehicles do not significantly deviate from those of conventional cars. There are some indications that the density of fast chargers should be increased, because of the fact that fast charging was the most frequently used way of charging. Maybe, this is connected to the character of car rental in Gotland: most people have rented a car for one day or 24 h, in which they want to see as much as possible from the island. Making long stops at destination chargers is less attractive for this specific market of car renters. When considering the groups of people who rented an electric Table 9 Stage-of-change at EV car rental and after car rental. During | After






Total during EV rental

Pre-contemplation Contemplation Preparation Action Maintenance Total after EV rental

2 3 0 0 0 5

4 24 6 1 0 35

0 3 4 0 0 7

0 0 0 0 0 0

0 1 2 2 1 6

6 31 12 3 1 53


Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

Table 10 Stage-of-change during and after car rental. Variable

Mean values




Stage-of-change during car rental - stage-of-change after car rental Threat appraisal during car rental – Threat appraisal after car rental Response efficacy during car rental - Threat appraisal after car rental Self-efficacy during car rental - Self-efficacy after car rental Pros during car rental - Pros after car rental Cons during car rental - Cons after car rental Knowledge during car rental – Knowledge after car rental

−0.118 −0.212 0.045 0.112 −0.070 −0.031 0.274

−1.134 −1.356 0.232 0.787 −0.474 −0.190 1.736

67 65 65 64 64 64 64

0.261 0.180 0.817 0.434 0.637 0.850 0.087

time, driving patterns are likely to be significantly different from the cars being rented from the Guest Harbour of Visby. There are some specific places that are popular to visit and rather far away from Visby, but in case of renting the car for more time, those places will be likely to be visited on one of the days, whereas other, maybe more nearby locations, are visited the other days. Due to a limited amount of data from car rentals that are not from Visby Guest Harbour, these GPS-tracks have not been incorporated in this analysis. The number of kilometres driven is influenced by several factors, out of which the time of rental is important. People renting a car for a few hours are not able to drive to the other side of the island and back. However, due to the fact that there is no information about the exact hours of car rental, and due to the fact that the aim of the analysis of driving data is to compare EVs with ICEVs, it has been assumed that these rental patterns are similar for people renting EV as for people renting ICEVs. The number of respondents (N = 158) was limited, which might have an influence on the representativeness of the study, even though no large differences between the group renting an EV versus the group renting an ICEV were found. Nevertheless, especially the high percentage of drop-offs between the first and the second wave of the study is a limitation for this study and resulted in a small number of observations for the second wave, making it difficult to draw statistical conclusions from the change of stage during and after the car rental event. Therefore, the sample size only allows drawing preliminary results. For future research, collecting information about the time of rental could add insights. Moreover, the effect of the length of car rental can be investigated, as the length of rental has might influence the distance travelled. Having access to more electric vehicles to rent would also enable to improve the data collection, as the analysis can then be carried out for different rental period lengths. Other variables that are worth collecting are whether the rental guests are returning visitors to the island of Gotland, as this might influence their travel patterns. Additional interviews can be used to test the preliminary results from this study. Moreover, heterogeneity of preferences could be investigated using segmentation analyses such as described in Marcuccci et al. (2017) and Valeri et al. (2016). Moreover, it is also worth investigating whether the company of the rental guests has any influence on their travel patterns. Finally, in future work, it should be taken into consideration whether the respondent is planning to purchase a new car within the coming year. Controlling for this variable might increase the correlation between renting an electric vehicle and moving forward in the process of behavioural change.

electric vehicle adoption. EV-renters are more positive about the characteristics of EVs and have more knowledge about EVs, where available charging infrastructure is located and which policy incentives exist. Despite the fact that the number of respondents having taken part in both the first wave and the second wave of the survey is rather small, the fact that there is no clear trend forwards in the process of behavioural change (RQ2) can be caused by the fact that a relatively small group of respondents is planning to purchase a vehicle in the coming six months. Therefore, there is a high threshold between the Contemplation stage and the Preparation stage. In the former, no commitment has to be made yet and the transition is still rather abstract. When comparing the driving patterns of people renting an EV and people renting an ICEV, no big differences have been observed (RQ3). Because of the fact that Gotland is an island with a maximum distance of around 160 km, the available EVs were able to accommodate the majority of the planned trips and the average values were very similar to those of the rented ICEVs. There were even signs of the EVs having been used more intensively. Policy measures tailored towards people in the Pre-contemplation are yet to be designed. EV-rental, even though no large commitment has to be made (after one or a few days, the vehicle is returned), is not likely to be chosen by persons who are in the Pre-contemplation stage. Alternative policy interventions that could be more effective should not require much active involvement of the person. As people are in the Pre-contemplation stage, they are not likely to actively do anything to bring themselves in contact with EVs. For these people, it can be useful to provide information in an easily accessible way. An example could be mass media campaigns where people's driving patterns are shown in order to explain that the EV is not a limiting factor for their daily travel behaviour. Another potential policy measure could be a billboard showing all public charging stations in a certain city and possibly the ones that are planned for the near future. If in the future, the market share of EVs would be much higher, people can also learn from the experiences of the people they know personally, which would imply that the efficiency of these measures would decrease. Electric vehicle rental could still be valuable for people in earlier stages-of-change. Even though people in the Pre-contemplation stage were less likely to rent an EV, four out of six who rented an EV went to more advanced stages of change in the after measurement. For the people in the Contemplation stage, the majority stayed in their stage, while three participants went back and four participants went forwards. In order to get a better sense of the potential of electric vehicle use for daily trips, an alternative strategy could be to stimulate car dealers to lend EVs when someone's car is at the garage, or to stimulate car sharing systems to use electric vehicles. In that way, people can get familiar with the electric vehicle in their own habitat, which might have a more positive effect on people's self-efficacy for using electric vehicles.

7. Conclusions and policy implications Overall, there are indications for EVs to be feasible rental cars in Gotland. The driving experience is considered to be rather positive, and the network of charging facilities facilitates reaching the different parts of the island. The fact that people being on vacation are more eager to try new things also facilitates EV rental to be able to be successful. Going back to the research questions posed, the results of this study showed that EV-renters differ from ICEV-renters in several ways (RQ1). On average, EV-renters are in a more advanced stage-of-change towards

Acknowledgements This project has been financed by the Swedish Energy Agency 81

Research in Transportation Economics 73 (2019) 72–82

J.H.M. Langbroek, et al.

(Elbilslandet Gotland, project number 37838-2).

households: An exploratory analysis. Transport Policy, Special Issue Sustainable transportation and electric vehicles, 1, 244–256. Lane, B., & Potter, S. (2007). The adoption of cleaner vehicles in the UK: Exploring the consumer attitude–action gap. Journal of Cleaner Production, The Automobile Industry & Sustainability, 15, 1085–1092. Langbroek, J. H. M., Franklin, J. P., & Susilo, Y. O. (2016). The effect of policy incentives on electric vehicle adoption. Energy Policy, 94, 94–103. Langbroek, J. H. M., Franklin, J. P., & Susilo, Y. O. (2017). Changing towards electric vehicle use in Greater Stockholm. European Journal of Transport and Infrastructure Research, 17, 306–329. Liao, F., Molin, E., & Van Wee, B. (2017). Consumer preferences for electric vehicles: A literature review. Transport Reviews, 37, 252–275. Lieven, T., Mühlmeier, S., Henkel, S., & Waller, J. F. (2011). Who will buy electric cars? An empirical study in Germany. Transportation Research Part D: Transport and Environment, 16, 236–243. Marcucci, E., Le Pira, M., Gatta, V., Inturri, G., Ignaccolo, M., & Pluchino, A. (2017). Simulating participatory urban freight transport policy-making: Accounting for heterogeneous stakeholders' preferences and interaction effects. Transportation Research Part E: Logistics and Transportation Review, 103, 69–86. Miwa, T., Kiuchi, D., Yamamoto, T., & Morikawa, T. (2012). Development of map matching algorithm for low frequency probe data. Transportation Research Part C: Emerging Technologies, 22, 132–145. Moons, I., & Pelsmacker, P. D. (2012). Emotions as determinants of electric car usage intention. Journal of Marketing Management, 28, 195–237. Noppers, E. H., Keizer, K., Bockarjova, M., & Steg, L. (2015). The adoption of sustainable innovations: The role of instrumental, environmental, and symbolic attributes for earlier and later adopters. Journal of Environmental Psychology, 44, 74–84. Prochaska, J. O. (1991). Assessing how people change. Cancer, 67, 805–807. Prochaska, J. O., & Velicer, W. F. (1997). The transtheoretical model of health behavior change. American Journal of Health Promotion, 12(1), 38–48. Rasouli, S., & Timmermans, H. (2016). Influence of social networks on latent choice of electric cars: A mixed logit specification using experimental design data. Networks and Spatial Economics, 16, 99–130. https://doi.org/10.1007/s11067-013-9194-6. Razeghi, G., Carreras-Sospedra, M., Brown, T., Brouwer, J., Dabdub, D., & Samuelsen, S. (2016). Episodic air quality impacts of plug-in electric vehicles. Atmospheric Environment, 137, 90–100. Rezvani, Z., Jansson, J., & Bodin, J. (2015). Advances in consumer electric vehicle adoption research: A review and research agenda. Transportation Research Part D: Transport and Environment, 34, 122–136. Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change. Journal of Psychology, 91, 93–114. Rogers, E. M. (2003). Diffusion of innovations (5th ed.). New York: Free Press. Schuessler, N., & Axhausen, K. (2009). Processing raw data from global positioning systems without additional information. Transportation Research Record: Journal of the Transportation Research Board, (2105), 28–36. Schuitema, G., Anable, J., Skippon, S., & Kinnear, N. (2013). The role of instrumental, hedonic and symbolic attributes in the intention to adopt electric vehicles. Transportation Research Part A: Policy and Practice. Psychology of Sustainable Travel Behavior, 48, 39–49. Spector, P. (1992). Summated rating scale construction. Newbury Park: SAGE Publications. Steg, L., & Vlek, C. (2009). Encouraging pro-environmental behaviour: An integrative review and research agenda. Journal of Environmental Psychology, Environmental Psychology on the Move, 29, 309–317. Tang, J., Liu, F., Wang, Y., & Wang, H. (2015). Uncovering urban human mobility from large scale taxi GPS data. Physica A: Statistical Mechanics and Its Applications, 438, 140–153. Valeri, E., & Cherchi, E. (2016). Does habitual behavior affect the choice of alternative fuel vehicles? International Journal of Sustainable Transportation, 10, 825–835. Valeri, E., Gatta, V., Teobaldelli, D., Polidori, P., Barratt, B., Fuzzi, S., et al. (2016). Modelling individual preferences for environmental policy drivers: Empirical evidence of Italian lifestyle changes using a latent class approach. Environmental science & policy, multidisciplinary research findings in support to the EU air quality policy: Experiences from the APPRAISAL. SEFIRA and ACCENT-Plus EU FP7 projects, 65, 65–74. Verplanken, B., & Wood, W. (2006). Interventions to break and create consumer habits. Journal of Public Policy and Marketing, 25, 90–103.

Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.retrec.2019.02.002. References Bockarjova, M., & Steg, L. (2014). Can protection motivation theory predict pro-environmental behavior? Explaining the adoption of electric vehicles in The Netherlands. Global Environmental Change, 28, 276–288. van Bree, B., Verbong, G. P. J., & Kramer, G. J. (2010). A multi-level perspective on the introduction of hydrogen and battery-electric vehicles. Technological Forecasting and Social Change, 77, 529–540. Bricka, S. G., Sen, S., Paleti, R., & Bhat, C. R. (2012). An analysis of the factors influencing differences in survey-reported and GPS-recorded trips. Transportation Research Part C: Emerging Technologies, 21(1), 67–88. Burgess, M., King, N., Harris, M., & Lewis, E. (2013). Electric vehicle drivers' reported interactions with the public: Driving stereotype change? Transportation Research Part F: Traffic Psychology and Behaviour, 17, 33–44. Carley, S., Krause, R. M., Lane, B. W., & Graham, J. D. (2013). Intent to purchase a plug-in electric vehicle: A survey of early impressions in large US cites. Transportation Research Part D: Transport and Environment, 18, 39–45. Carlson, L. E., Taenzer, P., Koopmans, J., & Casebeer, A. (2003). Predictive value of aspects of the Transtheoretical Model on smoking cessation in a community-based, large-group cognitive behavioral program. Addictive Behaviors, 28, 725–740. Choma, E. F., & Ugaya, C. M. L. (2017). Environmental impact assessment of increasing electric vehicles in the Brazilian fleet. Journal of Cleaner Production, 152, 497–507. Cocron, P., Buhler, F., Neumann, I., Franke, T., Krems, J. F., Schwalm, M., et al. (2011). Methods of evaluating electric vehicles from a user's perspective - the MINI E field trial in Berlin. IET Intelligent Transport Systems, 5, 127–133. Destination Gotland (2017). Destination Gotland. http://destinationgotland.se/, Accessed date: 9 August 2017. Egbue, O., & Long, S. (2012). Barriers to widespread adoption of electric vehicles: An analysis of consumer attitudes and perceptions. Energy Policy, Special Section: Frontiers of Sustainability, 48, 717–729. Elbilsstatistik (2017). Elbilen i sverige. http://elbilsstatistik.se/, Accessed date: 9 August 2017. Forward, S. E. (2014). Exploring people's willingness to bike using a combination of the theory of planned behavioural and the transtheoretical model. Revue Européenne de Psychologie Appliquée/European Review of Applied Psychology, Transport psychology: Identification of road users’ risks and attitudes and behaviour change, 64, 151–159. Franke, T., & Krems, J. F. (2013). Understanding charging behaviour of electric vehicle users. Transportation Research Part F: Traffic Psychology and Behaviour, 21, 75–89. Friman, M., Huck, J., & Olsson, L. E. (2017). Transtheoretical model of change during travel behavior interventions: An integrative review. International Journal of Environmental Research and Public Health, 14https://doi.org/10.3390/ ijerph14060581. Graham-Rowe, E., Gardner, B., Abraham, C., Skippon, S., Dittmar, H., Hutchins, R., et al. (2012). Mainstream consumers driving plug-in battery-electric and plug-in hybrid electric cars: A qualitative analysis of responses and evaluations. Transportation Research Part A: Policy and Practice, 46, 140–153. Haas, S., & Nigg, C. R. (2009). Construct validation of the stages of change with strenuous, moderate, and mild physical activity and sedentary behaviour among children. Journal of Science and Medicine in Sport, Physical Activity in Young People: Assessment and Methodological Issues, 12, 586–591. Jenelius, E., & Koutsopoulos, H. N. (2013). Travel time estimation for urban road networks using low frequency probe vehicle data. Transportation Research Part B: Methodological, 53, 64–81. Jensen, A. F., Cherchi, E., & Mabit, S. L. (2013). On the stability of preferences and attitudes before and after experiencing an electric vehicle. Transportation Research Part D: Transport and Environment, 25, 24–32. Kurani, K. S., Turrentine, T., & Sperling, D. (1994). Demand for electric vehicles in hybrid