Comparing high-end and low-end early adopters of battery electric vehicles

Battery electric vehicle adoption research has been on going for two decades. The majority of data gathered thus far is taken from studies that sample members of the general population and not actual adopters of the vehicles. This paper presents ﬁndings from a study involving 340 adopters of battery electric vehicles. The data is used to corroborate some existing assumptions made about early adopters. The contribution of this paper, however, is the distinction between two groups of adopters. These are high-end adopters and low-end adopters. It is found that each group has a different socio-economic proﬁle and there are also some psychographic differences. Further they have different opinions of their vehicles with high-end adopters viewing their vehicles more preferentially. The future purchase intentions of each group are explored and it is found that high-end adopters are more likely to continue with ownership of battery electric vehicles in subsequent purchases. Finally reasons for this are explored by comparing each adopter group’s opinions of their vehicles to their future purchase intentions. From this is it suggested that time to refuel and range for low-end battery electric vehicles should be improved in order to increase chances of drivers continuing with BEV ownership. (cid:1) 2016 The Authors. Published by Elsevier Ltd. ThisisanopenaccessarticleundertheCCBY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
The automotive sector is moving towards a transition from primarily petrol and diesel fuelled internal combustion engine vehicles (ICEVs) to more sustainable plug-in hybrid vehicles (PHEVs) and battery electric vehicles (BEVs) (Poullikkas, 2015;Sierzchula et al., 2014).BEVs are considered to be the most beneficial of these due to them having zero emissions, high efficiencies and having the potential to be fuelled entirely off renewable electricity (Helveston et al., 2014;Nordelöf et al., 2014;Offer et al., 2011;Schneidereit et al., 2015;Sierzchula et al., 2014;Thomas, 2009).In order for these vehicles to have the greatest effect on improving urban air quality, reducing carbon emissions and reducing energy use they need to be deployed in larger numbers than they are at present.Therefore a greater understanding of how to increase market penetration needs to be developed.It is possible to achieve this through understanding early adopters of BEVs (Schuitema et al., 2013).This will lead to an understanding of where the market for these vehicles lies and also how to ensure that BEVs appeal to these markets.This will inform policy makers and automotive OEMs on how best to grow the market of BEVs such that the societal benefits can be maximised.At present the market is at a very early stage of development with recent market introductions beginning in 2008-2010.Since then the BEV market has developed and grown both in terms of the numbers of vehicles available and the numbers being adopted by consumers.At the end of 2014 there were 665,000 BEVs deployed globally, with the top three markets for BEVs being the US (39%), Japan (16%) and China (12%).The market shares of BEVs in these nations are still low and of these only in the US did BEVs achieve a 1% share of 2014 vehicle sales.The highest market shares in terms of yearly sales are in Norway (12.5%) and the Netherlands (4%) (IEA, 2015).These numbers are promising for an early market but are still insignificant compared to the entire transportation market (Rezvani et al., 2015), clearly greater effort is needed in order to increase these numbers.
A significant change in the landscape of the BEV market occurred in 2012 with the introduction of the Tesla Model S. Prior to this all BEVs on the market where what are considered here to be low-end electric vehicles (Hardman et al., 2014(Hardman et al., , 2013)).These vehicles all have prices of $30-40,000 and ranges of <100 miles (Nissan, 2014).The Tesla Model S, which is considered here as a high-end BEV costs $70,000-105,000 and has a range of 270 miles (Tesla Motors Inc, 2014).Therefore the introduction of this vehicle resulted in a new market segment being created.So far, within the literature, adopters of BEVs have been considered as one homogenous group, with studies overlooking potential differences between high and low-end adopters.Existing studies have investigated barriers to the adoption of electric vehicles (Browne et al., 2012;Egbue and Long, 2012), how experience of an BEV relates to intention to adopt (Bühler et al., 2014;Franke et al., 2012;Graham-Rowe et al., 2012), purchase intentions and preferences of potential adopters (Chorus et al., 2013;Koetse and Hoen, 2014;Sierzchula et al., 2014) along with studies that identify potential early adopters of BEVs (Campbell et al., 2012;Plötz et al., 2014).Further literature investigates people with first hand experience with a BEV, such as (Lane et al., 2014).An insightful study by Caperello et al. (2014) used workshops involving BEV adopters and ICEV drivers in order to understand how to bridge the gap between early and later adopters.
At the beginning of this study it was hypothesised that the two groups of adopters would be different.This is due to the significant differences in the price and features of the vehicles (Table 1).It was believed that adopters would have different socio-economic and psychographic profiles.It was also believed that they would have differing opinions of their vehicles owing to their different attributes and features, which can be seen in Table 1.Further to this, an understanding of future purchase intentions of actual BEV adopters was needed.This should be in relation to the attributes of each vehicle in order to understand what circumstances lead to a high likelihood of continued adoption.Consumer intent to purchase a BEV has been investigated in some detail within the literature (Bühler et al., 2014;Chorus et al., 2013;Koetse and Hoen, 2014;Sierzchula et al., 2014).These studies investigate the intent of ICEV drivers to adopt a BEV and not BEV driver's future intention to continue with BEV ownership.In order for the market to grow early adopters will be required to remain users of BEVs and not abandon the technology.Repeat purchases are more important than initial purchases in maintaining long term growth of any new product (Crawford and Benedetto, 2011;Rogers, 2003).The overriding aim of this paper is to explore and understand the difference between adopters of low and high-end BEVs.This distinction between two different adopter groups of BEVs is the major contribution of this paper.The hope is that policy makers can use the results of this study to make more informed policy decisions and that OEMs are able to develop cars that are properly positioned for each market, in order to ultimately grow the BEV market.

Literature review
BEV adoption research has been on going since the early 1990s (Golob et al., 1997;Kurani et al., 1994Kurani et al., , 1996)), since then the body of literature has grown considerably with authors in many countries looking towards understanding the complexities of BEV adoption.The vast majority of the literature gathers empirical data from persons who are not adopters of BEVs, often sampling the general public and asking them questions about BEV perception (Egbue and Long, 2012;Krupa et al., 2014;Plötz and Gnann, 2011;Plötz et al., 2014).Only recently has literature begun to report samples of people who have actual experience with BEVs.This data can be more insightful as it is more representative of an actual decision to adopt a BEV, rather than a hypothetical one.Studies that involve actual adopters of BEVs include (Caperello et al., 2014;Lane et al., 2014;Tal and Nicholas, 2013;Tal, 2014;Turrentine et al., 2011).Whilst these studies are becoming more numerous they are still not abundant within the literature, and more studies are needed in order to guide the transition from ICEVs to BEVs.Indeed, a 2015 review by Rezvani et al. (2015) calls for more studies that use data from actual adopters.

Table 1
Comparison of the Nissan Leaf (low-end BEV), of which there were 152 in this study, and the Tesla Model S (high-end BEV), of which there were 153 in this study (Nissan, 2015;Tesla Motors Inc, 2015).A number of early publications explored characteristics of potential early adopters by interviewing multi-car households in the US (Kurani et al., 1994(Kurani et al., , 1996)).The authors predicted that early adopters of BEVs would be households with 2 or more cars and have a garage where at least one car can be parked.A study by Carley et al. (2013) predicted intent to purchase a BEV based on a sample of 2302 members of the general public.It was found that the main advantages of BEVs were high fuel economy, low environmental impacts, positive image and BEVs being viewed as cutting edge technology.From their sample they concluded that early adopters are likely to be highly educated, environmentally sensitive and would already tend to be owners of a hybrid vehicle.Studies by Plötz et al involving 210 people with ''high interests in EVs" (Plötz and Gnann, 2011;Plötz et al., 2014) predicted that early adopters would be middle-aged males, in technical professions, in rural or suburban multi-person households.
A study of US residents over 17 years of age by Hidrue et al. (2011) predicted that early adopters would be young, educated, have green life styles, and fuel cost concerns.Contrary to other studies, these authors found that income and multiple car ownership would not be key characteristics.They also found a high willingness to pay for a BEV with good range, fast charge time, with fuel cost savings, reduced emissions and good performance.Another study in the US sampled 1000 members of the general public in order to understand Plug-in Hybrid Electric Vehicle (PHEV) market entry (Krupa et al., 2014).Egbue and Long (2012) sampled 500 ''Technology Enthusiasts", who owned ICEVs.They found from their sample that sustainability was less important than purchase price and vehicle performance.A UK based study by Campbell et al. (2012) used census data in order to identify locations of early adopters.They identified early adopters as people who were homeowners, commute to work in their own vehicle, own 2 or more cars, have a high socio-economic profile, and are highly educated.Turrentine et al. (2011) investigated members of the Mini E trial in the US.The goal of the study was to understand user responses to BEVs and to identify a route to market for them.Members of the trial were required to lease the vehicle in order to participate.54 Mini E drivers took part in the study; data was collected using driving diaries, online questionnaires, and interviews.This was the largest data set of its kind at that time.From the study it emerged that users of the Mini E value the high performance of the vehicle, the sporty handling and the fact that these driving characteristics were available with low environmental impact.It also emerged that the regenerative braking meant that for much of the time acceleration and deceleration could be controlled using only one pedal, making it more convenient to drive.Members of the study were found to look favourably on BEVs post trial, with 100% agreeing that BEVs were suitable for daily use.The results of the trial were positive with 71% of the sample indicating that they were more willing to adopt a BEV after the trial.In addition to this, 64% of respondents indicated that they planned on purchasing a BEV in the next 5 years.A later study on the Mini E in Germany (Bühler et al., 2014) involved 79 participants in a 6 month trial.Respondents were interviewed before, during and after the study.According to the authors, this was the only study that recorded changes in BEV perception over time.They found that high purchase price and limited range still represent the main barriers to adoption.Another European study, this time in the UK, gathered data from 40 participants in an EV trial.In this study 20 people were given a PHEV and 20 people a BEV for a period of 7 days.All participants in the study were drivers of ICEVs.From qualitative interviews it was found that the respondents believed that purchase prices were too high (Graham-Rowe et al., 2012).
A US study by Lane et al gathered data from actual early adopters of BEV (Lane et al., 2014).In their sample 59 of the 76 respondents were BEV owners, and the remainder were fleet users.The study therefore used data from people who had adopted a BEV.It concentrated on early adopters of the THiNK City.The authors reported that users valued the environmental friendliness of the vehicles, and simply being early adopters of a new technology.The advantages of BEV ownership were found to be saving money, environmental protection, high-tech, low maintenance costs, and fun/enjoyable driving style.Three further studies that use data from actual adopters were identified.The publication by Caperello et al. (2014) investigated how to get later adopters or laggards interested in BEV through workshops that were populated with both BEV drivers and ICEV drivers.Tal and Nicholas (2013) explored who is buying BEVs and if these people are different from ICEV drivers.A second paper by the same authors explored the influence of high-occupancy vehicle lanes access for BEV buyers (Tal, 2014).These papers have the highest number of BEV adopters in their sample of any study the authors of this paper are aware of.
More recently, researchers have began looking into how the market for BEVs can be increased and how to encourage consumer adoption.Dumortier et al. (2015) suggested that high costs and deferred financial savings of BEVs lead to reduced rates of adoption.The authors suggest that providing total cost of ownership data could overcome this barrier.Helveston et al. (2014) investigated the impact financial incentives have on the adoption of BEVs.Gnann et al. (2015) suggest that there may be a significant market of BEVs in the commercial passenger vehicle sector.
An important goal of the literature is the identification of early adopters of BEVs.This information is useful in developing and growing the market.At present the literature suggests that early adopters will have pro-environmental and protechnology attitudes (Wolf and Seebauer, 2014), will be highly educated (Campbell et al., 2012;Carley et al., 2013;Hidrue et al., 2011), have a high economic status (Campbell et al., 2012;Hidrue et al., 2011), have two or more cars (Kurani et al., 1994), be young to middle aged (Hidrue et al., 2011;Plötz et al., 2014), would likely own a hybrid vehicle (Carley et al., 2013), have fuel cost concerns, and be mostly male (Plötz et al., 2014).These generalisations are made from data obtained from potential BEV users but without any empirical evidence from actual early adopters and hence need to be validated using such data.A summary of the assumptions made in the literature can be seen in Table 2.The table shows the authors, sample size, sample population and the expected socio-economic profile of early adopters along with the expected benefits of BEVs.
Currently within the literature there is no data that explores the future purchase intentions of current owners of BEVs.This data is important in order to understand the diffusion of BEVs through the market, and will also reveal if BEVs have enough benefits to convince adopters to continue with ownership.The majority of current BEV owners have purchased their vehicle as an initial purchase, having not owned a BEV previously.Subsequent purchases will be repeat purchases and it is known that the way in which an initial or repeat purchase decision is made is different (Crawford and Benedetto, 2008).One of the most significant limitations of the literature is that early adopters are considered to be one homogeneous group of consumers.Within the literature they are referred to as having shared socio-economic and psychographic characteristics.
No single study makes distinctions between different possible groups of early adopters.This is despite the price of BEVs ranging from $30,000 to $105,000 (Nissan, 2014;Tesla Motors Inc, 2014).It is unlikely that an adopter of a $30,000 vehicle would be similar to the adopter of a $105,000 vehicle.Therefore this paper addresses this major research gap; by identifying differences between high and low-end early adopters.This is achieved by understanding their socio-economic and psychographic characteristics; understanding how they respond to the vehicles that they have adopted and understanding their future BEV purchase intentions.

Questionnaire survey
By the end of 2014 there were 665,000 BEVs worldwide with 39% (275,000) of these being in the United States.For this reason, the questionnaire was targeted towards North American owners of BEVs.Nevertheless, the questionnaire was left open to all BEV owners across the world.Between July and December 2014, 340 fully completed surveys were collected.The method in which owners were recruited to participate in the questionnaire was via online forums.The following forums were identified and used: Telsamotors.com, the official Tesla forum.Reddit.com/r/teslamotors,an online forum with a sub-area for Tesla enthusiasts.Reddit.com/r/electricvehicles, an online forum with a sub-area for electric vehicle enthusiasts.Nissan and Infiniti Car Owners, a forum for Nissan and Infiniti owners, including the Nissan Leaf.Leaf Talk, a Nissan Leaf owner forum.Speak EV, a forum for owners of any electric vehicle.
The study divided the online questionnaire into three sections: the first gathers socio-economic data, the second psychographic information, the final section asks for information on respondents' opinions of their vehicle's attributes, and also asks them about their future BEV purchase intentions.The methods in which these questions were formulated are explained below.

Socio-economic and psychographic data
The socio-economic profile of respondents was measured to understand if there are any statistically significant differences between low-end and high-end adopters.Questions were developed based on the existing literature, outlined in the review, which makes statements about early adopters gender, age, income, level of education and the number of cars in the household; therefore these 5 attributes were used to understand respondent's socio-economic profile.
Rogers' theory (Rogers, 2003) makes some generalisations about early adopters' psychographic profile.These help towards identifying the types of people that may be adopters of new technologies in general.These generalisations along with findings from existing BEV literature were used to develop 20 questions, which measure respondent's psychographic profile.This allows a more quantitative method in deciding if low-end and high-end adopters are significantly different from one another.All 20 questions can be seen in Table 7.

Vehicle attribute opinion & future data
In order to understand what the benefits and shortcomings of BEVs are, a number of pilot interviews with UK BEV owners were undertaken.In total, 5 BEV adopters were interviewed, they were asked why they chose to adopt a BEV and what the benefits of ownership are.The following 10 attributes measured in this survey emerged as the perceived benefits of the vehicles and reasons for adoption of a BEV: 1. Brand 2. Vehicle image/looks 3. Purchase price 4. Vehicle range 5. Time to refuel 6. Vehicle performance 7. Fuel economy 8. Environmental impacts 9. Life style fit 10.Running costs Respondents were asked to compare their vehicle to an ICEV in order to ascertain in what manor the vehicles are worse, similar and superior.After respondents were asked about their vehicle attribute opinions they were asked two questions that measured if they would continue with BEV adoption in the future.The first measured the likelihood of continual ownership of any BEV, the second measured any brand loyalty to the current BEV that they own.This is a measure of consumer perceptions and not the actual performance of the vehicles.As is discussed in (Crawford and Benedetto, 2011), consumer perceptions are more important than actual product performance.Therefore the results reported here may not necessary be representative of the actual attributes of a BEV, but they are representative of how early adopters view these attributes.

Data analysis
In order to analyse the data three statistical techniques are used.In order to compare means between samples that use a Likert scale the T-test is used.In order to compare statistical differences between samples that do not use a Likert scale and use a nominal scale Chi-square is used.Finally linear regression is used to find out whether and which of a number of hypothetical independent variables have a significant impact on the dependent variable.The way in which these techniques are used is summarised in Table 3.
In order to identify the differences between high and low-end adopters the T-test is used.The T-test compares samples in order to understand if there is a statistically significant difference in the means.In this case it is used to compare questions that use a Likert scale.In order to reject the null hypothesis we use the standard 5% confidence level, meaning we require a significance value (p) of 0.05 or below.If a null hypothesis is rejected this means that there is a significant difference between the two sample populations.Independent samples T-test is calculated using the following: q T = obtained T value X 1 and X 2 = means for the two groups s 2 1 and s 2 2 = variances of the two groups n 1 and n 2 = number of respondents in each of the two groups Chi-Square is used to assess differences in the socio-economic data.Chi-square is used here, rather than the T-test, because the data is not measured using ordinal scales meaning the T-test would be inappropriate.The usual 0.05 significance is used to reject the null hypothesis.Chi-square goodness of fit is calculated using the following: In order to understand the data further, multiple regression is used.In this case it is used to understand how early adopter's views of the different attributes of their vehicles, as shown below in Table 8, may relate to their willingness to continue with BEV ownership as shown in Fig. 2 below.The fist stage of analysis is deciding whether the regression model has explanatory power.This is done by testing ANOVA (Analysis of Variance), which is rejected at the usual 0.05 level.In this case the null hypothesis for each variable is rejected at a significance of 0.1.This is less stringent than 0.05.This study focuses on early adopters and investigates the behaviours and opinions, therefore is a study within the field of social science.Within this field, due to the larger number of variables it is permissible to use this less stringent level of significance compared to natural sciences.The variables that are rejected at a significance of 0.1 are then further tested.This time only the values that are significant at a level <0.1 are retained in the regression model.For this final regression the null hypothesis is again set at 0.1.The variables less than 0.1 will be a significant influence on the dependant variable and the model is therefore a good predictor of future vehicle purchase intentions.This method of stepwise regression analysis is known as the backward elimination method.Multiple regression derives from simple Regression, which is calculated using below formula: where Y = the dependant variable X = are the independent variable b = the slope of the regression line a = the intercept Multiple regression, which is used in this study, is a more advanced technique and is used to predict the value of a dependent variable based on more than one independent variable.The equation for multiple regression is shown below.
where Y = the dependant variable X = are the independent variables b = the slope of the line a = the intercept

Results and discussion
The socio-economic profile of the 340 early adopters can be seen in Table 4.The sample is mostly male at 92.6%.Age is spread widely, however most are middle aged with 73.8% of the sample between 35 and 64 years of age.Level of education is high with 16.4% holding a doctorate or equivalent, 28.1% with a master's degree or equivalent and 40.6% with a bachelors or equivalent.This means that 85.1% of the sample has received a University level education.Level of income within the sample is high, with 76.5% earning more than $90,000 per year.The number of vehicles per household in this sample is 2.5, this is higher than the US average of 1.9 (US Department of Transportation, 2009).The sample consists of 359 electric vehicles, a breakdown of vehicles can be seen in Table 5.There are 19 more vehicles than BEV early adopters in this study as some owners have more than one BEV.The most common vehicles are the Tesla Model S (n = 153) and the Nissan Leaf (n = 152).
The remaining vehicles are all fully electric cars with the exceptions of 2 electric motorcycles.Further to this there are 9 vehicles that are range extended electric vehicles, of these there are 2 BMW i3's and 7 GM Volt's/Vauxhall Ampera's.
The questionnaire also asked adopters about their previous vehicles, this reveals an interesting trend.In order to compare prices of previous vehicles with prices of BEVs the data was standardised to 2014 vehicle prices.Therefore if the respondent's previous vehicle was a 1998 VW Golf, the 2014 price of this vehicle was compared to the 2014 price of the BEV that they currently own.It is found that the average purchase price of low-end adopters previous ICEVs would have been a mean of $25,553 and medium of $23,660.Low-end adopters paid a premium of $4195-5350 for their BEV compared to the ICEV they previously owned.This is an amount of money that could reasonably be recovered due to the low running costs of BEVs.High-end adopters previous vehicles have a mean of $45,144 and medium of $40,285.High-end adopters paid a premium of $37,614-41,575 for their BEV compared to the price of their previous vehicle.This is a significant leap, and goes against some suggestions that the cost of BEVs is too high (Graham-Rowe et al., 2012).

Socio-economic differences
Based on the results presented in Table 4 above it does appear that the groups have different socio-economic attributes.However in order to define whether these are statistically significant differences Chi-square was used, the results of this can be seen in Table 6.Firstly gender was compared, with the null hypothesis ''There is no difference in gender between high and low-end adopters".This was rejected at a significance of 0.019.It is found that whilst both groups have a low number of females, high-end adopters are comprised of more females than low-end adopters.11% of high-end adopters in this study were female, compared to only 4.3% of low-end adopters.There are, however, a low number of females in this study, only 25 in total, meaning this finding should be treated with caution until a larger sample of female adopters has been gathered.The age of adopters was compared using the null hypothesis ''There is no difference between high and low-end early adopters age".This was rejected with a significance value of <0.001.It is found that high-end early adopters are of a higher age than low-end early adopters.The level of education was then compared using the null hypothesis ''Level of education does not differ between high and low-end early adopters".This null hypothesis was rejected with a significant value of <0.001.Meaning that there are differences in the level of education between high and low-end adopters.High-end adopters are of higher education than low-end adopters.23.9% of high-end adopters held a doctorate or equivalent compared to 10% for low-end adopters, and 34.8% of high-end adopters held a masters or equivalent compared to 22.2% for low-end adopters.In total 92.3% of high-end adopters have received a University level education, compared to 78.9% of low-end adopters.The income of adopters was compared with the null hypothesis ''There is no difference between high and low-end early adopters level of income".This hypothesis was rejected at a significance of <0.001.It is found that whilst both sets of adopters are high income, the high-end adopters level of income is significantly higher than the low-end adopters.To illustrate this, 12.6% of low-end adopters earn more than $180,000 whilst 62.6% of high-end adopters earn more than this.The number of vehicles in the household does not differ between samples and the null hypothesis could not be rejected.Both household types have a higher car ownership than the US average.In summary there is no difference in car ownership, but clear differences in gender, age, education and income, with on average high-end adopters having higher socioeconomic status than low-end adopters.

Psychographic differences
In order to understand how high-end and low-end early adopters' psychographic profiles differ, a number of questions measuring psychographic differences were investigated using the T-test.From this it appears that in general both groups of adopters are typical early adopters based on the data presented in Table 7, however there are some differences.When comparing each group the null hypothesis was rejected for 2 questions.These were ''The level of empathy does not differ between high-end and low-end adopters" and ''There is no difference in the length of the innovation decision period between high and low-end adopters".It is found that high-end adopters have a significantly higher level of empathy, and that they often take less time before making a decision to invest in a new technology.In Rogers' theory (Rogers, 2003) it is stated that this would typically apply to early adopters.Therefore it appears that within this sample, high-end adopters are more representative of early adopters based on these two results.These two differences suggest that high-end adopters are slightly more aligned with Rogers theory.In addition to these two differences there is one variable with a significance of <0.1 and a further two that are close to 0.1.These numbers are not low enough for the null hypotheses to be rejected but they do indicate a number of additional subtle differences.The axis starts at ''Far Worse" in the centre, then through ''Slightly Worse" to ''Similar", then ''Slightly Superior" and finally ''Far Superior" on the outer most axis line.This means the closer the line to the outer edge the more superior adopters perceive this attribute.

Battery electric vehicle attributes
In order to understand how early adopters view their vehicles respondents were asked, ''Considering each of the following attributes how do you think your electric vehicle compares to an internal combustion engine vehicle?"The list of attributes given can be seen in Fig. 1 and Table 8.Respondents were given 5 answer options on a Likert scale.These were Far Superior, Slightly Superior, Similar, Slightly Worse and Far Worse.Fig. 1 shows a summary of how adopters view their vehicles in comparison with ICEVs.The scale of the spider diagram starts at ''Far Worse" in the centre goes through ''Similar" and to ''Far Superior" on the outer most axis.Therefore the closer each data point is located to the outer edge of the decagon the more superior the attribute is viewed.The graph shows that high-end adopters view their vehicles as being superior in the 7 following areas; brand, vehicle image/looks, vehicle performance, fuel economy, environmental impacts, lifestyle fit and running costs.This adopter group viewed purchase price, time to refuel and vehicle range as similar to ICEVs.Low-end adopters viewed their vehicles as being superior in the 5 following areas; performance, environmental impacts, fuel economy, lifestyle fit, and running costs.They viewed the vehicles as having similar vehicle image/looks and brand and as having worse purchase price, vehicle range and time to refuel.T-test results of the comparison between psychographic variables between high-end and low-end early adopters (Liker scale for questions is 1 = Agree Strongly, 2 = Agree Slightly, 3 = Neither Agree or Disagree, 4 = Slightly Disagree, 5 = Strongly Disagree).

Question
Adopter In order to understand how each adopter group's opinions compare, the T-test was used.The results of the T-tests can be seen in Table 8.The null hypothesis for brand, ''there is no difference between perceptions of brand between high-end and low-end early adopters" was rejected at a significance of <0.001.High-end adopters find their vehicles to have a superior brand compared to ICEVs whilst low-end adopters believed their vehicles to have a similar brand.The null hypothesis for vehicle image ''there is no difference between perceptions of vehicle image between high and low-end early adopters" was rejected at a significance of <0.001.High-end adopters view their vehicles as having a superior image and low-end adopters find their vehicles to be similar to ICEVs.The null hypothesis for purchase price was also rejected, this time at a significance of 0.028.It is found that high-end adopters view the purchase price of their vehicles as similar to ICEVs, whilst owners of low-end BEVs view their vehicles as slightly worse in terms of purchase price.This result is particularly intriguing as previously mentioned low-end adopters paid a far smaller premium for a BEV compared to high-end adopters.It is possible that each adopter group is not comparing their BEV to the ICEV which they previously owned rather they are making comparisons between their BEV and a vehicle which they perceive as being in a similar vehicle class.The null hypothesis comparing the means for vehicle range was rejected at a significance of <0.001.High-end early adopters believed that their vehicles have a similar range as compared to ICEVs.Low-end adopters believed their vehicle's range was worse than that of a comparable ICEV.This result is unsurprising as the EPA estimated range of a Tesla BEV is 270 miles, whilst the range of a low-end BEV can be less than 100 miles.
The null hypothesis ''there is no difference in how high and low-end adopters perceive their vehicles' time to refuel compared to that of an ICEV" was rejected.This was rejected at a high level of significance of <0.001.It is found that high-end adopters believe their vehicles have similar time to refuel as ICEVs, low-end adopters believe their vehicles to be worse time to refuel than ICEVs.It is surprising that high-end adopters perceive their vehicles to have similar time to refuel compared to ICEVs.The time to fully recharge a Tesla BEV is far longer than it takes to fill an ICEV with petrol or diesel.However the amount of time required for human interaction with the vehicle is similar, meaning that plugging in a BEV to a socket takes no longer than inserting a petrol or diesel pump into an ICEV.Additionally as will be shown below, some adopters viewed BEVs as being more convenient to refuel/recharge.
The null hypothesis comparing the means for performance was rejected at a significance of <0.001.Both groups of adopters believed their BEVs were superior to ICEVs in this respect, however high-end adopters viewed their vehicles as far superior, with low-end adopters viewing their vehicles as only slightly superior.The null hypothesis ''there is no difference between high and low-end adopters' perception of their vehicles' life style fit compared to ICEVs" was rejected at a significance of <0.001.It was found that both groups of adopters do believe their vehicles to be a better life style fit compared to ICEVs, however, high-end adopters found their vehicles to be a better fit than did the low-end adopters.The null hypotheses comparing fuel economy, running costs and environmental impact could not be rejected.There is no difference in how high and low-end adopters view these attributes.Both groups believe that their vehicles are superior to ICEVs in these three areas.Indeed BEVs do have superior fuel economy, running costs and environmental impact compared to ICEVs.Out of 10 vehicle Table 8 Table showing differences in answers to the question ''Considering each of the following vehicle attributes how do you believe your electric vehicle compares to an internal combustion engine vehicle?"Answers were measured using a Likert scale (1 = Far Superior, 2 = Slightly Superior, 3 = Similar, 4 = Slightly Worse, 5 = Far Worse).

Attribute
Adopter attributes tested, 7 null hypotheses can be rejected.This along with Fig. 1 clearly show that each adopter group responds to their vehicle is significantly different.High-end adopters are found to have more positive opinions of their vehicles compared to low-end adopters, they believe their vehicles are superior compared ICEVs.Further to the attributes that were tested using Likert scale questions respondents were also able to provide qualitative feedback with the question ''Please use the space below to list any advantages you think you (or your household) has experienced by using an electric vehicle?"This question revealed an additional benefit of BEV ownership.Without any prior cues 88 respondents (25.8%) said that BEV ownership had added convenience over ICEVs.Respondents reported that this saved them time and was more convenient for them.Respondent No. 8 answered, ''Not having to waste time to go to gas stations", and No. 44 responded ''Save Time by never going to the gas station".Both high and low-end adopters mentioned this as a benefit to BEV ownership.However a larger proportion of high-end adopters mentioned this as a benefit, with 37.9% of them mentioning this, compared to only 18.8% of low-end adopters.Fig. 1 and Table 8 show that high-end adopters view time to refuel more preferentially than low-end adopters.This difference may be due to high-end BEVs having a longer range and a shorter recharge time, which means that when charging more range is added in less time compared to the low-end vehicles.Additionally due to the longer range owners of high-end vehicles will be less likely to charge away from home and they will have to charge less often.

Differences in future purchase intentions
The way in which early adopters perceive the attributes of their vehicles will have an implication on the likelihood of repeat purchases.Two questions measured respondent's future vehicle choices.The first asked ''Will the next car be another battery electric vehicle?"With the following Likert scale for answers ''Definitely Not", ''Probably Not", ''Unsure", ''Probably Yes" and ''Definitely Yes".The second asked ''Will your next car be another Tesla?" or ''Will your next car be another vehicle of the same manufacturer as your current vehicle?"The possible answers for this were ''Yes", ''No" or ''Don't Know".Results from the first question (which used the Likert scale) were compared using the T-test to compare the means; the null hypothesis of there being no difference between the two groups was rejected at a significance of <0.001 for both questions.It was found that high-end adopters have a higher intent to continue with BEV ownership, and it is likely that they will continue to own a Tesla with their next BEV.67% of low-end adopters would probably or definitely continue with BEV ownership, compared to 81% of high-end adopters.Of the high-end adopters 59% said they would continue with Tesla ownership.Of the low-end adopters only 23% said they would continue with the same make of BEV.These findings are shown in Figs. 2  and 3.This demonstrates that low-end adopters are less likely to continue with BEV ownership than high-end adopters.This is concerning for the diffusion of BEVs through the market.This lack of willingness to continue with BEV ownership is not ''technological rejection" as Rogers defines this as ''The decision to not adopt an innovation".Within diffusion literature the decision to not continue with the adoption of an innovation is known as discontinuance.Discontinuance has previously been explored in the field of assistive technology for disabled persons (Philips and Zhao, 1993;Scherer, 1996).It has not received attention within automotive literature, and needs to be further understood.

Understanding differences in future purchase intentions
In order to build a greater understanding of why some adopters may not continue with BEV ownership multiple regression was used to understand how the results from the question ''Will the next car be another battery electric vehicle?" compare with how owners perceive the attributes of their vehicles (Table 8).This enables identification of which attributes of BEVs are a good indicator of likelihood of continuing with BEV ownership.The methodology for this is explained in Section 2.2.
The first multiple regression model for low-end adopters had an ANOVA significance value of <0.001 suggesting that the independent variables were a good predictor of the dependant variable.When comparing the significance values for each independent variable the null hypothesis was rejected for 6 attributes.These were vehicle image/looks, purchase price, time to refuel, environmental impacts, life style fit and running costs.Therefore these 6 attributes were tested again.The attributes that had a significance value of more than 0.1 were omitted from this regression analysis.The results of this can be seen in Table 9.The ANOVA significance value for this was <0.001 suggesting a high level of significance for the model.This shows that time to refuel, environmental impacts and running costs are the best predictors of future intent to own a BEV for low-end adopters.In order to confirm that these 3 attributes were the most significant contributors to willingness to continue with BEV ownership one final regression was done.This time only time to refuel, environmental impacts and running costs were included.The ANOVA value was <0.001 and the values for each of the three attributes were all less than 0.05 suggesting that they are indeed excellent predictors of willingness to continue with BEV ownership.The beta value for time to refuel is 0.198, environmental impacts is 0.201 and running costs is 0.173.This means that for every 1 unit increase on the Likert scale measuring opinions of environmental impacts there will be a 0.201 unit increase in willingness to continue with BEV ownership.This rate of increase is slightly higher than the other two significant variables, suggesting that environmental impacts are the most significant contributor to likelihood to continue with BEV ownership into the future for low-end adopters.The results from this can be seen in Table 10.
The same procedure was carried out for high-end adopters.Linear regression was applied to all 10 attributes as the independent variables against the dependant variable ''Will the next car be another battery electric vehicle?"The ANOVA significance value for this was 0.025 suggesting the model does have explanatory power.The first regression analysis allowed the null hypothesis for time to refuel, fuel economy, environmental impacts and running costs to be rejected.These 4 attributes were further tested in the absence of the 6 attributes whose significance value was more than 0.1.The ANOVA value for this linear regression was <0.001 suggesting that the model is significant.For these results the null hypothesis was rejected for time to refuel, fuel economy and running costs (Table 11).A second multiple regression was run with only these three attributes.This can be seen in Table 12, this model again had an ANOVA value of <0.001 suggesting the model is significant.In this multiple regression analysis time to refuel and running costs are the most significant contributors to likelihood to continue with BEV ownership into the future.In order to conclusively state that these two attributes are statistically significant a final regression model was run (Table 13), this again had an ANOVA of <0.001 and both independent variables were <0.1.Of these two attributes time to refuel had a Beta value of 0.153 and running costs 0.271.This suggests that for high-end adopters running costs are the most significant contributor to likelihood to continue with BEV ownership in future vehicle choices.Therefore due to the low running costs of a Tesla BEVs compared to the running costs of an ICEV in the same vehicle class, willingness to continue with ownership is high.
Multiple regression analysis suggests that for low-end adopters time to refuel, environmental impacts, and running costs are the best predictors of future intention to adopt.This suggests that these consumers are both motivated by functional considerations but also a social or emotional desire due to their environmental concern.Of these three attributes the beta value is highest for environmental impacts, suggesting that this is the best predictor of future likelihood to continue with BEV adoption.For high-end adopters time to refuel and running costs are the two best predictors of future intention to adopt,

High-end adopters
Will your next vehicle be another battery electric vehicle?suggesting that high-end adopters are more motivated by functional considerations than a social or emotional desire to adopt.Further to this for high-end adopters running costs has the strongest correlation to likelihood to continue with BEV ownership with a beta value of À0.238.Low-end adopters view running costs and environmental impacts as superior and these attributes do contribute to increased propensity to adopt.The third attribute to be found as a good predictor of willingness to continue with BEV ownership, for low-end adopters, was time to refuel.However long recharge times mean that time to refuel is viewed as inferior compared to ICEVs and therefore this is a potential barrier to low-end adopters willingness to continue with BEV ownership in the future.Therefore long recharge times are the most significant contributor to technological abandonment by low-end adopters.The model suggests if this attribute can be improved, or perceptions of this can be improved, willingness to continue with BEV ownership will increase.Therefore in future generations of low-end BEVs, recharge times should be significantly improved over current generations of the vehicles.

Conclusion
Based on data from 340 early adopters of BEVs it has been possible to corroborate a number of assumptions previously made within the literature.In this sample it was found that early adopters have a high-income, with 76.5% earning more than $90,000 per year, this is in agreement with (Hidrue et al., 2011).Early adopters are also highly educated with 85.1% having achieved a university level qualification, agreeing with (Campbell, 2014b;Campbell et al., 2012).They are also mostly male (92.6%) something which was suggested by Plötz et al. (2014).In this sample, of mainly US citizens, car ownership was higher than the US national average of 1.9 per household, with each household having 2.5 cars on average, this is in agreement with (Kurani et al., 1996;Plötz et al., 2014).Finally it was found that 25.3% had owned a hybrid vehicle prior to BEV ownership, whilst this is higher than average it suggests that hybrid ownership is not a prerequisite for BEV ownership, therefore the assumption by Carley et al. (2013) only partially holds true.In this sample there was no clear trend in terms of the age of respondents, however they are mostly between 35 and 65 (76.5%) years old suggesting that BEVs may be most popular with people who are around middle aged.
Previous literature overlooked the possibility of there being different groups of adopters.However, results from this investigation reveal two distinct groups, which are referred to here as low-end adopters and high-end adopters.The groups have significantly different socio-economic profiles, with high-end adopters being of higher income, higher education and of higher age.Both groups still align with the assumptions made in the literature (Campbell, 2014b;Hidrue et al., 2011;Kurani et al., 1996;Plötz et al., 2014), however high-end adopters have a far higher socio-economic status compared to low-end adopters.Two statistically significant psychographic differences were identified, with high-end adopters having greater empathy and taking less time to adopt a new technology.These two differences add to the evidence suggesting that both groups of adopters are not homogenous.
It was found that compared to ICEVs, BEVs have beneficial performance, running costs, life style fit, environmental impacts and fuel economy.This is in agreement with (Lane et al., 2014;Turrentine et al., 2011) who found performance to be a benefit, but goes against some suggestions that the performance of a BEV is viewed negatively compared to ICEVs (Schuitema et al., 2013).It agrees with (Lane et al., 2014), who found running costs to be a benefit and (Carley et al., 2013;Lane et al., 2014) who suggest environmental impacts and fuel economy would be benefits of BEV ownership.High-end adopters also found image and brand to be a benefit.Image has previously been suggested as a benefit by Carley et al. (2013).Despite both adopter groups agreeing that running costs, lifestyle fit, environmental impacts, fuel economy and performance are superior there are still statistically significant differences in the way in each group view these attributes.It was found that high-end adopters view their vehicles more preferentially than low-end adopters in these areas.High-end adopters did not believe their vehicles were worse than ICEVs in any area measured, but low-end adopters believed their vehicles had worse range, time to refuel and purchase price compared to an ICEV.
This paper adds further to the literature by measuring future purchase intentions of BEV owners.It was found that each adopter group has different future purchase intentions.High-end adopters appear likely to continue with BEV ownership with 81% continuing with BEV ownership in future purchases.Brand loyalty was also high with 64% stating their next vehicle will be the same make as their current vehicle.Low-end adopters are less likely to continue with BEV ownership with 67% likely to continue with owning a BEV, furthermore only 23% will continue owning a BEV of the same make as their current model.Therefore 33% of low-end adopters may abandon the technology with their next vehicle purchase, and 77% will choose a vehicle of a different brand, this could be harmful for the diffusion of BEVs and the creation of a more electrified transportation system.
It was found that low-end adopters future purchase intentions are significantly correlated to opinions of their vehicles and that time to refuel, environmental impacts and running costs are the most significant influences.Low-end adopters' opinions of environmental impacts and running costs were positive, as were their opinions of running costs.Their opinions of time to refuel however were negative, as they believed that this attribute was slightly or far worse than an ICEV.High-end adopters' future purchase intentions were related to running costs and time to refuel.Running costs were viewed as superior, time to refuel was viewed as similar, meaning it does not contribute to discontinuance.

Policy and managerial implications
The results from this paper can be used to make a number of policy and managerial implications.These are based on the results that suggest that the groups low and high-end adopters are not homogenous.Low-end adopters may abandon BEVs with their next vehicle purchase and the differing perceptions of the vehicles explain reasons for this abandonment.Even though the results presented here are representative of early adopters in the United States the data may be applicable to other markets globally.It has been previously suggested that early adopters of BEVs will be similar regardless of geographic location due to them having similar socio-economic characteristics, and also because diffusion processes are the same in different markets (Schneidereit et al., 2015).Therefore in markets where BEVs have little market share policy makers and OEMs can seek to target persons with similar socio-economic characteristics as the early adopters of this paper.
This paper has shown that there are two distinct BEV adopter groups.Therefore when introducing and promoting BEVs to markets, policy makers and OEMs should not view early adopters as one homogenous group.The results show that each group has a different socio-economic and psychographic profile, they respond to their vehicles differently and they have

Fig. 1 .
Fig.1.Figure showing differences in answers to the question ''Considering each of the following vehicle attributes how do you believe your electric vehicle compares to a internal combustion engine vehicle?"The axis starts at ''Far Worse" in the centre, then through ''Slightly Worse" to ''Similar", then ''Slightly Superior" and finally ''Far Superior" on the outer most axis line.This means the closer the line to the outer edge the more superior adopters perceive this attribute.

Fig. 2 .
Fig. 2. Comparison of future purchase intentions of BEVs between low-end and high-end adopters.

Table 2
Summary of the main literature that explores BEV adoption, by author, sample size and sample population and the main conclusions of these studies that explore the expected socio-economic profile of BEV adopters and the expected benefits of BEVs.

Table 3
Summary of the statistical techniques used in this paper, what they are used for and the reason they were selected.

Table 4
Table showing the socio-economic profile for each group of adopters and a summary of all of the data within the sample.Income is shown in US income brackets and in US$.

Table 5
Breakdown of the BEVs in this study by make and model.

Table 7
Will your next vehicle be another battery electric vehicle?