Electric Scooter - An Alternative Mode of Transportation for Malaysian Youth

Background: This study examined the effect of environmental concern, knowledge, attitude, subjective norms, perceived behavioural control, attributes of interest, and monetary benet on the willingness to purchase and actual purchase of electric scooter (ES) among Malaysian youth. This study adopted a cross-sectional design and collected quantitative data from 1,306 youth by sharing a Google form link through social media. Results: The ndings reveal that environmental concern and environmental knowledge have a signicant positive effect on attitude towards ES. Findings also show that attitude towards ES, subjective norms, perceived behavioural control, attributes of interest, and monetary benet have a signicant positive effect on the willingness to purchase and actual purchase of ES. Findings reveal the positive and signicant mediating effect of attitude towards ES and the willingness to purchase ES. The multi-group analysis revealed that the effect of attitude towards willingness was signicantly higher among the urban respondents. Moreover, the effect of environmental concern on attitude towards ES among rural respondents was signicantly higher than urban respondents. In contrast, the effect of environmental knowledge on attitude towards ES among urban respondents was signicantly higher than rural respondents. Conclusions: The environmental and transportation policymakers should focus on highlighting the attributes of ESs and their environmental and monetary benets to promote the mass adoption of ESs among Malaysian youth. Offering subsidies like no road tax and zero import duty and reducing the production of combustion-based two-wheelers are among the options to facilitate the adoption of ES among Malaysian youth. The adoption of ES is socially promoted to reduce the use of conventional two-wheelers. Another relevant area that needs attention is the charging outlet or fuel stations for ES. The adoption of ES remains low if there is a lack of charging or fuel stations.


Introduction
Global individual mobility is increasing at the turn of the century, and fossil fuels remain the core energy instrument that drives the world (Eccarius & Lu, 2020). Global transport signi cantly contributes to anthropogenic climate change. The contribution of the transport sector is about 23% of the global greenhouse gases (GHGs) emissions (Tu & Yang, 2019). Road transport accounts for more than 73% of the total of GHGs emissions from the transport sector (DellaValle & Zubaryeva, 2019). Urban areas suffer from transport-related air pollution and low emitting vehicles from early 2000. Cities like Jakarta, Bangkok, and Kuala Lumpur are known as hubs of motorbike users (Eccarius & Lu, 2020). Motorbikes offer exibility and economy in highly congested cities with ease of operation and maintenance. However, the GHGs emissions are high in the conventional combustion-based motorcycles engines that are currently used across the globe (Tu & Yang, 2019).
Moving towards green vehicles can address the situation and signi cantly reduce the GHGs from the transport sector. The global effort in reducing the GHGs emission from the transport sector can foster the efforts to offer electric vehicles (EV) (Lai, Liu, Sun, Zhang & Xu, 2015). Electromobility is the solution to address GHGs and reduce noise and local air pollution due to the use of conventional internal combustion-based engines (Peters & Dutschke, 2014). The increasing prices of fuels motivate the reduction of using conventional internal combustion-based engines. EV use is promoted by the government across the globe. The reduced registration or no road tax for EV encourages consumers to buy EVs and promote the idea of reducing GHGs (Carley, Krause, Lane & Graham, 2013).

Using Electric Scooter
Achieving sustainable society starts with the adoption of a sustainable lifestyle at an individual level.
Individual mobility is a critical topic in developed and developing societies. An individual becomes mobile, and the environment pays the price of enhanced individual mobility (Eccarius & Lu, 2020).
However, individual mobility is the key to the economy, leisure or holiday pursuits. About 25% of the GHGs emissions are caused by transportation (Peters & Dutschke, 2014). EV intensively reduces individual mobility impact on the environment. ES is the critical individual vehicle that is currently utilised in developed countries (Carley et al., 2013). Like other EVs, ESs can reduce GHGs emissions with low noise besides having elegant design and shape (Eccarius & Lu, 2020). However, the ES is in its early market stage, and general consumers' acceptance of ES is low. About 1.7 billion of the scooters (two-wheeler) are owned by individuals around the globe (Chen et al., 2018). However, there are fewer shares of ES. ES runs on electric charger or batteries (Eccarius & Lu, 2020). The rise in the purchase of ESs is observed in recent times, but it is not enough to signi cantly reduce GHGs from individual mobility (Higueras-Castillo, Molinilo, Cosa-Stefaniak & Liebana-Cabanillias, 2019). A total of 80% of the globally registered twowheelers are owned by Asians, and South East Asia has the highest number of two-wheeler owners in Indonesia, Taiwan, and Malaysia (Eccarius & Lu, 2020).
Having an individual mobility vehicle is on the rise in recent times, and it caused an increase in GHGs emissions (Leavy, Drossinos & Thiel, 2017). Southeast Asians are known for having a higher number of two-wheelers in the world due to the general tendency of the population to own two-wheelers (Jone, Cherry, Vu & Nguyen, 2013). In Malaysia, the adoption rate is still at its infancy stage among Malaysians.
Consumer awareness and lack of government support reduce the penetration of ES in Malaysia.
Government support may help to have clean air and less population from transport besides achieving the sustainable development goals (SDGs) 7, 11, and 15. This study aims to explore the willingness to purchase and actual purchase of ESs among Malaysian youth using the theory of planned behaviours (TPB) and extending the TPB by perceived monetary bene ts and perceived attributes of interest for ES.
Malaysian youth can play an essential role in reducing the effect of mobility by adopting ES.
The following section of the paper discusses the relevant literature and development of the hypotheses. Then, it is followed by summaries of the method, analysis, and results of the study. The following section offers the discussion and conclusion of the study. Page

Theory of Planned Behaviour
TBP is well-known, and it is one of the apparent theories for social science that deals with the behavioural intention and adoption of environment-friendly practices and products (Lai et al., 2015). Behavioural intention is a signi cant predictor of the behaviour or adoption behaviour for environmental practices or products (Maichum, Parichatnon & Peng, 2016). Moreover, behavioural intention is the outcome for the social-cognitive factors of attitude, subjective norm and perceived behavioural control ( TPB is used in different studies to explore the adoption of green products by the customers' intention to buy green products like hydrogen-electric motorcycles food (Chen, Tsai & Hsieh, 2018), green product purchase intention (Mostafa, 2006), and payment behaviour for environment-friendly products among low-income groups (Yadav & Pathak, 2016;. This study extended the TPB in the ecological business adoption like beekeeping and extended the application and predictive power of the TPB framework for the adoption of green business.

Environmental Concern (ENC)
The concern for anything can trigger some appropriate actions to solve the issues at hand. ENC is the individual's perception that his or her actions have consequences on the environment (Yadav & Pathak, 2017). ENC is a kind of evaluative acceptance that one's and other's behaviours can affect the environment (Maichum et al., 2016). An individual with a higher level of concern for the environment is more likely to embrace or participate in environmentally friendly practices or adopt green products. Yadav and Pathak (2016) stated that ENC signi cantly predicts the environmental attitude of purchasing ecofriendly products. Moreover, Maichum et al. (2016) claim that consumers' ENC (β = 0.342, p = 0.050) stimulates green consumption by impacting the environmental attitude among Thai consumers. Therefore, this study postulated the following hypothesis: Hypothesis (H 1 ): Environmental concern has a signi cant positive effect on the attitude towards ES among Malaysian youth.

Environmental Knowledge (ENK)
ENK provides the base to develop an attitude towards green behaviours (Mostafa, 2006). ENK is the general awareness of the facts, concepts, and relationships concerning the natural environment with the ecosystem (Mostafa, 2016). ENK develops a personal understanding of the environment, and it is about the vital association between environmental aspects and impact (Yadav & Pathak, 2016). ENK can lead to the appreciation of the collective responsibility to achieve a sustainable lifestyle (Lai et al., 2015).
Individual environmental action depends on the concrete understating of environmental issues that can lead to environmental actions. The lack of ENK signi cantly reduces environmental actions. Mostafa (2006) reported that the individual ENK signi cantly leads to (β = 0.563, p = 0.000) green product purchase intention among the Egyptian sample. Therefore, this study hypothesised the following: Attitude reveals the psychological assessment of the product or service (Yadav & Pathak, 2016). TPB particularly deals with the attitude towards the behaviour. Attitude exposes the subjective evaluation of the given object based on the summary assessment (Maichum et al., 2016). For green products, the attitude towards green products or services are greatly in uenced by the subjective internal assessment of the environmental products and initiated the internal stimuli to own the product (Lai et al., 2015). Mostafa (2006) stated the positive in uence of the attitude of green products on the purchase intention among Egyptian consumers. However, Yadav and Pathak (2017) reported that ATT signi cantly (β = 0.350, p = 0.001) in uences the intention to purchase green products among the poor. Hence, this study hypothesised the following: Hypothesis (H 3 ): Attitude towards ESs has a signi cant positive effect on the willingness to purchase ES among Malaysian youth.

Subjective Norms (SBN)
Social pressures play a signi cant role in the development of the intention to behave in a particular manner (Lai et al., 2015). Subjective norm is the perceived social pressure stimulation to perform or not to perform a speci c behaviour. Individual decision making is in uenced by the perceived social norm of

Perceived Monetary Bene t (PMB)
Adoption as a personal decision is in uenced by the perception of monetary bene ts from the adoption of the product or practice (Peters & Dutschke, 2014). A consumer may consider converting to a new green product based on the reduced running cost of the product or the economic bene ts of using the new innovative product (Tu & Yang, 2019). Lai et al. (2015) reported that the PMB from the use of electric vehicles (β = 0.435, p = 0.000) in uences the intention to purchase electric vehicles among Chinese respondents. Reduced fuel consumption and incentives associated with the purchase of electric vehicles promote the intention to purchase electric vehicles (Razak et al., 2014). This study proposed the following hypothesis: Hypothesis (H 7 ): Perceived monetary bene t has a signi cant positive effect on the willingness to purchase ES among Malaysian youth.

Willingness to Purchase Electric Scooter (WTP)
Positive willingness to purchase or the intention to behave in a particular manner can lead to the performance of the behaviour. The critical evidence was available on the willingness as the critical precursor of the purchase behaviour. Yadav and Pathak (2017) postulated that the intention to purchase green products (β = 0.563, p = 0.001) signi cantly in uences the purchase of green products among Indian respondents. Therefore, this study proposed the following hypothesis:

Mediating Effect of Attitude towards Electric Scooters
TPB directs that the attitude towards a speci c behaviour signi cantly in uences the intention towards a speci c behaviour (Yadav & Pathak, 2017). Moreover, the attitude is impacted by individual environmental concerns and environmental knowledge. Attitude towards green products signi cantly mediates the individual environmental concern and willingness to purchase eco-friendly products (Maichum et al., 2016). The relationship between environmentally friendly products and the willingness to purchase is mediated by the attitude towards environmentally friendly products (Yadav & Pathak, 2017).
Thus, this study hypothesised the following: Hypothesis (H 1M ): Attitude towards ESs mediates the effect of environmental concern on the willingness to purchase ESs among Malaysian youth.
Hypothesis (H 2M ): Attitude towards ESs mediates the effect of environmental knowledge on the willingness to purchase ESs among Malaysian youth.

Mediating Effect of the Willingness to Purchase Electronic Sooter
TPB postulates that the willingness mediates the relationship between the predictors, i.e., attitude towards behaviour, subjective norm, and perceived behavioural control for the particular behaviour (Yadav & Pathak, 2017). However, Lai et al. (2015) reported that behavioural intention is mediated by the factors of TPB and the actual environment-friendly product buying behaviour. Moreover, this study expanded the TPB by factors for the attributes of interest and monetary bene ts for the purchase of ES. Therefore, this study hypothesised the following:

Research Methodology
This study adopted the cross-sectional design and collected quantitative data from Malaysian youth through an online survey to examine the effect of environmental concern, knowledge, attitude, subjective norms, perceived behavioural control, attributes of interest, and monetary bene t on the willingness to purchase and actual purchase of ES among Malaysian youth. Figure 1 shows all hypothesised and tested associations.

Population and Sample
The target population of this study were Malaysian youth (aged between 15-40). In order to collect the data about the willingness and purchase of ES, this study adopted the cross-sectional design and collected data through an online survey during the rst two weeks of April 2020. This study designed a Google form, highlighted the purpose and reporting procedure of the study, and collected informed consent from all respondents before they participated in the survey. The questionnaire was distributed by sharing the link of the questionnaire form using social media. A total of 1,306 valid responses were yielded for the study.

Measurement and Scales
Explicit and straightforward statements were designed to gauge responses for the constructs. This approach could obtain an appropriate and accessible understanding of the respondents. Five questions measured the environmental concern were adopted from several studies (

Assessment of Common Method Variance (CMV)
Social science research methods are connected with common method biases that are initiated by the use of single source and single point of time data collection technique (Podsakoff, Mackenzie Lee & Podsakoff, 2003). Harman's (1976) one-factor test proposed a strategy to access the impact of CMV on study constructs (Podsakoff et al., 2003). The use of one-factor Harman's test con rmed that CMV was not a critical issue for the study as the highest factor accounted for 36.71% variance, which was less than the suggested limit of 50% (Podsakoff et al., 2003).

Multivariate Normality
SEM-PLS is not associated with multivariate normality in the data as it is a non-parametric analysis instrument (Hair, Risher, Sarstedt & Ringle, 2019). However, multivariate data normality was suggested by Peng and Lai (2012) on the use of web power (https://webpower.psychstat.org/wiki/tools/index) to con rm data normality. The test results con rm that the data set is not as normal as Mardia's multivariate coe cient p-values that is less than 0.05 (Cain, Zhang & Yuan, 2017).

Data Analysis Method
Partial least squares structural equation modelling (PLS-SEM) with the Smart-PLS software 3.1 was exploited to examine the study data. PLS-SEM is a multivariate analysis instrument to assess path models that have latent constructs . PLS-SEM allows scholars to work with non-normal and small data sets. Furthermore, the casual-predictive nature of the PLS-SEM can work with complex models that have composites and without the assumption of goodness-of-t estimation than the covariance-based SEM (Chin, 2010). Two-step techniques were proposed to analyse data with PLS-SEM, and the rst measurement was performed on the model to test the reliability and validity of study constructs (Hair et  The importance-performance map analysis (IPMA) de nes the study constructs into relatively high to low by importance and performance for the endogenous construct (Chin, 2010). IPMA detects the possible area of enhancements that necessitate contemplation from policymakers and scholars. IPMA analysis is built on the total effect of rescaled variables scores in the unstandardised procedure (Ringle & Sarstedt, 2016). Rescaling can settle for each latent variable score that is between 0 and 100. The mean value of the latent variable score signi es the performance of latent variable, where 0 represents the least and 100 represents the most signi cant in the performance of the endogenous construct . Furthermore, multi-group analysis (MGA) in PLS-SEM allows scholars to differentiate the variances in prede ned groups (Henseler, Ringle & Sinkovics, 2009). MGA is an agile method to evaluate the variances between the groups inside the data set (Hair et al., 2014). MGA supports the investigators to assess the variations among the structural paths of various groups in the data . The rst step is to generate groups based on the categorical variables of interest like age, gender, or income. Then, the path coe cients of the groups are analysed, and the two groups are signi cantly diverse from each other based on the guidelines by Henseler et al. (2009). The differences that exist within the data set are based on the characteristics that may not be evident in aggregated data, and the path coe cients of the group data can demonstrate the statistical variance using MGA to establish the differences that are statistically signi cant among the data based on categorical bases (Henseler et al., 2009).

Demographic Characteristics
The data were collected from the majority of the respondents who were female (59.3%). A total of 23.2% of the respondents were below 21 years old. The majority of the respondents were between 21-25 years old (61.7%). Respondents between 26-30 years old were 8.9%, 31-35 years were 2%, and the remaining were 36-40 years. The majority of the respondents were single (92.3%), and the remaining were married. The following are the percentage for education level: the majority of the respondents completed their bachelor's degree or equivalent (49.5%), secondary school level (19.3%), diploma or technical school level (28.5%), master level (2.5%), and the remaining completed their doctoral level. The following are the percentage for income: monthly income of less the RM2,500 (76.1%), income between RM2,501-RM5,000 (16.3%), income between RM5,001-RM7,500 (4.2%), income between RM7,501-RM10,000 (1.8%), and the remaining have an income of more than RM10,000. The majority of the respondents live in urban areas (87.8%). The following are the percentage for the place of origin: Chinese origin (74.6%), Malaysian (6.3%), Indian origin (5.4%), and other origins (13.7%). About 88.3% of respondents do not have ES, and the remaining respondents have ES.  (Chin, 2010). These results signify that the latent constructs have achieved adequate reliabilities and performed well for the subsequent analysis. The AVE for all items for each construct must be above 0.50 score to con rm convergent validity (Hair et al., 2019). The items display that the constructs have acceptable convergent validity (see Table 2.). All the VIF values for each construct are below the threshold of 3.3., displaying no issue of multicollinearity (Chin, 2010). The item loading and cross-loading are reported for the construct discriminant validity in Tables 3 and 4, respectively.   Table 3). Furthermore, the Fornell-Larcker criterion (1981) and HTMT ratio were employed to evaluate the discriminant validity of study constructs. The Fornell-Larcker criterion was estimated with the square root of the respective construct's AVE, and the square root of AVE for the construct must be higher than the correlation among other constructs . HTMT ratio needs to be less than 0.90 to establish discriminant validity for each construct (Henseler et al., 2015). Tables 3 and 4 show that the study has discriminant validity for each construct.  (Chin, 2010). The adjusted r 2 value for the one exogenous construct (i.e., WTP) on PES explicates the per cent 0.4 of change in the purchase of ES. The predictive relevance (Q 2 ) value for the part of the model is 0.003, indicating no predictive relevance (Chin, 2010).  PES is mediated by WTP. The result reveals that WTP mediates the relationship between MV and PES (β = 0.020, CI min = 0.008, CI max = 0.032, p = 0.003). The mediation results are presented in Table 7.

Multiple Group Analysis
Multiple group analyses were executed to match the results for different groups based on gender, living area, and education. One non-parametric test was employed to evaluate the differences in the vital association between the model based on the gender, areas of living, and education. Table 8 shows the path values for two groups with the differences within groups and the p-values as recommended by

Effects of Gender on the Groups
The results of the two groups are based on the gender of the sample. The gender has no signi cant difference in the relationship of the model. The variance of gender does not in uence the relationship between study models. The results are depicted in Table 8.

Effects of Living Area on the Groups
The results of the two groups are based on the living area of the sample. The living area shows a signi cant difference in the relationship between ATT and WTP for ES. The urban respondents have more ATT in in uencing WTP for ES. However, the variance of the living area (i.e., urban and rural respondents) does not in uence the variance between the study's other paths. Study results are provided in Table 8.

Effects of Education on the Groups
Page 20/32 The results of the two groups are based on the education of the sample. Education has a signi cant difference in the relationship between ECN and ATT, ENK, and ATT for ES. The variance of education does not in uence the variance between the study's other paths. Analysis results are shown in Table 8. 4.6. Importance Performance Matrix

Predictive Assessment
The sample predictive power of the model was estimated with the PLS Predict with ten folds and one repetition. This assessment establishes the performance of the PLS model with new predictive observations. The distribution of error is presented in Fig. 3. Most of the endogenous constructs' indicators outperform the naïve benchmark (Shmueli et al., 2019). The Q 2 predict value is above 0. Then, the prediction error is analysed in detail to evaluate the relevant prediction statistic. The error distribution plots show that the error distribution for all items of WTP (Fig. 3) is not highly non-symmetric (Shmueli et al., 2019). Therefore, this study assessed the predictive power based on the RMSE value. The error distribution for PES is highly non-symmetric, and this study used the mean absolute error (MAE) values to assess the predictive power based on PES.  Table 10). However, there is no predictive power for PES as the error is equal for the PLS-SEM and naive LM benchmark.

Factors Affecting Attitude towards ESs
The study's rst two hypotheses assessed the effects of ENC and ENK on ATT. The study ndings support the argument that ENC (f 2 = 0.106) and ENK (f 2 = 0.081) have a signi cant and positive effect on the ATT among Malaysian youth. The effect of ENC and ENK on ATT for ES is small (Chin, 2010). This study nding is parallel to the studies by Maichum et al. (2016) and Mostafa (2006)

Conclusion
Global environmental education is rising, and energy conservation, emission reduction, and environmentfriendly product adoption are on the rise. Our lifestyle and our transport mode are unsustainable (Rezvani et al., 2015). Individual mobility was on the rise in the past, and transportation remains a signi cant contributor to GHGs emissions. South Asian countries are known as the hub of people that use twowheelers (Eccarius & Lu, 2020). The rise of individual mobility needs to be sustainable like the use of EVs or speci cally individual mobility on the use of ES. Southeast Asians use two-wheelers for individual short distance travel (Rezvani et al., 2015). The use of two-wheelers is associated with the factors of weather, economy, and culture (Eccarius & Lu, 2020). Youth are more inclined to use ES.
Malaysia is among the top ten markets for two-wheelers. Converting individual mobility usage of twowheelers to ES can promote environmental bene ts that enable Malaysians to achieve SDGs 15. This ). An informed individual is always more inclined to use innovative and environmentally friendly products.
The associated limitations of this study are as follows. The consumers are different based on their personal attributes like willingness to experiment, innovativeness, and awareness. Future research may include these personal attributes to estimate the willingness to purchase ESs. The bene ts can be realised on the actual use of the innovation. This cross-sectional study lacked the actual use of ES by consumers. Therefore, future research models may be apt to use the longitudinal study design to fully grasp the attributes of ES that in uence the adoption of ESs. At the early stage of ES market, the government incentive role should be explored by future research. Besides that, innovation adoption is complex and requires the social-economic policy research to understand the adoption barrier like ES.

Declarations
Ethics approval and consent to participate Local ethics committees (Universiti Malaysia Kelantan, Malaysia) ruled that no formal ethics approval was required in this particular case. This study has been performed in accordance with the Declaration of Helsinki. Written informed consent for participation was obtained from respondents who participated in the survey. For the respondents who participated the survey online (using google form), they were asked to read the ethical statement posted on the top of the form (There is no compensation for responding nor is there any known risk. In order to ensure that all information will remain con dential, please do not include your name. Participation is strictly voluntary and you may refuse to participate at any time) and proceed only if they agree. No data was collect from anyone under 16 years old.

Consent to publish
Not applicable Availability of data and materials All data generated or analysed during this study are included in this published article (Submitted with the Manuscript: additional supporting le)

Competing interests
The author(s) declare(s) that they have no competing interests.
Funding Figure 1 Research Framework Figure 2 Importance Performance Map Figure 3