Students’ Acceptance for e -Learning and the Effects of Self-Efficacy in Malaysia

The study investigated the e-Learning acceptance among Malaysian higher education students. There are three exogenous variables involved, namely, performance expectancy, social influence, and perceived enjoyment. A mediating effect of self-efficacy was correspondingly tested to build a different connection point on the research area. The target population of the study are active students of Malaysia higher education institutions. Data was collected using an online platform, and out of the 557 responses received , only a total of 414 were valid and subsequently used for data analysis. The results indicated that, performance expectancy, social influence, perceived enjoyment, and self-efficacy have a positive direct statistically significant relationships with e-Learning acceptance among students. Additionally, there was a partial mediating effect of self-efficacy between performance expectations and perceived enjoyment on e-Learning acceptance. Meanwhile, social influence was found to have no mediating effect, since there was no statistically positive relationship between social influence and self-efficacy. Students with a positive feeling about the usefulness of e-Learning tend to have a positive acceptance of the e-Learning method, and this, Lecturers and students will be benefited from this study by considering significant factors and the importance of individual self-efficacy towards achieving an excellent understanding of the lessons. Hence, it is expected that, institutions and regulatory bodies can contribute towards a more productive and acceptable learning system to improve the quality of the education to be delivered.


Introduction
Technological advancement has transformed the ways daily activities are performed. From the past decade, the world of tertiary education has been rapidly involved in the advancement of internet technologies and the revolution of computer software (Tayebinik & Puteh, 2012). Like other developing countries, Information Communication Technology (ICT) in Malaysia has become a significant part of the national initiatives to maintain and improve the quality of public education, while fast becoming a competitive advantage for the institutions of higher learning. The development of ICT and the positive society response resulted in the adoption of e-Learning in the local education system, and Malaysia higher education began implementing it in the late 1990s (Hussin, Bunyarit, & Hussein, 2009).
E-Learning is one of the technical-based tuition and training platforms in telecommunication technology used to deliver information in education. Along with the progressive information and communication development, e-Learning is considered a paradigm in modern education. The e-Learning mode is offered for formal and informal education programmes in some local universities, such as University Tun Abdul Razak, Malaysia (UNIRAZAK), and Open University Malaysia (OUM). It has proven to be practical to address the concern of time and space limitation for interactions between learner and instructor, through asynchronous and synchronous network models (Katz, 2002;Katz, 2000).
E-Learning is one way of learning methods, where students can learn individually at their preferred time, unlike the traditional classroom learning method. It is home-based and the courses designed can be altered to suit learners' needs and preferences (Al-Rahmi et al., 2018). However, The absence of face-to-face communication among learners and the new e-Learning environment are challenges which must be overcome by the learners (Tayebinik & Puteh, 2012). Therefore, individual effort and readiness are vital to ensure an excellent delivery of knowledge. As self-efficacy is related to individual belief in own capabilities to organize and execute the courses of action required to produce given attainments (Bandura, 1997), it builds a bridge of potential connections on the e-Learning acceptance. People who have high self-efficacy in technology will have a higher perception that, learning by the use of technology is useful to themselves. In contrast, those with low self-efficacy perceive using learning by technology as a burden and this could negatively affect their acceptance of e-Learning. This is a critical issue to address, as it could potentially hinder students' understanding and adaptability to e-Learning, while the actual usage of e-Learning is determined by the self-efficacy factor (Lwoga & Komba, 2015).
As e-Learning in Malaysia is still at the infancy stage, addressing the factors that influence students' acceptance to it is vital. The continued usage of this method is mainly because of performance expectancy, self-efficacy, social influence, and other factors (Lwoga & Komba, 2015). Importantly, demand for e-Learning continually increases as this approach can reach a global audience, unique functionality, accessibility, and flexibility in a long time (Azhari & Ming, 2015).
Moreover, as the globe embarks on Industrial Revolution 4.0 (IR 4.0), it is clear that technology in learning enables students to be more dependent in acquiring knowledges which meet industry requirements (Chee, Ab Jalil, Ma'rof, & Saad, 2020).
Therefore, this study aimed to establish the determinants of e-Learning acceptance and to understand the potential effects of students' self-efficacy for a better understanding and development of e-Learning in Malaysia using the unified theory of acceptance and use of technology (UTAUT) and technology acceptance model (TAM) as a basis of the study. The objectives of the study are as stated below: • To investigate the predictors of e-Learning acceptance among students in Malaysia.
• To examine the mediating effects of self-efficacy between predictors and e-Learning acceptance.
Literature Review E-Learning E-Learning is a method of learning and teaching which encompasses fully or partially education models, based on the use of electronic media and devices as a tool for increasing access for communication and interaction, training, and it facilities new ways to understand and establish learning (Selvarajah, Krishnan, & Hussin, 2017). It can be used and accessed using several types of technology without specific time and barriers. Any electronic media web technologies can deliver e-Learning efficiently, making it to be more readily accepted by academic institutions and business organizations, compared to web-based media (Hiltz & Turoff, 2005). Indeed, e-Learning has become the most crucial part of competitive educational service. Besides, to meet the huge demand of educational customers, learning institutions offer online lessons, online tests, and online educational consulting (Lee, Yoon, & Lee, 2009). E-Learning, similar with other education platforms, contains its own strength, weakness, opportunity, and threats (Schroeder, Minocha, & Schneider, 2010). It can be used to establish a good community spirit among the learners, to reduce cost (travelling cost), to improve communication among learners and instructions, and to improve independent problem-solving skills among students. However, it has a limitation of interaction quality that leads to less trust in terms of feedback and team activities, compared to traditional learning methods. E-Learning also increases the workload of students and instructors, especially the time and efforts taken for the preparation purposes. Indeed, it is difficult to ensure the reliability of the learning services provided (Schroeder et al., 2010). That is the reason it is vital to assess students' acceptance of the e-Learning method. In addition, fear of technology, lack of technical skills, lack of technical support for both students and lecturer may also cause some concerns (Ali & Magalhaes, 2008). Nevertheless, e-Learning enables students to produce high-quality work and to be actively involved in alumni community activities. Educational institutions too benefit from it, by gaining an exposure and adding values to their programmes around the globe, besides responding towards IR 4.0 (Schroeder et al., 2010).

Underline Theory Technology Acceptance Model (TAM).
The technology acceptance model (TAM) was been developed in 1989 and has since become the most popular research model to determine and predict use and acceptance by an individual regarding information systems and technology. It is an adaptation from the previous model, the Theory of Reasoned Action (TRA), to the field information system. For this study, one relevant variable, namely, perceived enjoyment was adopted.

Unified Theory of Acceptance and Use of Technology Model (UTAUT).
The unified theory of acceptance and use of technology (UTAUT) has received significant attention in e-Learning and technology in education, as it provides a reliable basis to investigate individual perceptions related to technology in education (Venkatesh, Morris, Davis, & Davis, 2003). It is used to evaluate the success of new technology applications and it was derived from previous models of technology acceptance (TAM). The UTAUT has been used extensively in past studies to investigate user's technology acceptance (Tagoe, 2012). It was applied in this study via two relevant variables, namely, performance expectancy and social influence as independent variables.

Variables of the Study Performance Expectancy
Performance expectancy is an individual's belief from the advantage and usefulness they gain through the use of technology and system (Venkatesh et al., 2003). By adjusting performance expectancy in the e-Learning context, it can be great assistance, because learners can complete their learning activities and directly enhance their education skills and performance (Salloum & Shaalan, 2018). Performance expectancy has become one of the factors which influences behavioural intention among students to use e-Learning in their studies (Mahande & Malago, 2019;Zawaideh, 2017). It has been proven to produce the most impact on student's positive acceptance toward usage intention and the highest significance, compared to the other variables in UTAUT (Chung, Shen, & Qiu, 2019). Another study reported that, performance expectancy was an essential contributor in foretelling students' intention to use the mobile learning system (Bharati & Srikanth, 2018). Thus, it is critical for the study to verify the relationships between social influence and e-Learning acceptance.

Social Influence
Social influence refers to how much an individual perceives that others believe themselves should use the system as people around influence, individual action, and reaction (Venkatesh et al., 2003). Thus, social influence potentially affects e-Learning acceptance, as the technology in education is affected by social rather than technological factors. Relatively, social behaviour could affect a user's opinion, adoption, and performance, especially in a collectivist culture (AlMarshedi, Wanick, Wills, & Ranchhod, 2017). A previous study revealed that, social influence affects individual intention to use technology (Tan, 2013;Yoo, Han, & Huang, 2012). Based on the findings in some past studies, social influence has positive relationships with the students' perception and attitude towards their readiness to use e-Learning. (Mahande & Malago, 2019;Ngampornchai & Adams, 2016). Indeed, social influence has been found to be an important influence on students' decision to use e-Learning (Lwoga & Komba, 2015) continuously. Thus, it is critical for the study to investigate the relationships between social influence and e-Learning acceptance.

Perceived Enjoyment
Perceived enjoyment is defined as the extent, to which the activity of using a system is perceived to be enjoyable in its own right, aside from consequences resulting from the system. It can explain behavioural intention as being an acceptance to use information systems (Punnoose, 2012). Students' subjective feelings of joy, relaxation, and positive experience also play roles in explaining user acceptance and usage behaviour of e-Learning. If students do not enjoy the e-Learning process, they will certainly not be involved again and this will negatively affect their learning performance. Perceived enjoyment directly has its impacts on behaviour attention and indirectly affects influences through attitude (Lee, Cheung, & Chen, 2005). However, some other studies also highlighted that, perceived enjoyment has no direct influence on intention to use (Venkatesh et al., 2003;Yi & Hwang, 2003), Thus, it is critical for this study to investigate the relationships between perceived enjoyment and e-Learning acceptance.

Self-efficacy
In general, self-efficacy refers to an individual's belief in his or her ability to perform a particular behaviour (Bandura, 1997). Self-efficacy in the online learning context refers to an individuals' judgement on his or her ability to use online learning in daily activities, including the use of the internet, computers, web-based instructional, and learning tools. People who have high selfefficacy in technology will have a positive perception of e-Learning and vice versa. Students' computer anxiety is one of the critical factors affecting their satisfaction. Once dissatisfied, their belief to use the technology as a medium will be directly affected (Sun, Tsai, Finger, Chen, & Yeh, 2008). Self-efficacy has a positive relationship with and is a significant factor for students' intention to use e-Learning (Al-Rahmi et al., 2018). Despite the potential of self-efficacy as a mediator towards e-Learning acceptance among students, there is a lack of study to test this relationship. Thus, this study is essential to investigate the mediating effect of self-efficacy between potential relevant variables affecting e-Learning acceptance and the e-Learning acceptance itself.

Research Design
The study adopted a correlational research design to investigate potential relationships between variables involved without manipulating them. SPSS and AMOS software were used to examine the causal relationships between these variables.

Study Population and Sampling Procedure
The study was conducted in higher education institutions in Malaysia. The target population of the study encompassed active students currently enrolled in undergraduate and postgraduate levels (N -1,343,830) (Ministry of Education Malaysia, 2018).
To calculate the required sample size, a 95% confidence level and a 5% margin of error were applied as it is acceptable in education research and categorical data use (Ary, Razavieh, & Jacobs, 1996). The minimum sample size, according to the population of the study, is 384 (Sekaran & Bougie, 2016). The data was collected using convenience sampling. An online self-report questionnaire was used to ensure broad accessibility and generalizability. An online approach has increasingly become important because people are more inclined towards technology with their desires to know more about a wide range of topics (Brick, 2011). The online survey is also able to reduce the social desirability bias associated with the traditional face-to-face survey.
The link of the survey was subsequently distributed through social media and the participating students' online group. Filter questions were used to verify that the respondents are currently active students of higher institutions before they could proceed to the next question. The data was collected in a period of almost one week, with a total of 557 responses received, out of which, 414 responses were found to be valid, still meeting the minimum required sample size of 384.

Respondent Profile
Based on the descriptive analysis in Table 1, the majority of respondents are female (77.3%), compared to males (22.7%). They are mainly aged between 18 -21 years old (67.9%), pursuing undergraduate courses (Diploma and Bachelor's Degree). The students are mostly from public institutions of learning, including Colleges and Universities.

Measurement and Instrumentation
The questionnaire used in this study was adapted from previous literature obtained through an in-depth analysis of literature addressing the research objectives. The demographic profiles instrument of the study was developed based on the research objectives. Meanwhile, of the five remaining instruments, four of them, namely, (i) Performance expectancy, (ii) Social influence, (iii) Perceived enjoyment and (iv) E-Learning acceptance were adapted from well-established instruments and fairly tested for validity and reliability (Efiloğlu Kurt & Tingöy, 2017;Ngampornchai & Adams, 2016;Niehaves & Plattfaut, 2014). while the last one, namely, (v) Self-Efficacy, the mediator variable of the study, was adapted from the development and validation of a students' self-efficacy scale (Schmitz, 2013).

Data Screening
Before a further analysis was conducted, the raw data was coded, and the missing value analysis had been conducted through a minimum and maximum analysis to ensure that, the data was coded accurately. The normality of the data had been tested using Box Plot analysis in SPSS and outlier analysis in AMOS to remove extreme values that could affect the result to be obtained. The final number of valid respondents was 414 out of 557 of the total responses received.

Full Measurement Model
The full measurement of the research model was conducted to test the model's fitness and validity. The Confirmatory Factor Analysis (CFA) with scores of model fitness is as presented in figure  1. The CFA reported a good model fit score with CMIN/DF = 2.673; CFI = .951; TLI = .946 and RMSEA = .064 (Awang, Hui, & Zainudin, 2018). The validity of the construct on Table 2 indicates a good convergent validity with the AVE score of all construct score of 0.50 and above. Similarly, construct validity as the model fitness scores a good result. The reliability test of CR similarly achieves a good reliability with a CR score 0.60 and above. Therefore, the construct is validated.

Structural Equation Modelling (SEM) Analysis
To examine the direct relationships and the mediation effect of the construct to describe the effect of performance expectancy, perceived enjoyment, social influence, and self-efficacy toward e-Learning acceptance. As illustrated in Figure 2, the result of model fitness shows that, the data collected fitted the model, where a minimum fitness required was achieved with CMIN/DF =2.673; CFI=.951; TLI .946 and RMSEA =.064 (Awang et al., 2018). Therefore, a further analysis can be administered using the structural model.

The Direct Effects Hypothesis
The summary of the results obtained as attached in Table 2. The results revealed that, performance expectancy had a positive and statistically significant effect on self-efficacy (β = .342; C.R = 3.098; p = .002). When performance expectancy went up by 1 standard deviation, self-efficacy went up by .342. Thus, H1 was accepted (The performance expectancy has a significantly positive relationship with students' self-efficacy). H2 was also accepted (The perceived enjoyment has a significantly positive relationship with students' self-efficacy). The analysis indicated a positive and statistically significant relationship of the H2 (β = .320; C.R = 3.511; p = .000). When perceived enjoyment went up by 1 standard deviation, self-efficacy went up by .320. Somehow, the result revealed that, social influence was statistically insignificant with self-efficacy (β = .155; C.R = 1.637; p = .102). Thus, H3 was rejected (The social influence has a significant positive relationship with students' self-efficacy).
H4 was accepted (Self-efficacy affects e-Learning acceptance positively). The analysis indicates a positive and statistically significant relationship of the H4 (β = .115; C.R = 3.640; p = .000). When Self efficacy went up by 1 standard deviation, e-Learning acceptance went up by .115. Besides, Table 2 shows a positive and statistically significant effect of social influence on e-Learning acceptance (β = .226; C.R = 4.120; p = .000). When social influence went up by 1 standard deviation, e-Learning acceptance went up by .226. Thus, H5 was accepted (Social influence affects e-Learning acceptance positively).
Correspondingly, H6 was also accepted (Perceived enjoyment affects e-Learning acceptance positively). The analysis indicates a positive and statistically significant relationship of the H6 (β = .484; C.R = 8.871; p = .000). When perceived enjoyment went up by 1 standard deviation, e-Learning acceptance went up by .484. Lastly, the analysis showed a positive and statistically significant effect of performance expectancy on e-Learning acceptance (β = .179; C.R = 2.768; p = .006). When performance expectancy went up by 1 standard deviation, e-Learning acceptance went up by .179. Thus, H7 was accepted (Social influence affects e-Learning acceptance positively). Table 3. Analysis of the direct relationship Note: B (unstandardized regression weight); S.E (standard error); β (standardized regression weight; C.R (critical ratio).

The Bootstrapping and Hypothesis of the Mediation Effect
The bootstrapping test was applied to examine the mediating effect of self-efficacy purposely to assign measures of accuracy to sample estimates (Darren & Paul, 2020). The study selected a 2000 bootstrap sample with 90% of the bias-corrected confidence interval. The summary of the results is as in Table 4. Generally, all three independent variables tested (Performance expectancy, perceived enjoyment, and social influence) had a statistically significant relationship with e-Learning acceptance (x → y) with a p-value of less than 0.05. However, to confirm the mediation effect, the indirect effect and the significant value between any relationship must be assessed. The first mediation relationship (self-efficacy) to be tested was between performance expectancy and e-Learning acceptance. There was a positively standardized indirect effect with a score of 0.037 (p = .000) which supports H8 (Selfefficacy mediates the relationship between performance expectancy and e-Learning acceptance). Thus, there is a partial mediation effect of the first mediation relationship tested.
Correspondingly, the analysis in Table 4 supports H9 of the study (Self-efficacy mediates the relationship between perceived enjoyment and e-Learning acceptance). There is a positive standardized indirect effect of the relationship tested for H9 with a score value of 0.039 (p = .05). Hence, there was a partial mediation effect of the second mediation relationship tested. However, H10 of the study was rejected (Self-efficacy mediates the relationship between performance expectancy and e-Learning acceptance) because of the standardized indirect effect score of 0.018 (p = .126). As the p-value score is more than 0.05, the relationship is insignificant. In conclusion, there is no mediation effect of the third mediation analysis tested.

Conclusion and Discussion
Based on the analysis, performance expectancy, perceived enjoyment, and social influence were found to be significant with students' acceptance of e-Learning. Meanwhile, only performance expectancy and perceived enjoyment were found to be significant with students' self-efficacy. Correspondingly, self-efficacy was significant with students' acceptance of e-Learning, and mediating effects of self-efficacy were identified between performance expectancy and perceived enjoyment toward e-Learning acceptance.
The results showed that, performance expectancy has significant effects on students' acceptance of e-Learning, in agreement with findings from other studies which revealed that, performance expectancy has significant effects on students' intention to use e-Learning (Salloum & Shaalan, 2018;Masa'deh, Tarhini, Bany Mohammed, & Maqableh, 2016). Theoretically, students with a positive feeling about the usefulness of e-Learning will have more intention to use it (Tarhini, Hone, Liu, & Tarhini, 2016). Thus, when the performance of e-Learning meets students' perceived performance, it is considered useful.
Similarly, social influence was found to be significant towards e-Learning acceptance. Social influence and facilitating condition positively influence continuance intention to use e-Learning among higher education students, and this explains the results obtained (Bakar, Zaidi, & Abdul, 2014), which might be due to a cultural setting of Malaysia as a collectivist country as the technology in education is influenced by social rather than technological factors. Relatively, social behaviour could affect a user's opinion, adoption, and performance, especially in a collectivist culture (AlMarshedi et al., 2017).
Meanwhile, perceived enjoyment was found to significantly influence students' acceptance of e-Learning. According to Al-Gahtani (2016), there is a significant relationship between perceived enjoyment with perceived ease to use. Students with a high perceived ease to use will have positive effects on their intention to use e-Learning. However, this is in contrast with the finding in a study by Hussein (2018), which showed that, perceived enjoyment was not a significant factor in students' engagement with e-Learning. Similarly, for the intended use of e-Learning, previous studies highlighted the insignificant effect of perceived enjoyment (Venkatesh et al., 2003;Yi & Hwang, 2003). Somehow, perceived enjoyment does not just directly affects individual behaviour, it also indirectly gives influences through attitude (Lee et al., 2005). Thus, this explains the contradictions in the findings.
Besides, the mediator analysis confirms the significant effect of the direct relationship between self-efficacy and e-Learning acceptance. The finding is in agreement with that in a previous study in an analysis between factors of student acceptance and intention to use e-Learning (Al-Rahmi et al., 2018;Lwoga & Komba, 2015). People who have high self-efficacy in technology will have a positive perception of e-Learning and vice versa. However, only performance expectancy and perceived enjoyment have a partial mediation effect upon self-efficacy toward e-Learning acceptance. High self-efficacy will enhance students' acceptance of e-Learning. However, social influence is not directly significant towards self-efficacy and this leads to a no mediating relationship. The self-efficacy effect as a mediator is supported by a previous researcher who found the impacts of students' belief to use the technology as a medium for their learning (Sun et al., 2008). As selfefficacy mainly concerns on individual belief and internal self-control rather than social belief, this thus explains the insignificant mediating effect of self-efficacy on social influence.

Recommendation and Implication
This study donates toward a new theoretical contribution by incorporating self-efficacy as a mediator in between the field of e-Learning research theories of TAM and UTAUT. An additional theoretical link was developed from the hypothesis derived from a rationale application of selfefficacy. Subsequently leading to the improvement of the rational theory by giving a new contribution for the current study. Besides that, a theoretical linkage was tested with the exogenous construct and self-efficacy construct. A test which had been minimally tested in previous researches. The result of the test has produced an empirical contribution for the current study. Furthermore, the current study had correspondingly determined the degree of self-efficacy which mediated the relationship between exogenous variables and endogenous variables. Thus, by employing the suitable sampling procedure which increased the generalisability of research and the targeted population, who are students of Malaysian higher institutions. As such this has essentially contributed towards the methodological gap of the current study.
In the current digital era, students are very confident in the digital domain and they are now searching similar exposure and ease in their academic life (Newland & Byles, 2014). Their acceptance is influenced by their perception that this system can improve their performance in study, support from people around them, and their feeling towards this system. Thus, internal self-control factors, such as self-efficacy are critical to be investigated in an online learning environment (Alqurashi, 2016). However, a study of e-Learning acceptance and self-efficacy is deficient in the Malaysia setting, and it could affect e-Learning acceptance positively or negatively (Alqurashi, 2016). Thus, it is expected that, this study will contribute towards new theoretical and practical contributions for related stakeholders.
The results of this study highlighted several implications for stakeholders involved. This study can benefit researchers to increase their understanding regarding e-Learning and the relationship of self-efficacy that could be used to potentially build a bridge for a more in-depth study on the area, to ensure the effectiveness of e-Learning. It can encourage more studies related to e-Learning, and perhaps this can be a source of reference for future research to respond towards the shifting of learning approach preferences and development of technology aligned with IR 4.0. Lecturers and students can be facilitating the effectiveness of e-Learning sessions by considering significant factors and the importance of individual self-efficacy towards achieving an excellent understanding about the taught lessons. This study also provided an overview for the institutions of higher learning on what can be done to ensure that, this learning system is more effective and acceptable to students as they can be influenced by lecturers and their institutions. Indeed, it is also able to increase student satisfaction that is vital in the current competitive era to ensure institution sustainability (Latip, May, Kadir, & Kwan, 2019). Lastly, this research can help increase the number of studies related to e-Learning, besides being a source of reference for future research work.