An empirical study of Chinese students' behavioral intentions to adopt 5G for smart-learning in Covid-19

At present, the digital revolution and virtualization have dramatically altered all economic, social, and formal education systems (Lorenzo & Gallon, 2019; Winthrop et al., 2016). We have also discovered and intensely studied technologies used to help students learning. Today, advanced technology is used as a means or tool to access learning content (Daniel, 2012), research, construction, communication (Bruce & Levin, 1997), assessment (Meyer & Latham, 2008) and expression (Goodman, 2003). With the mobile and advance internet technology, mobile learning has become an essential mode of learning. Compared with traditional static education, mobile learning emphasizes students' mobility. Further, the support of ubiquitous technology has brought about other changes, changing the way of learning from mobile to ubiquitous, emphasizing that learning can be carried out anywhere, regardless of time, location, or environment (Hwang et al., 2008). Today, Smart learning is a novel model of using IT in educational institutions. The global covid-19 has instigated large-scale behavioral and institutional shock effects in almost all parts of life. The influence of this epidemic on learners is extraordinary (Chang et al., 2021). It restricted more than 150 million students worldwide to their homes (Teräs et al., 2020). Due to large-scale shutdowns, affected countries have been enforced to seek out speedy solutions and to shift to online learning (Jandrić 2020; Presti et al., 2021). The prompt change from classroom teaching to online has left behind deeper insights related to national education policy, theoretical basis and premise (Sunarto, 2021). This new environment enabled students to use advanced technology and techniques of learning (Cheung, 2014; Fayez et al., 2021; Marinova et al., 2017). Researchers and educators argue that this concept (smart learning) should not be restricted to only smart devices use. This kind of environment is a digital environment or virtual place, through a structured approach or a self-regulated process, fully supervised or semi-supervised to learning and teaching (Cook, 2010). With the increase in technical complexity, the concept of smart learning is evolving into a contextualized and virtualized ubiquitous environment. The modern concept of smart learning has become so clear that the limits between informal and formal learning have almost disappeared (Annoni & Kozovska, 2010). Smart institutions are converting into social and personal centers of learning, where teachers, students, and entire communities can take part in activities (Lorenzo & Gallon, 2019). Hwang (2014) and Scott and Benlamri (2010) believe that smart learning is ubiquitous and context related technique. Gwak (2010) suggested the idea of smart learning as it has more content and learners than device-oriented learners and it is efficient, intelligent, and personalized learning. Kim et al., (2012) believe that this type of learning is service-oriented and learner-centered, rather than using the equipment. Middleton (2015) also illustrates learner-centred smart learning and how they can benefit from smart technology. According to the Korean Ministry of Education, Science and Technology Smart Learning is as follows:

S: Self Directed: It means that the system of education moves towards a system of self-learning more than ever before. The role of students changes from adopters to creators of knowledge.

M: Motivated: It refers that education seeks creative problem solving and personalized assessment while being mindful of the experience.

A: Adaptability: Adaptability means increasing the flexibility of the system of education and adapting learning to personal preferences for future careers.

R: Resource enriched: Resource enriched says that smart-learning uses rich content for both the public and private sectors.

T: Technology embedded: Technology embedded means that students can learn anytime and anywhere through advanced technology in the education environment.

The government, academic, and industrial researchers should find ways to integrate existing technologies into 5G technology. They should also find possible ways to build an adoption environment as this will overcome all the limitations in the existing technologies.

Learning through advanced technology provides a platform for students to learn regardless of their location (Khan et al., 2019). Some empirical research has discovered how such type of learning affects pedagogical tactics, endorsing that new pedagogical approaches in education could help students learn (Khan et al., 2019). According to growing studies, internet technology and mobile devices are increasingly being used to assist students all around the world (Churchill et al., 2016). Several studies (Alharbi et al., 2017; Khan et al., 2019) assert that learning assisted by advanced technology has worked as a significant accelerant in education systems, where the evolution of new learning approaches has profited from a momentous evolution that boosts the learning efficiency of both educators and students. The integration of internet and mobile technologies in education appears to have yielded positive results. Global revenue through E-learning is predicted to reach $65.41 billion by 2023. However, as a result of the covid-19 epidemic and the evolution of the 5G internet technology, education has fundamentally changed its essence. When using the smart learning concept, it is beneficial to undertake a more detailed study to finalize various learning strategies (Putnik, 2016). Before applying smart learning in education, student’s perceptions about the technology should be explored. The majority of past research has concentrated on students' adoption of learning through technology, with intent as a dependent variable. Using the Unified Theory of Acceptance of Technology (UTAT), (Abu-Al-Aish & Love, 2013; Venkatesh et al., 2003) discovered that effort and performance expectancy, quality of service and personal innovativeness affect mobile learning acceptance. When Briz-Ponce et al., (2017) evaluated the variables of learning technology acceptance using TAM, they discovered that perceived ease of use and perceived usefulness influenced the adoption of such technology. Lin et al., (2020) explored students learning intention using TAM ∪ TPB models. Smart learning, on the other hand, is a brand-new approach to education that has emerged in the previous decade and got importance, particularly after Covid-19. As a result, we anticipate that users will assent or reject smart learning based on their willingness to adopt smart learning technologies. Therefore, our study extended the traditional TAM by incorporating three new factors from social practice theory (SPT). Although there have been several studies in this field, research findings of student’s adoption of smart learning in covid-19 are still few. Research on Chinese students’ intentions of smart learning is a gap that needs to be filled.

PEOU is an essential and repetitive factor proposed in TAM and has been used widely in accepting technology (Hamidi & Chavoshi, 2018; Shah et al., 2021). PEOU indicates how little effort is expected to be necessary to use a system (Davis, 1989). It is an individual's belief in eradicating physical and mental stress in a specific area. Joo et al., (2014) revealed that it is a student’s belief of using technology without any difficulties. In this study, PEOU is related to the easy access, use, interface, and flexibility of 5G smart learning technology. It is the degree of effortlessness associated with using 5G technology. In the initial phases of practising technology, simplicity of use is essential as the expectation of effort is considered necessary for using the technology (Wang et al., 2016). Moreover, users will find easy technology use more beneficial (Huang & Hsieh, 2012). Therefore we assume that:

H1: PEOU of 5G technology for smart-learning is positively related to the student’s intention (UI).

PU is another key factor derived from TAM and is used in new technologies acceptance models (Hamidi & Chavoshi, 2018). According to Davis (1989) and Trikoilis (2021), the assumption that using a specific technology would improve performance is referred to as PU. Althunibat (2015) also explained PU as the degree to which a system is personally dependent on improving technology performance in a defined sphere. In a smart learning context, PU signifies a certain level of confidence that smart learning will lead to improved student performance (Hao et al., 2017; Uwantege et al., 2021). Joo et al., (2014) also defined PU as the belief that technology will help to achieve student’s educational objectives. Collaboration with teachers and peers will improve efficiency in completing certain tasks. Sabah (2016). Students will embrace technology if they believe that 5G technology will improve their learning performance (Ali & Arshad, 2016; Chai et al., 2021). According to the previous research, PU has a positive influence on adoption intention. Therefore:

H2: PU of 5G technology for Smart-learning is positively related to the student’s intention (UI).

BI defines the intentions of a user to perform a particular behavior (Davis, 1989). Studies have proven that these intentions are highly correlated with acceptance and use (Hassanzadeh et al., 2012; Mohammadi, 2015; Shah & Zhongjun, 2021). Besides, most theories in the field of technology adoption use behavioral intention as a prerequisite for user acceptance (e.g. TRA, TPB, and UTAUT). Behavioral intention (BI) is considered the core structure of TAM that predicts student’s smart learning acceptance.

"Material" is an essential part of the practice. In practice the material elements are the infrastructure, objects, hardware etc. Materials can be seen as spanning spaces of individual and social opportunity, as the availability of certain devices may explain the differences in technology usage behaviour. Changes in behaviour are linked to technological advancement. In Social practice theory, "things" are not only discourses of symbolic meaning or identity (Shove & Pantzar, 2005; Warde, 2005), but also directly participate in the behavior and reproduction of everyday life (Shove & Pantzar, 2005). In the current study, the 5G technology quality content contains rich and continuously updated learning content and facilitating conditions for smart learning practice (Almaiah et al., 2016). 5G technology will help in delivering content that is highly abstract and difficult to reconstruct. With the help of 3D virtualization technology, students will gain a profound understanding of reading. It will produce computer-generated images same to real-world content. This technology will integrate simulation and animation into education to provide the most challenging educational content better. Similarly facilitating conditions mean the existence of organizational as well as the infrastructure to back the system (Venkatesh et al., 2003). In the field of IT, It includes knowledge, resources, and support personnel. Users may not be able to practice 5G for smart learning in the absence of these conditions (Iqbal & Qureshi, 2012; Shove & Pantzar, 2005). Compared with traditional learning, smart learning is a new concept. Therefore, its execution needs users who have an understanding of the applications and services. Base on this, we hypothesis that;

H3a: The Material Access (MAA) to 5G technology for smart learning is positively related to Perceived Usefulness (PU).

H3b: The Material Access (MAA) to 5G technology for smart learning is positively related to the Perceived Ease of Use (PEOU).

'Meanings' are primarily based on Bourdieu (1984) concept of habitus, which proposes that a group's perception of importance is shared, thereby unifying the group. Meaning is specific to an act or thing. Shove et al., (2012) explains that this theory stresses tacit and unconscious forms of experience and knowledge, through which a collective form of understanding is established. In the context of 5G technology for smart learning, these are the individual beliefs on learning concerning the classmates, teachers, and parents etc. In general, people may think about how people close to them view their practice of technology. This inspiration may come from academic personnel or people with high social status. Further, people make self-sacrificing purchases for the benefit of others or society. The covid-19 pandemic is an unprecedented situation, which has a great significance for understanding consumer moral decision-making during and after a long-term epidemic. These environments provide users with an opportunity to imitate the fundamental meaning of consumption and its effect on themselves, society, and the environment. The covid-19 epidemic surprised consumers that their basic needs might not be completed because they may not have access to food and basic needs. At the same time, it has changed the consumer's view on how to meet social and personal needs. In terms of 5G technology consumers will consciously consider consumption and will make adoption decisions to be responsible to themselves and society. There may be a significant shift towards responsibility and pro-social consumption. Based on this solid empirical support, we assume:

H4a: The Meaning Access (MEA) to 5G technology for smart learning is positively related to Perceived Usefulness (PU).

H4b: The Meaning Access (MEA) to 5G technology for smart learning is positively related to the Perceived Ease of Use (PEOU).

Competences mean "embodied knowledge," which originates from the studies of Bourdieu (1984) and Shilling (1991). Shove et al., (2012) define Competencies as understanding and knowledge to signify the type of experience necessary to practice successfully. Practice constitutes a "block," and its existence must depend on reality and existence. The specific interdependence between these elements cannot be condensed to any of these unique elements (Reckwitz, 2002). The competence part of practice may relate to the social-psychological theories. In the context of 5G technology for smart learning, it means the desire of an individual to accept and use new technology or risk-taking and attempts to practice new technologies (Hao et al., 2017). Highly innovative people are more willing to give a positive response to new technologies (Milošević et al., 2015). Besides, those who learn technology, are more likely innovative than others (Joo et al., 2014; Milošević et al., 2015). Kim et al., (2017) stated that such people are more concerned with attaining information about using novel technologies. Students with innovative behavior want to assent the risk of using 5G technology and are more inclined to use it. Similarly, some people think that when innovative technology is deemed to be compatible with work practice, they may realize the practicability of technology (Chau & Hu, 2001; Moore, 1991). In the context of Smart-learning through 5G, increased competencies access will positively impact as new 5G technologies will be compatible with existing technologies. Based on these arguments, we assume that:

H5a: The Competency Access (COA) to 5G technology for smart learning is positively related to Perceived Usefulness (PU).

H5b: Competency Access (COA) to 5G technology smart learning is positively related to the Perceived Ease of Use (PEOU).

The questionnaire of this study consisted of five questions about demographic information and 24 queries about smart-learning influencing factors (MAA, COA, MEA, PEOU, PU, and BI). These items were selected from the technology adoption literature. The TAM (Davis, 1989) is used as a base theoretical model here. Hence the factors PU and PEOU are adopted from it to know the experiences of consumers' willingness to use 5G technology for learning. Material access (MAA) items are adapted from Almaiah et al., (2016) and Cho et al., (2009). The measurement items of the Competency access (COA) are adjusted from Chau and Hu (2001) and Hao et al., (2017). Similarly, the measurement items of the Meaning access (MEA) are derived from Venkatesh et al. (2003) and Du et al., (2015). For the above measures, the five-point Likert scale (1 = strongly disagree to 5 = strongly agree) was adopted. The survey was conducted in Chinese to cater to local environments. To verify the items, a pilot survey was undertaken.

We examined 4G and 5G users to verify the hypothesis. Whether in electronic or paper forms, the questionnaire was optional. In addition, people were asked to use the snowball method to share questionnaires with classmates and friends. To maintain anonymity, all respondents were notified that their responses would be kept anonymous and used solely for academic purposes. To represent our proposed model design, the questionnaire was divided into demographic details and measuring parameters. From September 2020 to November 2020, through an online questionnaire website http://​www.​sojump.​com, a formal survey was conducted online. Finally, 375 valid responses were selected. Table 1 illustrates the demographic details of these respondents. The study results showed that both men (54.9%) and women (45.06%) were involved almost equally, with more than 71% of participants aged 20 to 40. About 29.06% of respondents have higher education and 62.9% have a university degree. In terms of user experience and similarity, both 4G (56.5%) and 5G (43.4%) users were examined.

The validity and reliability of an external model's can be evaluated through the items' reliability, convergent, and discriminant validity. Reliability is the degree to which an item set designated for a specified construct measure a similar construct, stay reliable in distinct situations. The term validity means how to fit the instrument chosen items of a given construct are realistic. Reliability can be measure through Cronbach's alpha, composite reliability (CR) and factor loadings, convergent validity is through Average Variance Extracted (AVE), while discriminant validity is through the square root of AVE. According to Hair Jr et al. (2016) indicators, outer loading values greater than 0.6 should be retained, while the rest should be removed to increase the AVE or CR values. A total of 24 items were verified, as presented in Table 2 and Fig. 2. Cross-loadings were examined to check the items discriminant validity (Hair Jr et al., 2016). As the cross-loadings amongst constructs were more significant than the determined critical point, it verified the discriminant validity.

Construct validity (CR) is another technique to assess the outer model. It determines that these processes are essential tools for expressing and measuring the investigative constructs (Hair Jr et al., 2016; Gefen and Straub, 2005). Further Convergence validity is the degree that relates or converges measures of a similar construct (Hair Jr et al., 2016). When the explained AVE's value is equivalent to or exceeds 0.5, convergence's effectiveness is verified (Fornell & Bookstein, 1982). The AVE scores of all constructs were more than 0.5, which justifies the convergent validity (Table 3). It can also be inspected through the CR of the constructs (Fornell & Bookstein, 1982). By surpassing the 0.60 threshold value, all constructs verified the composite reliability (Hair Jr et al., 2016). When calculating Cronbach's alpha to determine internal consistency, the reliability assessments should be more than 0.70 (Field, 2009). As mentioned in Table 3, all the Cronbach's alpha values surpassing the 0.70 thresholds recommended by Field (2009), Hair Jr et al. (2016) and Shah and Zhongjun (2021), thereby achieving the convergent validity second condition. Generally, the model is suitable and effective for inspecting the significance of the paths associated with these variables.

Discriminant validity also examines how distinct the latent variable is from other factors (Hair Jr et al., 2016; Shah & Zhongjun, 2021). Examining the correlation matrix between constructs was one popular approach for determining discriminant validity. Especially, each prospective construct's AVE should be higher than its topmost square correlation with any other potential construct (Hair Jr et al., 2016). Table 4 demonstrates discriminant validity since all constructs in the proposed model fulfil this requirement because no off-diagonal element surpasses the diagonal element.

The PLS method evaluated the structural model to determine the path relevance and predictive effect and then used the bootstrap procedure to determine the path coefficients' significance level by evaluating standard errors, confidence intervals, and T statistics (Hair Jr et al., 2016). Table 5 highlights the study hypothesis and displays the path coefficient of the latent variable as well as the critical bootstrap ratio. The bootstrap T statistic defines the estimate's stability. A 95% confidence interval above 1.96 is considered acceptable (Hair Jr et al., 2016). In the proposed model, all research hypotheses were supported. The next part will explicate the outcomes of each path. Figure 3 also illustrates the tested and verified proposed model on SmartPLS3.0 software. These conclusions offer the foundation for discussion.