Introduction

The rapid advancement of information and communication technology (ICT), such as the artificial intelligence (AI), Internet of Things (IoT), cloud computing, big data, and sensor technology, has transformed the educational landscape, particularly within university campuses (Krouska et al., 2024). These transformations have revolutionized traditional models of learning, teaching, research, and administration, enhancing the overall quality of campus life and satisfaction. Therefore, universities worldwide are increasingly embracing digitalization to achieve higher standards of excellence (Yang et al., 2024). One of the most prominent manifestations of this shift is the rise of the smart campus concept. Evolving from the digital campus, it is often regarded as a microcosm of the smart cities paradigm (Dana et al., 2022; Polin et al., 2023). According to Min-Allah and Alrashed (2020), a smart campus typically utilizes and integrates smart physical and digital spaces to deliver responsive, intelligent, and improved services, creating a productive, creative, and sustainable campus environment. This exemplifies the innovative integration of education and advanced technologies. As the primary mission of a university is to provide educational services and cultivate talent, many researchers have advocated for a learner-centered design in smart campuses (Baba et al., 2024). By offering a smart learning environment, a smart campus not only helps transform citizens into a smart workforce to boost the knowledge economy but also plays a crucial role in the overarching framework of smart cities (Dong et al., 2020).

Recent research has suggested that smart campuses, through the integration of intelligent technologies, are revolutionizing traditional educational strategies. They create a smart learning environment that offers personalized, interactive, informative, ubiquitous, and effective learning experiences, meeting the diverse educational needs of students (Zhang and Li, 2021). For example, AI-powered educational platforms can personalize learning plans based on historical data and individual student characteristics, thus adapting to each student’s learning style and ability. This helps students acquire knowledge, track progress, and receive real-time feedback. Innovative learning scenarios, such as smart classrooms and smart laboratories, substantially enhance the interactivity and engagement of learning through simulations, gamified learning, and adaptive learning platforms (Li et al., 2024). Smart technologies also eliminate geographical and physical barriers, enhancing interactions and knowledge sharing between teachers and students, promoting educational equity, and expanding access to knowledge. Moreover, the application of big data and cloud computing on campus management optimizes resource allocation and improves the convenience and efficiency of educational services.

Smart campus typically consists of three layers: infrastructure (ICT equipment), technology (intelligent systems), and services (direct offerings to users) (Chagnon-Lessard et al., 2021). Smart devices, serving as a crucial component of the service layer, act as the bridge between users and the diverse services within the smart campus ecosystem. These devices range from personal smart devices (such as smartphones, tablets, and laptops) connecting the smart campus network and accessing intelligent services, such as educational services and knowledge resources, to IoT devices deployed in various campus settings (smart classrooms, smart libraries, etc.) to provide an intelligent physical learning environment. With the extensive use of these devices by users, large-scale data collection has become an important means of improving services (Domínguez-Bolaño et al., 2024). However, the data collected on campus often contain sensitive information, such as academic materials and personal data, raising concerns about information security in the smart campus environment. According to previous research, ensuring information security is not only a vital aspect of constructing smart campuses but also a key factor influencing user trust in the smart campus system (Cheong & Nyaupane, 2022).

Research on educational technology acceptance has long been a prominent topic, emphasizing the importance of effectively integrating technology into educational environments. Current research on factors influencing the use of technology on campus has focused on specific technologies and systems, such as generative artificial intelligence (Saif et al., 2024), IoT technologies (Gökçearslan et al., 2024), mobile learning (Alshurideh et al., 2023), and learning management systems (LMS) (Al-Mamary, 2022). Some researchers have also explored it from the perspective of smart learning environments, including the smart classroom environment (Dai et al., 2024). However, the broader concept of the smart campus involves a comprehensive integration of technologies and systems through smart devices. While previous studies have provided valuable insights into the application of specific technologies, research on students’ perceptions and attitudes toward using smart devices within the larger smart campus context is notably lacking. In addition, scholars focusing on smart campuses have highlighted that current research has often prioritized technological innovations and external form changes while neglecting the actual experiences and perceptions of users, particularly students. As such, the actual requirements of users have not been adequately met, and the effective development and construction of smart campuses has been hindered (Zhang et al., 2022). Consequently, many scholars have emphasized the importance of conducting smart campus research from a human perspective (Zhu et al., 2016; Dong et al., 2020; Zhang et al., 2022). Despite the increasing prevalence of smart devices on university campuses, empirical studies that thoroughly analyze how students perceive and use these devices within the smart campus environment, particularly from the perspectives of technology acceptance and learning remain scarce.

To address this research gap, this study aims to explore the factors that influence university students’ attitudes toward using smart devices (ATT), including both personal and campus-deployed devices, to access services within the smart campus environment. The study integrates technology acceptance model (TAM) with task-technology fit theory (TTF) and perceived risk theory (PRT) to develop a theoretical framework. Specifically, three key variables, namely learning support (LS), perceived interactivity (PI), and information security (IS), are introduced to extend the TAM. Building on the TTF, the study assesses the alignment between the capabilities of smart campus technologies and the learning tasks they are intended to support, emphasizing the importance of a strong fit between technological features and user needs for successful adoption. In addition, based on PRT, the study investigates how students’ perceptions of information security in the smart campus environment influence their ATT. To test this framework, a survey was conducted among students from public universities in Shenzhen, China, and the data were analyzed using partial least squares structural equation modeling (PLS-SEM).

The present study contributes to the literature on educational technology acceptance in several key ways. First, it expands the research scope from isolated educational technologies or narrow smart learning environments to a comprehensive investigation of smart device usage within the broader and integrated smart campus environment. Second, by theoretically depicting and empirically testing the hypothesis, this study enhances the understanding of TTF in the educational context. Specifically, LS and PI are conceptualized as concrete manifestations of task-technology alignment. In addition, this study highlights the critical role of IS in shaping students’ ATT within the smart campus environment. Third, by emphasizing students’ actual experiences and perceptions, the study advances a human-centered research paradigm in smart campus studies. These insights are crucial for optimizing device deployment, enhancing smart campus systems and services, and ultimately improving students’ academic performance, thereby contributing to the sustainable development of higher education.

Literature review and hypothesis development

Smart devices usage in the smart campus environment

In the ongoing digital transformation of the education sector, the IoT is playing an increasingly crucial role in smart campus development. It drives the growth of the “Internet of Education Things”, which involves integrating diverse smart devices into education (Kassab et al., 2020). By integrating advanced technologies across campuses, the IoT breaks down information silos among devices, establishing an efficient service system, seamless connectivity, and intelligent decision-making mechanisms (Asgharinezhad et al., 2024). This comprehensively elevates the quality of campus smart services.

In higher education settings within smart campuses, smart devices are widely utilized to create personalized and smart learning environments for students. Leveraging data collection and intelligent analysis, these devices adapt students’ learning behaviors to meet their individual needs. Refer to Badshah et al. (2024) for a summary of IoT smart devices in education. Specifically, the widespread use of personal smart devices, such as smartphones, tablets, and laptops, enables students to access a broad range of online learning resources (Al-Adwan, 2020). Moreover, these devices are integrated with LMS, facilitating streamlined interactions between teachers and students through course-related activities, such as learning, assignment submission, and assessment (Al-Mamary et al., 2024). In the physical campus environment, smart devices also enhance teaching and learning. Smart whiteboards enrich instruction by displaying multimedia content and promoting interactive learning, thereby increasing engagement (Alturki et al., 2021). Radio frequency identification devices automate attendance tracking, improving efficiency and accuracy. Surveillance cameras support campus safety and enable lecture recordings for post-class review. Smart audio systems improve classroom interaction, whereas virtual reality (VR) and augmented reality (AR) technologies immerse students in experiential learning, helping them better understand complex concepts (Faqih and Jaradat, 2021).

However, the widespread adoption of smart devices represents a double-edged sword. On the one hand, some studies have suggested that the excessive use of personal mobile devices can negatively affect learning outcomes. For example, Yao and Wang (2023) observed that excessive use of personal mobile devices can lead to information overload and technostress, which in turn can negatively impact students’ sleep quality and academic performance. On the other hand, the evolution of mobile learning applications has prompted a revaluation of the role personal mobile devices play in learning. Lin et al. (2021) conducted a large-scale study that revealed the positive, direct impacts of using mobile learning on academic performance and the adverse effects of using entertainment applications. This suggests that, in a smart learning environment, the proper use of smart devices as learning aids can lead to positive learning outcomes, a finding supported by other studies (Lee et al., 2022; Hossain et al., 2020, 2019). Regarding devices in the campus physical environment, research has consistently affirmed their positive influence on learning. For instance, Wang et al. (2022) revealed that students’ individual perceptions of classroom process quality were significantly related to their engagement in the smart classroom environment. Lu et al. (2021) observed that students’ positive perceptions of smart environments improve classmate interaction, motivation, and learning strategies, thereby fostering higher-order thinking. Similarly, Asad et al. (2024) highlighted that IoT-based smart laboratories promote academic performance through interaction, creativity, motivation, and hands-on learning.

Theoretical foundation

The integration of emerging educational technologies poses challenges for educational institutions in terms of technology acceptance and utilization. In recent years, several theoretical frameworks, including TAM, the theory of planned behavior (TPB), and the unified theory of acceptance and use of technology (UTAUT), have been widely used to study technology adoption in education. Among these, TAM is one of the most widely applied and influential models. Two constructs lie at its core: perceived ease of use (PEU) and perceived usefulness (PU) (Davis, 1989). If users believe that technology is easy to use, they will be more willing to continue using it in the future. On the other hand, if users believe that technology is useful for work, their attitude and intention to continue using it will strengthen. Moreover, many researchers have extended TAM by introducing external variables to gain a more context-specific understanding of technology adoption.

Unlike previous research that has focused on the adoption of a single technology, this study emphasizes how the core inherent characteristics of the smart campus ecosystem influence students’ attitudes. This research perspective has guided our theoretical foundation. Compared to TAM, the TPB and UTAUT models explain user behavior through external factors, such as subjective norms, perceived behavior control, social influence, and facilitating conditions. Therefore, these models were not chosen for this study. First, because these models incorporate many external social factors, they may shift the focus of the study toward the organizational level (Hu, 2022). Second, social factors are difficult to assess in advance, as the different functions of smart devices in the smart campus environment are often not fully known in advance (Lagstedt et al., 2020). Moreover, neither the TPB nor the UTAUT model directly addresses the core design attributes of a system. To maintain the simplicity of the model and emphasize the core inherent characteristics of the technology, as well as to keep the focus on a learning-center perspective, this study adopts TAM as the primary theoretical foundation.

Since its introduction in 1989, TAM has been extensively validated across different technologies and user groups, demonstrating its robustness and broad applicability. Particularly in the educational field, TAM has become the dominant model for studying the acceptance of educational technologies (Al-Mamary, 2022). Additionally, the TAM framework is widely used for research on IoT-based devices in education, particularly in guiding studies on smart campus technologies.

To better understand how the core inherent characteristics of smart devices influence students’ attitudes toward using them in the smart campus environment, this study extends TAM by incorporating TTF and PRT. TTF posits that when the characteristics of technology align well with task demands, users’ acceptance and usage intentions will significantly increase (Przegalinska et al., 2025). By analyzing the interaction between tasks (learning) and technology, the actual efficacy of the technology can be measured more effectively. Specifically, this study introduces two variables based on TTF: LS and PI, which are considered as key predictors of PU and PEU. Additionally, drawing on the PRT and considering the extensive data collection in smart campus environment, IS is selected as a direct predictor of PU and ATT. Figure 1 shows the research model of this study.

Fig. 1
figure 1

Theoretical Framework.

Full size image

Learning support (LS)

The concept of LS was first proposed by Sewart (1993), and it was initially referred to as a series of informational, resource-based, personnel, and facility-based services provided by educational institutions to guide, assist, and promote personalized and self-directed learning among students, ultimately aiding learners in achieving their educational goals. Its original intention was to address issues such as decreased student motivation and insufficient engagement in distance learning environments (Tait, 2003).

However, with the rapid development of digital technologies, the scope and context of LS have undergone marked changes. LS is no longer confined to distance education but has been deeply integrated into higher education, becoming a key component in driving educational transformation (Wei et al., 2021). In the context of a smart campus, the integration of digital technologies provides students with more comprehensive and personalized LS. Smart devices, both personal and within smart learning environments, create convenient and diverse learning conditions for students, promoting the development of self-directed and personalized learning capabilities (Zhang et al., 2022) and enabling students not only to acquire knowledge but also to become the creators of knowledge.

To better understand the factors determining students’ attitudes toward the use of smart devices in smart campus environment, we incorporate LS into the research model and combine it with TTF. According to TTF, the characteristics of technology closely parallel task requirements, which aligns well with LS in educational technology contexts. In this context, LS refers to the degree to which the use of smart devices in the smart campus environment fits the requirements of students’ personalized learning and self-directed learning tasks. Specifically, it encompasses the provision of ubiquitous learning environments, as well as customized learning resources and feedback facilitated by smart devices in smart campus environment.

Currently, most research on LS has focused on e-learning, with existing studies showing that LS has a significant positive impact on students’ engagement (He et al., 2019), learning performance (Wongwatkit et al., 2020), and system satisfaction (Zhao et al., 2022). However, research on LS in the context of blended online–offline learning environments within smart campuses is relatively limited. Notably, LS serves as a key manifestation of alignment between task (learning activities) and technology, studies across various educational technologies in higher education, such as e-learning spaces (Wang et al., 2024a), video-based learning platforms (Pal and Patra, 2021), MOOCs (Kim & Song, 2022), and educational robots (Suhail et al., 2024), have demonstrated that task-technology alignment is crucial for technology adoption, as it enhancing both PU and PEU. Specifically, if students believe that smart devices can effectively support their self-direct and customized learning tasks, they are likely to perceive these devices as offering higher value in terms of enhancing learning efficiency and outcomes while reducing the cognitive load and effort required when using these devices for learning, thereby increasing their PU and PEU of the devices. The following hypotheses are thus proposed:

H1: LS has a positive effect on PU

H2: LS has a positive effect on PEU

Perceived interactivity (PI)

In an IoT-empowered smart campus environment, using smart devices to enhance collaboration and interactivity during learning is a prominent feature (Haleem et al., 2022c). Existing education research has subdivided interaction into learner–learner and learner–instructor interactions (Pan et al., 2023). Smart devices create favorable conditions for students to deeply engage in classroom activities, effectively enhancing interaction and communication between students, classmates, and teachers (Cheung et al., 2021). For instance, mobile learning applications on personal smart devices facilitate better communication between students and teachers, substantially improving the frequency and quality of interaction between both parties (Criollo-C et al., 2021). Moreover, multimedia tools in smart classrooms, such as screen sharing, digital whiteboards, and virtual discussion platforms, enhance the learning atmosphere, making classroom learning more engaging and interactive (Bilotta et al., 2021).

Knowledge resources acquisition and feedback are only one part of learning, and interaction and collaboration with instructors and classmates are equally vital components of the learning process (Sun et al., 2022). The interactive features of smart devices provide strong support for the successful completion of learning tasks. Therefore, based on TTF, PI can be defined as the degree of alignment between the technology characteristics and the task requirements. Specifically, in this context, PI refers to students’ perception of how smart devices enhance their interaction with instructors and classmates in the smart campus environment.

In the field of education, numerous studies have already established a link between interactivity and learning performance (Sun and Wu, 2016; Oyarzun et al., 2018). Some empirical research has also confirmed that, from the perspective of learner–learner and learner–instructor interactions, PI has a significant positive impact on PU and PEU of educational technologies (Girish et al., 2022). Similar to research on LS, that on PI has focused on online education, whereas studies on blended learning environments, which combine both online and offline learning, are relatively scarce. Specifically, when students perceive that the high-quality interaction facilitated by smart devices during learning, they tend to believe that these devices can improve learning performance and enhance learning efficiency, thus increasing their evaluation of PU. Meanwhile, the enhanced interactivity allows students more easily solve the problems associated with the use of the devices, reducing the cognitive load when using the smart devices, and thereby improving their PEU. Therefore, the following hypotheses are proposed:

H3: PI positively influences PU

H4: PI positively influences PEU

Information security (IS)

Ongoing security concerns in smart learning environments make it crucial to study how students perceive the risks associated with learning in such environments (Jiang et al., 2022). Risk perception is widely regarded as one of the major barriers to user adoption of innovative technologies. Numerous studies have identified information security as a key factor affecting the use of IoT devices (Jaspers and Pearson, 2022). In smart campus environment, data form the basis for services, and many IoT-based smart campus services rely on the collection and analysis of increasingly specific personal information (e.g., trajectories, learning documents, and biometric data) (Jurcut et al., 2020). The pervasiveness of sensing technologies that handle large amounts of personal sensitive data increases the risk of privacy breaches significantly, thereby hindering the effective application of smart campus technologies (Bates and Friday, 2017). Meanwhile, studies have shown that IoT systems in smart campuses face various attacks and threats, as well as many security vulnerabilities (Zhang et al., 2022). If these data are leaked, not only are users’ privacy severely infringed, but their personal and financial security are also threatened (Gill et al., 2021). In university environments, unauthorized access or hacking could alter or leak sensitive information, such as exam papers, student grades, and academic papers, thereby damaging the institution’s academic integrity, affecting students’ academic careers, and tarnishing the institution’s reputation (Rajab and Eydgahi, 2019). Moreover, information security incidents may also have negative psychological effects on students and faculty, leading to anxiety, fear, and distrust (Aqeel et al., 2022).

The role of IS in the success of smart campuses is further underscored by PRT, which suggests that how individuals perceive potential risks associated with technology markedly influences their attitudes and behaviors toward adoption (Li, 2025). In the context of smart campus environment, IS refers to the degree to which students perceive that the smart campus system can protect their information from unauthorized access. Some scholars have argued that the success of future IoT-based smart campus applications depends on users’ perceptions of privacy, security, and trust (Sneesl et al., 2022a). Previous studies in various contexts have indicated that perceived information risk negatively influences not only users’ attitudes regarding technology use but also PU due to its association with negative consequences and uncertainty (Al-Adwan et al., 2023; Samadzad et al., 2023). Conversely, users who exhibit lower aversion to potential costs and losses are more likely to adopt a given technology; meanwhile, those who perceive lower risks in adoption tend to view the technology as more useful and form more positive attitudes toward it. Based on this, the following hypotheses are proposed:

H5: IS positively affects PU

H6: IS positively affects ATT

Perceived ease of use (PEU)

According to Davis (1989), PEU refers to the degree to which an individual believes that using a particular technology or system requires minimal physical and mental effort. In this study, PEU refers to university students’ subjective perception of the ease with which they can operate smart devices within the smart campus environment for learning-related activities. According to the TAM, PEU is a direct predictor of both PU and attitude toward using technology. Similarly, when students perceive smart devices in the smart campus environment as easy to use, they are more likely to enhance their perception of the effectiveness of the devices in improving learning efficiency and outcomes while also developing a more positive emotional and cognitive response to their use within the smart campus environment. This relationship has been widely validated in recent studies on educational technology adoption (Tan et al., 2023; Ma et al., 2024; Sawiji, 2024). Accordingly, the following hypotheses are proposed:

H7: PEU positively influences PU

H8: PEU positively influences ATT

Perceived usefulness (PU)

PU refers to the degree to which an individual believes that a particular technology can enhance job performance, particularly in terms of increasing efficiency and effectiveness (Davis, 1989). In the original TAM framework, PU is considered as a key antecedent of both attitude and behavior intention to use technology. Numerous studies have shown that PU positively influences the student’ attitudes and intentions regarding the use of educational technologies, including various tools and platforms within smart campuses (Zhang et al., 2023; Wang et al., 2024b). Specifically, when students perceive a technology as capable of improving their efficiency, they are more likely to regard it as useful and demonstrate a more positive attitude and intention to adopt it (Dhingra and Mudgal, 2019). In this study, PU is defined as university students’ perception of the effectiveness of using smart devices in the smart campus environment. As students recognize tangible benefits, such as increased learning efficiency or improved academic results, their cognitive and affective evaluation of using smart devices tends to become increasingly favorable, thereby reinforcing a positive attitude toward using them. Based on this, the following hypothesis is proposed:

H9: PU has a positive effect on ATT

Methodology

Measurement

This paper employed a questionnaire consisting of two sections to test our research model. Section A collected participants’ demographic characteristics, including gender, age, academic stage, and frequency of using smart devices in the smart campus environment for learning activities. Section B contained 21 items measuring six constructs: Learning Support (4 items) adopted from Kim and Song (2022); Perceived Interactivity (3 items) from Camilleri and Camilleri (2022); Information Security (3 items) from Al-Adwan et al. (2023); Perceived Usefulness (4 items) from Davis (1989) and Al-Emran et al. (2020); Perceived Ease of Use (3 items) from Davis (1989) and Lavidas et al. (2020); and Attitude Toward Using Smart Devices (4 items) from Lavidas et al. (2020). All items were measured using a five-point Likert scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). The detailed items of each construct are presented in Table A1 in the Appendix.

To ensure the reliability of the survey, several measures were implemented. First, the questionnaire items were developed by drawing upon existing literature and considering the specific context of the smart campus environment. Second, the English version of the questionnaire was translated into Chinese. Two professors with relevant academic expertise reviewed the translation result to ensure equivalence between the two versions. Third, a group of 20 undergraduate students (10 male and 10 female) with sufficient experience in using smart campus services, along with three professors experienced in the relevant research field, reviewed and revised the questionnaire. Finally, the questionnaire was refined and further improved based on the feedback received.

Data collection

Data were collected using an online questionnaire platform. The target participants for this study were students from public universities in Shenzhen, Guangdong, China, who had prior experience using smart campus services through smart devices. According to the Bureau of Public Works of Shenzhen Municipality, public universities in Shenzhen have more advanced smart campus development than those in other cities in China and worldwide. As such, a purposive sampling method was employed, with specific screening criteria to ensure that all participants met the requirements. The survey was conducted from April to July 2024. During administration, participants were provided with a comprehensive explanation of the study’s objectives and detailed guidelines for completing the questionnaire, thereby enhancing their understanding and ensuring data quality and reliability. This study strictly adhered to the principle of voluntary participation, and rigorous confidentiality measures were implemented throughout the data collection process to safeguard participants’ personal information.

The data collection process involved several key steps. Initially, the online questionnaire was distributed to public universities in Shenzhen, with a prescreening phase to ensure that only students from public universities were included. Subsequent screening steps eliminated participants who were unfamiliar with or had no prior experience using smart devices to access smart campus services for learning activities. Accordingly, a total of 556 responses were received. Invalid responses were filtered out based on the aforementioned criteria and additional screening methods, such as response time and IP address. Regarding missing data, because we distributed the survey online, any incomplete responses were prevented from being submitted. Consequently, there were no missing values in our dataset. Ultimately, 428 valid questionnaires were returned for further analysis, yielding an effective response rate of 77.15%. To further examine the adequacy of the sample size, a power analysis was conducted using G*Power software, which confirmed the sample’s adequacy by considering model structure, expected effect sizes, and significance levels. The calculation indicates that the sample size of our study is appropriate.

Descriptive statistics

As listed in Table 1, the demographics of the participants are as follows: 237 males (55.42%) and 191 females (44.58%). In terms of age, most of the participants ranged from 18 to 25 years. Specifically, 144 individuals (33.64%) were aged 20 years or younger, 235 individuals (54.9%) were between 20 and 25 years old, and 49 individuals (11.45%) were aged 25 or above. Most participants were undergraduate students (76.92%), and the largest proportion majored in engineering (40.09%). Regarding the frequency of using smart devices in the smart campus environment for learning activities, 28 participants (6.54%) reported infrequent use, 151 participants (46.03%) reported use several times a month, 191 participants (35.28%) reported several times a week, and 52 participants (12.15%) reported multiple times a day.

Table 1 Participants’ profile.
Full size table

Data analysis

The PLS-SEM was selected to analyze the respondents’ data as an alternative to the covariance-based structural equation model, providing greater flexibility in data requirements, model complexity, and relationship specification (Hair et al., 2019). The PLS-SEM technique was implemented using SmartPLS 4 software (Version 4.1.0.3) to analyze the data. Specifically, a widely used two-step approach was used to evaluate the proposed theoretical model, as suggested by Leguina (2015). In the first step, the outer model was assessed for convergent and discriminant validity to ensure the adequacy of the measurement model. In the second step, the inner model was evaluated to test the hypotheses, focusing on the structural relationships between constructs.

Result

Common method bias

To address common method bias (CMB), we implemented both procedural remedies before and statistical remedies after data collection. Prior to data collection, participants were assured of their anonymity, and we emphasized the importance of providing honest responses (e.g., the survey introduction included the following statement: “There are no right or wrong answers to any of the questions in this questionnaire”) to mitigate social desirability bias. After data collection, an assessment was performed to identify the potential presence of CMB. Harman’s single-factor test was employed to assess the presence of CMB. The results of the unrotated exploratory factor analysis revealed six factors with eigenvalues greater than one, and the largest factor accounted for 38.273% of the variance, well below the 50% threshold that would indicate significant CMB (Hair et al., 2019), confirming the absence of substantial CMB in the data.

Model fit

To assess the goodness of fit of the model, several key fit indices were examined, including SRMR, d_ULS, d_G, and Chi-Square. The results revealed that SRMR < 0.08, d_ULS < 0.95, and d_G < 0.95 (see Table 2), indicating that the fit of the theoretical model with the collected data is satisfactory and acceptable.

Table 2 Model fit.
Full size table

Reliability and validity of mesurement model

Prior to hypothesis testing, the reliability and validity of the measurement model needed to be examined. Reliability was assessed through Cronbach’s alpha and composite reliability (CR) (Sarstedt et al., 2021). Additionally, Rho_A was used as an alternative indicator to assess internal consistency reliability. As listed in Table 3, the Cronbach’s alpha values range from 0.812 to 0.879, which all exceed the benchmark of 0.7. CR values range from 0.889 to 0.916, which are all higher than the threshold of 0.7. Moreover, the Rho_A values for all constructs range between 0.812 and 0.892, all exceeding 0.7. These results indicate an adequate degree of reliability.

Table 3 Summary result for the reflective outer model.
Full size table

Convergent validity was evaluated by measuring the average variance extracted (AVE) and factor loading for all items associated with specific reflective variables. As listed in Table 3, the AVE values range from 0.713 to 0.763, exceeding the threshold of 0.5 (Sarstedt et al., 2014). Furthermore, all factor loadings are above 0.721, surpassing the benchmark of 0.7, thereby demonstrating satisfactory convergent validity. Discriminant validity was confirmed by applying the Fornell–Larcker criterion and the Heterotrait–Monotrait ratio (HTMT) method, as suggested by Leguina (2015). As shown in Table 4, the bold diagonal values represent the square roots of AVEs. Since the square root of the AVE for each construct is greater than the highest correlation coefficient between that construct and any other construct, the Fornell–Larcker criterion is satisfied. In addition, Table 5 shows that all HTMT values between all dimensions are less than 0.85, confirming that discriminant validity has also been achieved.

Table 4 Discriminant validity (Fornell Larcker criterion).
Full size table
Table 5 Discriminant validity (HTMT criterion).
Full size table

Structural model evaluation

After a satisfactory evaluation of the measurement model, the next step is to assess the structural model. Standard assessment criteria include the coefficient of determination (R2), blindfolding-based cross-validated redundancy measure (Q2), and statistical significance and relevance of the path coefficients. Before evaluating the structural model, collinearity should be measured to ensure that regression results are not adversely affected. Variance inflation factor (VIF) values are used to detect collinearity problems, with values lower than 3 indicating no significant collinearity concerns. The inner VIF values of this study are between 1.235 and 1.653 (see Table 6), indicating that collinearity does not affect the structural model.

Table 6 Inner VIF values.
Full size table

The explanatory and predictive abilities of the model were examined using R2 and Q2. As listed in Table 7, the R2 values of ATT, PU, and PEU are 0.306, 0.400, and 0.161, respectively. Noticeably, the R2 values for ATT and PU are moderate, whereas that for PEU is relatively small, as PEU is a more complex construct that may be influenced by external factors outside the scope of the model. The ATT, PU, and PEU values of Q2 are 0.302, 0.394, and 0.157, respectively, all > 0. Therefore, according to the results of R2 and Q2, the structural model has good explanatory and predictive power. Furthermore, effect size f2 evaluates whether one exogenous construct has a substantive impact on the endogenous construct, with f2 values below 0.02 illustrating a small effect. The f2 value for PEU and PU to ATT are 0.044 and 0.096, respectively (see Table 8), suggesting a moderate effect. The effect size of IS on ATT is 0.017, suggesting its effect on ATT is weak, but it has a strong effect on PU (f2 = 0.133).

Table 7 Predictive accuracy and predictive relevance.
Full size table
Table 8 Effect size assessment.
Full size table

The proposed hypotheses were evaluated using the bootstrapping algorithm in SmartPLS4, involving 5000 bootstrap resampling procedures. The results of the hypotheses test are listed in Table 9 and Fig. 2. As expected, LS significantly positively impacts both PU (β = 0.163, t = 3.67, p < 0.001) and PEU (β = 0.172, t = 3.751, p < 0.001), indicating that LS is a key enabler of PU and PEU, thus supporting hypotheses H1 and H2. Similarly, PI emerges as a critical determinant of PU (β = 0.146, t = 3.293, p < 0.01) and PEU (β = 0.295, t = 6.317, p < 0.001), confirming hypotheses H3 and H4. Moreover, IS exhibits a significant positive effect on PU (β = 0.348, t = 6.915, p < 0.001) and ATT (β = 0.141, t = 2.61, p < 0.01), thereby supporting hypotheses H5 and H6.

Within TAM, PEU has a significant positive effect on PU (β = 0.171, t = 3.772, p < 0.001), confirming hypothesis H7. Among the predictors of ATT, PU exerts the most substantial positive influence (β = 0.320, t = 6.493, p < 0.001), followed by PEU (β = 0.209, t = 4.605, p = 0.001), supporting hypotheses H8 and H9.

Table 9 Hypothesis test.
Full size table
Fig. 2
figure 2

Assessment of the structural model (***p < 0.001, **p < 0.01, *p < 0.05).

Full size image

Total indirect effect

Table 10 presents the total indirect effects of the independent variables on the dependent variables. The total indirect effect of LS on ATT (β = 0.098, t = 4.436, p < 0.001) is significantly positive. This suggests that LS provided by smart devices in a smart campus environment enhances students’ ATT by improving both their PU and PEU. In addition, LS indirectly affects PU through PEU (β = 0.029, t = 2.718, p < 0.01), indicating that PEU is a key pathway through which LS enhances usefulness.

Table 10 Total indirect effect.
Full size table

Interestingly, PI shows a stronger indirect effect on ATT (β = 0.125, t = 5.478, p < 0.01) than LS (β = 0.098, t = 4.436, p < 0.001). PI improves ATT by simultaneously enhancing PU and PEU, highlighting the crucial role of students’ PI with other learners and instructors enabled by smart devices in shaping ATT. Additionally, PI also indirectly contributes to PU by improving PEU (β = 0.05, t = 3.166, p < 0.01).

Furthermore, IS has a significant indirect effect on ATT through PU (β = 0.111, t = 4.669, p < 0.01), suggesting that when students perceive the smart campus system as secure in protecting their information, they are more likely to consider smart devices useful. This perception of usefulness then positively influences their ATT. Finally, PEU indirectly influences ATT via PU (β = 0.055, t = 3.477, p = 0.001), reinforcing that ease of use enhances perceived usefulness, which in turn strengthens students’ overall ATT.

Discussion

Impact of LS on PU and PEU

The results suggest that university students’ perceptions of LS provided by smart devices in the smart campus environment have a significant direct effect on PU and PEU. This finding aligns with those of studies that have examined technology acceptance in education from the perspective of TTF (Suhail et al., 2024; Wang et al., 2024a). Additionally, the results are consistent with research exploring LS from the perspective of service quality. For instance, Kim et al. (2025) observed that learning resource availability significantly affects PU and PEU in smart classroom environments, whereas He et al. (2023) revealed that educational support on E-learning platforms can significantly affect continuous usage intention through PU and PEU. However, past research has focused on specific technologies in education, but relatively few comprehensive explorations of smart learning environments have been conducted. In contrast, this study applies the TTF to deeply examine the overall LS provided by smart devices as service mediaor in the smart campus environment. Accordingly, it broadens the perspective of existing research and further validates and supplements the conclusions of previous related studies.

The goal of educational innovation in smart campuses is learning-oriented (Dong et al., 2020). First, the smart campus environment enables ubiquitous and seamless learning via smart devices (Criollo-C et al., 2021). Students can access intelligent LS in classrooms, libraries, living spaces, and even off-campus, greatly expanding the accessibility and flexibility of learning (Hossain et al., 2022) and facilitating “learning in life.” Second, smart devices integrate information from multiple platforms to construct an open and shared resource system. By leveraging AI algorithms to accurately analyze students’ learning histories, ability levels, and interest preferences, they can recommend personalized learning resources and paths, helping students complete learning tasks more efficiently (G. M. et al., 2024). This is highly consistent with the proposition of the situated learning theory that learning should occur in real-world and task-relevant situations. By providing intelligent support tailored to students’ learning tasks, the smart campus creates an environment that is both highly realistic and closely related to learning tasks, enabling students to deeply internalize and apply knowledge. In this context, the LS of smart devices meets students’ diverse needs during learning, significantly improving their learning efficiency and, thus, enhancing their perception of the devices’ usefulness. Meanwhile, extensive LS can reduce students’ cognitive load when using smart devices, thereby strengthening their perception of the ease of use of the device.

Effect of PI on PU and PEU

Similar to LS, PI has a positive, significant impact on PU and PEU. This result is highly consistent with previous studies in the field of E-learning (Croxton, 2014; Girish et al., 2022). However, relatively few studies have been conducted on the impact of PI on PU and PEU in an offline learning environment. In contrast, this study utilizes TTF to verify the applicability of this conclusion in a complex smart campus environment that integrates online and offline learning, further expanding the theoretical framework of the impact of PI on the learning experience.

A notable feature of smart campus environment is their ability to enhance teaching interactions, enabling students to interact more effectively with classmates and instructors (Haleem et al., 2022a). This enhanced interaction provides a more efficient bridge for communication and collaboration between students, teachers, and classmates. The integration of smart devices into smart campuses has successfully transformed the traditional learning environment into a digital learning space. With the diverse applications of smart whiteboards, mobile LMS, collaborative tools, and campus social media (Lee et al., 2019; Ma et al., 2024), and relying on the powerful support of network platforms, smart devices have made cooperation and idea sharing between teachers and students more seamless. This transformation not only improves classroom interactions but also promotes collaborative learning. As Burns et al. (2013) demonstrated, collaborative learning enhances student engagement, boosts their confidence, and improves learner–instructor relationships. In the digital learning environment created by smart campuses, students are no longer merely passive recipients of knowledge. Instead, they actively construct their own knowledge systems through diverse interactive means. In the new educational model, students even have the opportunity to teach others, and this role transformation redefines the functions and roles of teachers (Dong et al., 2020). The interactions with classmates and instructors provide students with effective solutions when they encounter learning difficulties. Smart devices in the smart campus environment significantly enhance students’ interaction abilities with their classmates and teachers, offering students timely help, especially when solving learning problems. From the perspective of knowledge construction, this process not only provides students with more practical opportunities, helping them to consolidate and apply knowledge while solving problems, but also allows students to restructure and expand their existing knowledge, thus deepening their understanding of knowledge and skills. These interactions enhance students’ perception of the effectiveness of smart devices in learning (i.e., PU) and improve students’ PEU of the devices by reducing operational barriers.

Influence of IS on PU and ATT

Our results demonstrate that IS can significantly influence PU and ATT from the perspective of PRT. This finding is consistent with those from previous studies on technologies in smart education environments, such as E-learning (Jiang et al., 2022; Madni et al., 2022), smart learning systems (Al-Adwan et al., 2023; Jo and Park, 2024), AI (Zhang et al., 2025), cloud services (Arpaci et al., 2015), and biometric recognition technology (Rukhiran et al., 2023). As Sneesl et al. (2022b) highlighted, privacy concerns are key factors influencing the adoption of smart campus technologies, and this study further validates this idea through a larger perspective of smart devices in the smart campus environment. IS is a core element in promoting university students’ acceptance of smart devices, and it has a significant positive impact on students’ PU and ATT.

The implementation of smart campuses relies heavily on the collection and analysis of extensive data, and realizing system intelligence largely depends on the availability of high-dimensional data, particularly personal data reflecting subjective opinions (Dong et al., 2020). A primary concern regarding information security is the risk of data breaches. Once sensitive data are compromised, the privacy of students and faculty members could be severely impacted, potentially facilitating illegal activities. Consequently, students perceive data breaches as a significant source of privacy and security risks. When students regard these risks as excessive, they tend to develop negative attitudes toward the smart campus initiative. Therefore, safeguarding the data of students and faculty in smart campuses, especially on platforms utilizing emerging technologies, such as cloud computing and IoT, presents a substantial challenge (Bates and Friday, 2017).

Roles of PU and PEU in enhancing ATT

In this study, PEU was found to have a significant positive impact on PU. This result is consistent with the propositions of TAM and those of similar studies on smart education environments (Kim et al., 2025). This suggests that when students perceive smart devices in the smart campus environment as easy to use, they are more likely to consider these devices as beneficial to their learning, as they are likely to believe that these technologies can enhance their learning efficiency and outcomes. In other words, the more intuitive and accessible the devices are, the easier it becomes for students to recognize their learning value, thus enhancing perceived usefulness.

Further analysis revealed that both PEU and PU significantly influence students’ attitudes toward using smart devices. Specifically, the impact of PEU on ATT indicates that when students find smart campus technologies user-friendly, their attitudes become more favorable. This finding contrasts with those of some studies on the metaverse (Al-Adwan et al., 2023) and VR (Wang et al., 2024c) in educational contexts, which argue that university students already familiar with digital tools may not regard PEU as a significant determinant of attitude or behavioral intention. This may be due to the complexity of smart campuses, which integrate multiple technologies and systems, imposing cognitive loads on university students. As such, ease of use may shape the attitudes of even technologically adept students, highlighting PEU’s continued relevance in the smart campus environment.

Meanwhile, PU exerts a stronger effect on ATT than PEU, which is consistent with TAM, as proposed by Davis (1989), as well as findings from prior studies on emerging technologies in higher education (Lefrid et al., 2023; Sungur Gül and Ateş, 2023; Watanabe et al., 2023; Zhang et al., 2023; Wang et al., 2024b). This result underscores that when students perceive smart devices as highly useful, they tend to maintain a positive attitude toward their usage, even if the devices require additional effort to learn how to use. Therefore, PU has emerged as a more influential predictor of ATT, suggesting that students’ attitudes are predominantly driven by their cognitive evaluations of usefulness.

Mediating roles of PU and PEU

Beyond the direct effects, the mediating roles of PU and PEU also offer valuable insights. First, LS indirectly enhances students’ attitudes via both PU and PEU, indicating that LS influences students’ attitudes toward using smart devices in smart campuses by enhancing PU and PEU. This mechanism highlights the importance of LS in cultivating favorable attitudes toward technology use in the smart campus environment. Additionally, LS influences PU through PEU, suggesting that PEU is one of the mechanisms through which LS impacts perceived usefulness.

Notably, the total indirect effect of PI on students’ ATT is even stronger than that of LS. By simultaneously enhancing PU and PEU, PI exerts a more substantial influence on ATT. Presumably, in the smart campus environment, the social interaction and collaborative experiences introduced by smart devices are more likely to be directly experienced by students, outweighing their support for personalized and self-directed learning. Thus, these aspects are central in shaping students’ attitudes toward smart devices. This finding emphasizes the significance of enhancing interactive capabilities in blended learning environments, including smart campuses. Additionally, PI indirectly enhances PU through its positive impact on PEU, strengthening its multi-faceted contributions to the learning experience.

IS also demonstrates a significant indirect effect on ATT through PU. Specifically, IS has a strong positive effect on PU, which in turn positively influences ATT. This finding further suggests that students’ trust in the security of the smart campus system is a critical precondition for developing favorable attitudes toward smart device use. In an era of growing concerns over data privacy, information security becomes a key factor in shaping students’ willingness to adopt educational technologies.

Finally, the mediating role of PU in the relationship between PEU and ATT was also confirmed. Although PEU has a direct effect on ATT, its indirect effect via PU remains significant, thereby supporting the foundational structure of the TAM. This underscores PU as a pivotal mediator in students’ decision-making processes regarding the adoption of educational technologies.

Implications

Theoretical implications

This study has several theoretical implications. First, while previous research on educational technology acceptance has often focused on specific technologies or narrow smart learning environments, this study represents one of the few attempts to adopt a broader perspective by examining determinants of university students’ attitudes toward using smart devices within the smart campus environment. This approach provides a more comprehensive understanding of technology acceptance mechanisms in complex, technology-integrated smart learning environments.

Second, this study proposes an theoretical framework by extending TAM through the integration of TTF and PRT, and incorporating LS, PI, and IS as key external variables. This study focuses on the influence of LS and PI of smart devices in the smart campus environment from the perspective of TTF. Specifically, this study regards LS and PI as specific manifestations of task-technology alignment, deepening the application of TTF in the educational field. The results reveal that both LS and PI are key determinants of university students’ attitudes toward using smart devices in the smart campus environment. This verifies the crucial significance of the synergistic effect between task characteristics and technology features in attitude formation. Based on PRT, and in response to concerns about information security in the IoT environment of smart campuses, this study incorporates IS into the model. This inclusion highlights the necessity of addressing security concerns in the development of smart campuses and provides an insight for research on trust mechanisms in smart campuses.

Finally, by advocating for a contextualized and human-centered approach to smart campus development, this study contributes to advancing the theoretical development of smart campus research and offers valuable guidance for model construction and variable selection in future studies.

Practical implications

The results and discussion of this study have practical implications for university students, educators, university administrators, and smart campus developers to promote the effective application and widespread acceptance of smart devices in the smart campus environment.

From the perspective of university students, the significant influence of PU and PEU on ATT suggests that students should actively explore and utilize the functional capabilities of smart learning systems, such as personalized recommendations and intuitive interface designs, to better meet individual learning needs and enhance their motivation for technology use. By acquiring skills in operating smart devices and understanding their advanced features, students can improve both self-directed and collaborative learning outcomes. Moreover, given the importance of IS in the smart campus environment, students must raise their awareness of cybersecurity and acquire basic knowledge of personal data protection, equipping themselves to identify and respond to potential security threats.

From the perspective of educators, it is imperative to embrace the integration of smart devices into pedagogical practice and fully leverage their potential to enhance teaching effectiveness. By designing interactive and intellectually stimulating learning activities, such as through gamification or flipped classrooms (Haleem et al., 2022b), educators can facilitate deeper students’ engagement, promote collaborative learning, and foster meaningful instructor–student interaction. Similarly, instructors should remain attentive to students’ challenges in using educational technologies, offering timely guidance and support to maximize learning outcomes. In addition, educators bear the responsibility of cultivating students’ information security awareness, providing instruction on the safe and ethical use of digital tools, and improving their own cybersecurity literacy to proactively identify and mitigate potential risks in the smart campus environment.

From the perspective of university administrators and smart campus developers, this study highlights the need to optimize both PU and PEU through user-centered design strategies. System interfaces should be simplified, operational processes streamlined, and application functions stabilized to ensure a seamless user experience. Given that the largest impact of PI on ATT is through both PU and PEU, developers are encouraged to enhance interactive functionalities within smart devices and software platforms. Incorporating immersive technologies, such as VR and AR, can further elevate student engagement and foster more authentic learning experiences. Moreover, by customizing educational processes to meet university students’ learning needs and providing high-quality intelligent learning tools and resources, smart campuses can increase the likelihood of students using smart devices to support their learning. This enhanced perception of usefulness, coupled with improved learning outcomes, will strengthen students’ overall attitude toward and acceptance of smart devices in the smart campus environment. Furthermore, information security remains a core pillar of smart campus development. Developers should strengthen cybersecurity infrastructure, increase investments in risk prevention mechanisms, and implement real-time monitoring and rapid response systems to detect and address vulnerabilities promptly (Cheng, 2023; Salamzadeh et al., 2022; Cormack, 2019). Drawing upon prior research, the formulation and refinement of comprehensive cybersecurity policies and protocols are essential for maintaining a safe and trustworthy smart campus ecosystem (Dong et al., 2020).

Conclusions

This study identifies and evaluates the determinants of students’ attitudes toward using smart devices within the broader, integrated smart campus environment. Utilizing the extended TAM, the study organically integrates TTF and PRT, incorporating three key external variables, LS, PI, and IS to develop a human-centered framework for understanding technology acceptance within smart campuses. Based on 428 valid survey responses from university students in Shenzhen, Guangdong, China, the study employed PLS-SEM to establish the model. The findings indicate that PU, PEU, and IS significantly influence students’ ATT in the smart campus environment, with PEU also having a significant effect on PU. Furthermore, LS and PI exert significant indirect effect on students’ ATT by enhancing PU and PEU, with PI demonstrating a stronger impact than LS in shaping students’ ATT.

Overall, this study provides empirical support for understanding students’ adoption of smart devices within the smart campus environment. It extends and enriches the existing educational technology acceptance literature and offers targeted practical implications for higher education institutions’ deployment of smart campuses. By emphasizing students’ actual experiences and perceptions, this study aims to advance a human-centered research paradigm for smart campus. This approach contributes to enhancing the educational value of smart campus initiatives and promotes the sustainable development of higher education.

Limitations and future research directions

Although this study provides insights into university students’ attitudes toward the use of smart devices in smart campuses, several limitations should be noted. First, the research sample is limited to students from universities in Shenzhen, Guangdong Province, China, potentially limiting the generalizability of the findings. Second, the study employs a cross-sectional design, reflecting attitudes and behaviors at a specific point in time, which makes it difficult to capture their dynamic changes over time. Furthermore, the study focuses on the learning perspective, incorporating only three variables: LS, PI, and IS, without fully addressing the multidimensional factors that influence attitudes. Finally, in measuring the PI variable, the study focuses solely on interpersonal interactions, while the actual interactions also include interactions with content and human-computer interactions, which may have significantly different impacts on attitudes. In light of these limitations, future research should broaden the sample to include diverse cultural backgrounds and different types of universities to enhance the external validity of the model. Longitudinal research designs could also be employed to track the evolving attitudes and behaviors of students throughout the development of smart campuses. Additionally, perspectives from other stakeholders, such as teachers and parents, could be incorporated to obtain more comprehensive insights. Lastly, future studies should adopt a mixed-methods approach, combining interviews, focus groups, and field observations to gain a deeper and more nuanced understanding of different stakeholder experiences within the smart campus environment.