“Teaching engagement” is derived from ‘work engagement’. Kahn (1990) defines work engagement as “the control of an organization member’s self over his or her work role” and the use and expression of energy, cognition, and emotion in the workplace. Schaufeli et al. (2002) defined ‘work engagement’ as “a positive, work-related emotional and cognitive state, included three core dimensions: vigor, dedication, and absorption.” He also noted that the three core components correlate to Kahn’s proposed energy, emotion, and cognition. This view was widely recognized (Salanova and Schaufeli, 2009).

This study defines online teaching engagement as “the sum of energy, emotion, and cognition that teachers invest in online teaching” based on the definition of Schaufeli et al. It includes the three characteristics of vigor, dedication and concentration. Studies have also demonstrated that work engagement is associated with positive outcomes at individual and organizational levels, such as higher work performance, higher health level, and higher organizational commitment goals (Innstrand et al., 2012; Bakker and Demerouti, 2008).

Terms like “school support,” “support for innovation” and “institutional support” are frequently used when analyzing the integration of technology into teaching. This study employs the term “institutional support.” The institutional support examined in this study is based on Moreira-Fontán et al.’s (2019) definition as teachers’ perceived school support for innovation and the use of information technology. In the context of online teaching, when information technology is fully used, this concept has two dimensions. First, support for innovation and creativity, as well as the use of innovative teaching methods. Second, support for the use of technology in online teaching.

Institutional support is a type of job resource. Job Demands-Resources Theory hypothesizes that job resources reduce the effects of job demands on stress, including job anxiety (Bakker and Demerouti, 2017). Job resources are also a relevant predictor of work engagement. Some studies have provided additional evidence for this interaction. Guglielmi et al. (2016) found a positive correlation between teachers’ job resources and work engagement. There is a motivational connection between the employee and the workplace that triggers a higher level of work engagement (Sortheix et al., 2013). A study of home care professionals found that some job resources can slow down job demands and burnout (Xanthopoulou et al., 2007). Furthermore, institutional support was found to be a negative predictor of teachers’ ICT stress but a positive predictor of teachers’ positive views, attitudes toward ICT (Joo et al., 2016), and self-efficacy in teaching using technology (Admiraal et al., 2017). School support for innovation provides teachers with psychological comfort. The higher the teachers’ satisfaction with the support, the more positive their attitudes toward teaching (Cross and Hong, 2012). Therefore, this study proposes the hypothesis 1.

Teachers’ emotions play a major role in their lives. In addition to being crucial for teachers’ mental health, it is strongly correlated with teaching behaviors, teacher-student relationships, and student outcomes in the teaching and learning process (Schutz and Zembylas, 2009). According to Farouk (2012), teachers’ emotions are essential to their interactions with students, colleagues, and parents rather than being “inner feelings” that stay fixed within the body. It is defined as a person’s mental activity, comprehension of other people’s emotions, capacity for emotion regulation, and reaction to emotional activity. Online teaching is a process of interaction with others. Due to various uncertainties in the teaching environment, teachers may experience both positive and negative emotions during this process. According to this study, teachers’ emotions during online teaching are defined as temporary, dynamic, and interactive emotional states brought on by a variety of uncertainties in online teaching. Additionally, it refers to the psychological experience of attitudes toward individuals, things, and objects that are associated with online teaching, including both positive and negative emotions. Negative emotions related to online teaching can be caused by a variety of factors, such as teachers’ lack of expertise in it, the digital environment, and their low level of digital literacy. Happiness, anger, and anxiety are the three emotions most closely associated with teaching; happiness is the most positive emotion, anger is the most negative emotion, and anxiety is the most interesting to researchers (Frenzel et al., 2016). Teachers experience anger at least once in about 15–20% of their classrooms. Anxiety is also one of the most concerned emotions related to the use of technology, which can prevent teachers from using technology (Joo et al., 2016).

Emotion plays a crucial role in all teaching activities of technology applications (Azzaro and Martínez Agudo, 2018). Wang (2014) emphasized the importance of emotions and emotional awareness in teaching and learning when using technology. Massive online teaching is the most powerful sign of technology-enabled teaching and learning. In this context, teacher emotions would have an impact on teachers’ confidence and intention to use technology, as well as the teaching and learning process. Technology-related anxiety or fear among teachers is caused by the unconsidered use of technology in the teaching and learning process, which hinders learning (Oluwalola, 2015). Ultimately, this has a negative impact on the effective application of technology. Perceived ease of use of technology is positively connected with pre-service teachers’ enjoyment (Teo and Noyes, 2011). In addition, pre-service teachers’ anxiety was negatively correlated with their technology knowledge (Kay, 2008).

However, negative emotions have been neglected in studies of teacher work engagement. Since work engagement has been defined as a positive work-related affective-motivational state (Schaufeli et al., 2002), it is important to explore the relationship between teachers’ negative emotions and work engagement in online teaching. Schaufeli et al. (2001) concluded that employees with high work engagement have higher positive emotions and lower negative emotions. The study of novice teachers’ anxiety found that there is a negative relationship between teachers’ work engagement and anxiety (Zhang, 2010). Therefore, hypothesis 2 was proposed in this study.

Digital self-efficacy (DSE) in online teaching contexts in this study is defined as teachers’ domain-specific confidence in their ability to effectively integrate digital technologies into pedagogical practices, encompassing both technological operation and instructional design capacities. This concept originates from Bandura’s (1977) self-efficacy theory but is contextualized to digital education, emphasizing dual competencies in technical operation and instructional strategies. And according to Moreira-Fontán et al. (2019), there are two components of digital self-efficacy in teaching. One is digital technology competency, which includes technological knowledge and the capacity to use digital technology resources (e.g., online platforms, multimedia resources, data analytics software). The other is pedagogical competence, which can also be considered content knowledge or pedagogical knowledge. Encompasses the capacity to translate technology into pedagogical resources, such as designing interactive online activities, leveraging data-driven feedback to optimize instruction, and sustaining student cognitive engagement in digital environments. This construct diverges from general self-efficacy in three critical dimensions: (a) task specificity (focused on technology-mediated teaching scenarios rather than general life challenges) 3; (b) competency duality (requiring simultaneous mastery of technological tools and pedagogical principles) (Ulfert-Blank and Schmidt, 2022); and (c) dynamic responsiveness (adapting to rapidly evolving digital ecosystems) (Hwang et al., 2023).

In addition, the concepts of technological competence and digital self-efficacy are also different. Although technological competence focuses on objective skill acquisition (e.g., coding, software operation), digital self-efficacy incorporates individual beliefs. For instance, two teachers with equivalent technical skills may exhibit divergent innovation intentions in online teaching due to differences in self-efficacy (Sehar and Alwi, 2023). Empirical evidence suggests that technological competence forms the foundation of digital self-efficacy, but the latter also requires synergistic support from teaching experience and institutional support (e.g., training, peer collaboration) (Hatlevik and Hatlevik, 2018).

With the development and use of technology, there are many studies on computer self-efficacy (Hatlevik et al., 2018; Wartella and Jennings, 2000) and Internet self-efficacy (Chang et al., 2014; Joo et al., 2000). Digital self-efficacy is associated with learning and developing new skills and also influences learners’ willingness to engage in digital systems (Ulfert-Blank and Schmidt, 2022).

Based on the JDR theory, self-efficacy is considered one of the main personal resources that predicts work engagement. A study showed that a higher level of personal resources, including self-efficacy, correlates with higher levels of work engagement (Xanthopoulou et al., 2009). People with low self-efficacy would experience more negative emotions (Schwarzer and Hallum, 2008). Research suggests that self-efficacy for teaching correlates positively with positive emotions and negatively with negative emotions (Taxer and Frenzel, 2015). Additionally, Moreira-Fontán et al. (2019) showed that teachers’ ICT-related positive emotions mediate the effects of digital self-efficacy on work engagement and that teachers’ digital self-efficacy positively predicts teachers’ ICT-related positive emotions. Therefore, the following hypothesis 3 was proposed in this study.

Drawing from previous research, this study aims to analyze the structural relationships among four variables: work engagement, institutional support, negative emotions, and digital self-efficacy. As shown in Figure 1, six hypotheses of direct relationship and three hypotheses of mediated relationships were proposed.

In total, 246 teachers were contacted. They were from different primary and secondary schools in China. All the participants had experience with online teaching. After the removal of incomplete and invalid questionnaire responses. There were 232 valid responses. The specific statistics are shown in Table 1.

The questionnaire included two parts. The first part was socio-demographic variables, such as gender, age, teaching year, and educational level. The second part was target variables. And the items were presented.

The data analysis was conducted with a three-step methodology. In the first step, the reliability and validity of the measurement model were tested by Cronbach’s alpha (Cronbach’s α) and confirmatory factor analysis (CFA) using SPSS19.0 software. In the second step, a t-test and analysis of variance (ANOVA) were conducted to investigate whether gender and teaching age were related to work engagement, institutional support, teachers’ negative emotions and digital self-efficacy. In the third step, the research model (Figure 2) was analyzed using structure equation modeling (SEM), supported by AMOS 26.0. The paths between the four variables were modified.

Cronbach’s α was calculated to validate the internal consistency reliability. The value of Cronbach’s α ranged from 0.813 to.899. Additionally, the validity was tested through confirmatory factor analysis (CFA). The meanings of all the items were clearly described and distinguished from one another. All the items included in the four common variables were consistent with the classification.

The potential relationships were tested by SEM. AMOS 26.0 software was used to test the hypothesized model. The measurement model is shown in Figure 2. The measurement model tests of goodness of fit were a prerequisite condition for valid interpretations of the variables in structural relationships (Kline, 2005). As shown in Table 2, the measurement model needed to be modified to satisfy goodness of fit. Therefore, the measurement model was modified twice based on data and hypotheses.

The indices of the final model revealed that the model fit the data well. The fit indexes of the revised structural equation model are shown in Table 3, and the final structural model in Figure 3. The results showed that institutional support significantly predicted teachers’ digital self-efficacy (β = 0.47, p < 0.01) and negative emotions (β = −0.82, p < 0.01). Digital self-efficacy significantly predicted negative emotions (β = 0.15, p < 0.01). Teachers’ negative emotions significantly predicted work engagement (β = −0.90, p < 0.01). Therefore, H1b, H1c, H2a, and H3b are valid and H1a and H3a are not valid.

The study conducted a mediation effects test through Bootstrap 5,000 times in AMOS 26.0. As shown in Table 4 and Figure 4, this study supports two chain mediation effects.

Path 1: Institutional support → Negative emotions → Work engagement.

BootstrapCI 95% of the mediating effect of online teaching negative emotions did not contain 0 [0.726, 1.132]. And the effect value was 0.913 accounting for 91.94% of the total effects.

Path 2: Institutional support → digital self-efficacy → negative emotions → work engagement.

BootstrapCI 95% of the chain mediation effects of digital self-efficacy and negative emotions did not contain 0 [0.011, 0.166], with an effect value of 0.080, or 8.06% of the total effects.

Therefore, H2b and H3d were valid. And H3c was not valid. In addition, the Bootstrap CI 95% for the comparison of the two paths was [0.810, 1.198], respectively, neither of which contained 0. It indicated a significant difference in the chain mediation effects of the two paths.

An independent samples t-test was conducted to investigate whether gender was related to institutional support, digital self-efficacy, negative emotions and work engagement. The results showed that there was a significant difference between genders in institutional support (p = 0.002 < 0.05) and work engagement (p = 0.011 < 0.05). Specifically, female teachers perceived more institutional support for online teaching than male teachers. And female teachers had higher work engagement in online teaching than male teachers.

ANOVA was conducted to investigate whether age was related to the four dimensions. Teachers between the ages of 20–29 and 30–39 had significant differences in negative emotions (p = 0.002 < 0.05), institutional support (p < 0.001) and work engagement (p = 0.001 < 0.05). Specifically, 20–29 year old teachers experienced less negative emotions than 30–39 year old teachers, but better than 30–39 year old teachers in perceiving of institutional support, and work engagement. In addition, there was a significant difference between teachers aged 20–29 and 40–49 in terms of negative emotions (p = 0.026 < 0.05), digital self-efficacy (p = 0.010 < 0.05), and institutional support (p < 0.001). Specifically, teachers aged 20–29 experienced fewer negative emotions than teachers aged 40–49, and were better in digital self-efficacy, and perceiving institutional support than 40–49 teachers. There was a significant difference between teachers aged 30–39 and 50 + in negative emotions (p = 0.003 < 0.05), and institutional support (p < 0.001). Specifically, teachers aged 50 experienced less negative emotions than teachers aged 30–39, and better than teachers aged 30–39 in perceiving of institutional support. There was a significant difference between teachers aged 40–49 and 50 + in negative online emotions (p = 0.003 < 0.05), institutional support (p < 0.001), and work engagement (p = 0.005 < 0.05). Particularly, teachers aged 50 + experienced less negative emotions than teachers aged 40–49, and were better in perceiving institutional support, and online teaching engagement than teachers aged 40–49 teachers. In addition, there was no significant difference between 20 and 29 and 50+, 30–39, and 40–49-year-old teachers in the four dimensions.

ANOVA was conducted to investigate whether teaching years were related to the four dimensions. The results showed that there was a significant difference between teachers with less than 5 years of teaching experience and those with 6–10 years of institutional support (p = 0.007 < 0.05). Specifically, teachers with less than 5 teaching years were better than teachers with 6–10 teaching years in perceiving of institutional support. And there was a significant difference between teachers with less than 5 years and teachers with more than 10 teaching years in negative emotions (p = 0.013 < 0.05), digital self-efficacy (p = 0.003 < 0.05), institutional support (p < 0.001), and work engagement (p = 0.001 < 0.05). In particular, teachers with less than 5 teaching years had less negative emotions than those with more than 10 teaching years, and were higher in digital self-efficacy, perceiving innovation support and work engagement than teachers with more than 10 teaching years. In addition, there were no significant differences between teachers with more than 5 teaching years and teachers with more than 10 teaching years on the four dimensions.

ANOVA was conducted to investigate whether education level had a significant influence on the four dimensions. In terms of teachers’ education level, all the participants had an associate degree or above. There was a significant difference between teachers with undergraduate degrees and those with graduate degrees in negative emotions (p = 0.002 < 0.05), digital self-efficacy (p = 0.035 < 0.05), institution support (p = 0.002 < 0.05), and work engagement (p = 0.002 < 0.05). Specifically, teachers with an undergraduate degree experienced less negative emotion in online teaching than teachers with a graduate degree, and also higher in digital self-efficacy, perceiving innovation support and work engagement than teachers with a graduate degree. Additionally, there was no significant difference in the four dimensions between teachers with associate degrees and teachers with undergraduate degree or graduate degree.

The study found that online teaching negative emotions mediated the relationship between institutional support and work engagement in online teaching. Institutional support supports teachers not only with substantive technical support, but also with psychological support. It was in order to help teachers reduce negative emotions associated with online teaching by relieving the stress of technology use. The finding was consistent with previous researches (Joo et al., 2016; Vongkulluksn et al., 2018). Joo et al. (2016) emphasized institutional support’s dual role in providing technical resources (e.g., reliable platforms, IT assistance) and psychological reassurance (e.g., peer mentoring, stress management workshops). For instance, technology-specific anxiety is mitigated when schools offer real-time technical hotlines, while burnout is alleviated through workload redistribution policies (Vongkulluksn et al., 2018). Unlike anxiety, which often arises from transient technical challenges, anger is more deeply rooted in systemic inequities or perceived institutional neglect. For example, teachers reported anger when forced to adopt poorly tested technologies without consultation, echoing Ding (2025) observation that anger emerges when autonomy is undermined. In addition, teachers can conduct online teaching with a positive attitude through the reduction of negative emotions of online teaching, resulting in increased work engagement in online teaching. Moreira-Fontán et al. (2019) demonstrated that ICT positive emotions mediate the relationship between innovation support of ICT and work engagement, which was consistent with the findings of this study.

The differences between the two mediating effect paths in this study were found to be significant. And the mediating effect of negative emotions is much higher than the chain mediating effect of digital self-efficacy and negative emotions. It suggests that negative emotions were more relevant in online teaching. A study conducted by Azzaro and Martínez Agudo (2018) showed that emotions play an important role in teaching environment where technology is used. Therefore, the impact of emotions in online teaching environments needs to be paid more attention.

The three strategies are proposed based on the findings. First, school administrators have a decisive influence on teachers’ intentions to use technology for learning and teaching (Instefjord and Munthe, 2017; Teo and Noyes, 2011). And they should develop mechanisms to motivate teachers in order to be more positive in online teaching. Second, schools should provide more opportunities for teachers to communicate, as effective communication can also contribute to a positive working environment which helps teachers to work with greater ease and pleasure. For example, establish transparent communication channels for teacher feedback during technology rollout and ensure equitable workload distribution (Giray, 2025). Third, deploying AI-Driven Emotional Support Tools (AI Chatbots et al.) trained on cognitive-behavioral therapy (CBT) principles to reduce teacher workloads. Furthermore, other negative emotions such as frustration and burnout can also affect teachers’ engagement in online teaching (Aldahdouh et al., 2024). Strategies to reduce teachers’ frustration and burnout will be examined in future research.

According to the JD-R theory, job resources and personal resources may support each other and reduce stress caused by work demands. Institutional support is a job resource. Institutional support is a type of job resource. Digital self-efficacy is a type of personal resource. Emotions, such as job-related anxiety, are also sources of stress. The JD-R confirms the negative effects of digital self-efficacy and institutional support on negative emotions in online teaching.

Online teaching is a new form of teaching different from traditional classroom teaching. Online teaching is a novel way of teaching that differs from traditional classroom teaching. To adapt to the new job requirements, teachers must make adjustments to their instructional resources, methods, and other aspects. Teachers would have negative emotions when switching from their previous teaching mode. Institutional support and teachers’ digital self-efficacy can help reduce negative emotions and improve work engagement. It is consistent with the findings of Xanthopoulou et al. (2009), who discovered that job resources predict personal resources (self-efficacy) and work engagement.

The four evidence-based strategies were proposed. First, institution could establish regional hubs for sharing digital teaching resources and support strategies, mirroring Japan’s GIGA School Program (MEXT, 2022), which provides uniform devices and cloud-based lesson templates, minimizing variability in institutional support (Tanaka and Yamada, 2023). Second, school administrators are required to develop a teacher professional development plan to enable the successful integration of information technology and education that includes an online instructional plan (Vongkulluksn et al., 2018). In Romania, the central administration of education contributes to the development of teachers’ competences through their participation in continuous training programs (Tripon, 2022). And certification systems should be advocated that standardize digital teaching skills while integrating emotional resilience training. In addition, certification should be linked to career advancement. Third, Teachers must also conduct a learning session on an online course instructional approach and extensively complete the instructional design of the online courses. In fact, teachers have a high demand for training in online teaching knowledge and competency. Song et al. (2020) proposed that more than 70% of teachers need to be trained in the use of platforms and access to digital resources. Fourth, establish observation groups where exemplary teachers share recorded lessons and problem-solve collaboratively. According to Song et al., another strategy to improve digital self-efficacy is to observe other teachers who use information technology well in the classroom. In practice, teachers who are more effective in online teaching should be encouraged to record their online teaching for other teachers to reference and learn from. Each group should include excellent teachers to share their successful online teaching experience. The members share their learning experiences and discuss the problems related to online teaching. It helps to improve their work engagement and online teaching effectiveness.

In this study, teacher work engagement was investigated as an outcome variable. But it can in turn actively moderate the relationship between digital self-efficacy and negative emotions. The influence is reflected in two aspects.

First, high engagement should be buffer low digital self-efficacy. Teachers with high work engagement exhibited lower levels of negative emotions even when their digital self-efficacy was moderate or low. It aligns with the buffering hypothesis within the JD-R framework (Bakker and Demerouti, 2017), where engagement—a personal resource—mitigates the detrimental effects of limited self-efficacy. Specifically, teachers with high engagement reframed technological challenges as opportunities for pedagogical innovation, thereby reducing frustration and anxiety (Schaufeli, 2015). The adaptive response pattern corresponds with the transactional stress model (Lazarus and Folkman, 1984), suggesting that engagement facilitates the development of constructive coping strategies.

Second, low engagement exacerbated negative emotions triggered by low self-efficacy. Teachers with reduced engagement levels were more likely to attribute technical failures to personal inadequacy (e.g., “I lack inherent technical competence”) rather than situational constraints, intensifying emotions of anger and burnout (Bandura, 1997). Crucially, sustained engagement appears to foster self-efficacy through two complementary pathways: (1) mastery experiences derived from overcoming technical challenges (e.g., resolving live-streaming disruptions), and (2) vicarious learning through peer observation (e.g., adopting colleagues’ successful digital strategies), collectively disrupting negative emotional cycles (Salanova et al., 2016).

Within the landscape of Chinese online education, platforms such as Tencent Meeting and DingTalk have emerged as dominant tools. The participants in this study have experience teaching online on Tencent Meeting or DingTalk (China Internet Network Information Center, 2020). These platforms exemplify how institutional support intersects with pedagogical innovation, shaping teachers’ digital self-efficacy and emotional experiences. Based on the online teaching platforms, three types of digital teaching practices were proposed.

Type 1, Innovative Practices for Synchronized Interaction. Tencent Meeting, widely adopted for synchronous instruction, facilitates real-time interactive teaching through features like breakout rooms for small-group collaboration, shared digital whiteboards for co-constructing knowledge. These tools reduce frustration by enabling dynamic student-teacher interactions, but require institutional training to mitigate technology-specific anxiety (Wang and Li, 2022). In the future, it is important to continue to optimize and use smart technology in online teaching. Design targeted training based on teachers’ instructional needs and create dynamic learning programs to reduce technology anxiety (Wang and Li, 2022).

Type 2, Innovative Practices for asynchronous adaptation. Conversely, excels in supporting asynchronous adaptation via its “Smart Homework” system, which automates grading and generates personalized feedback using optical character recognition (OCR) and machine learning. Such practices enhance digital self-efficacy by providing data-driven insights (e.g., learning analytics dashboards) (Chen M. et al., 2022), yet may provoke negative emotion if workload balancing mechanisms are absent (Li et al., 2024). Therefore, future strategies should prioritize enhancing personalization through multimodal data integration (text, voice, video) while expanding analytics dashboards with real-time emotional feedback mechanisms to improve teacher digital self-efficacy (Chen I. H. et al., 2022; Li et al., 2024).

Type 3, innovative practices for hybrid pedagogical models. Tencent Meeting’s API compatibility allows seamless embedding of third-party tools like VR science labs, while DingTalk’s “DingTalk Classroom” combines live streaming with asynchronous discussion boards and cloud-based resource libraries. Teachers often hybridize these platforms: hosting synchronous lectures on Tencent Meeting, then assigning DingTalk’s AI-graded reflections to consolidate learning. The hybrid pedagogical model foster engagement through creative autonomy but demand robust institutional infrastructure to prevent anger from systemic inequities (Hill and Reimer, 2023).