In 2020, schools around the world were forced to respond to the COVID-19 outbreak, which caused chaos and uncertainty (Howard et al., 2021).This global shift, though necessary, introduced a variety of challenges, including technology access and teacher preparedness, which varied significantly from one country to another. Initially, some governments, such as Singapore, sought to adjust the timing of extended breaks to avert school closures. However, due to the rapid spread of the virus and the implementation of social distancing policies, many countries quickly realized that closing schools was an essential measure for controlling the outbreak (Ng, 2021). Taiwan was no exception to this trend, and when the Delta variant broke the borders, the Taiwan Central Epidemic Command Center (CECC) raised the epidemic alert to level 3 (i.e. strict social distancing), and all levels of schools were forced to suspend in-person learning and switch to online courses between May 19th and July 27th. As a result of the emergency shutdown, tech-savvy teachers initiated grassroots efforts and shared resources and teaching strategies online. For instance, online groups such as https://teachers-bar.com/ and https://bit.ly/3RGtMIX attracted over 200,000 teachers, or more than two-thirds of all teachers in Taiwan, during the school shutdown. However, the sudden shift still left many Taiwanese teachers grappling with online technologies without sufficient training, leading to increased anxiety and stress.

The dramatic change in teaching routines during the pandemic was later defined as Emergency Remote Teaching (ERT) (Hodges et al., 2020). ERT differed from regular online education in three significant ways, including being underprepared, under the threat of a pandemic, and uncertain about its duration (Cutri et al., 2020). During ERT, teachers' daily instructional minutes were replaced with increased planning, paperwork, and interactions with colleagues and parents (Jones et al., 2022). Although ERT brought extreme work pressures to teachers during school shutdowns, schools soon returned to normal operation in late 2021, after the delta and omicron variants of the COVID virus decreased in the western hemisphere and most East Asian countries. However, researchers speculate whether teachers' psychological wellbeing will return to normal as policymakers embrace the "new normal" after the post-pandemic world (Clark & Larson, 2022; Fegter & Kost, 2023; Kar & Kar, 2023; Singh et al., 2022). Some studies in India and Ecuador have suggested that teachers experience continuing emotional disorders after returning to regular face-to-face teaching, while others have found that teachers were less stressed after the COVID-19 pandemic (Arias-Flores et al., 2022; Kamath et al., 2022; Yogapriya et al., 2022; Zinskie et al., 2023). Due to such inconsistencies, there is a need to investigate the impact of ERT on digital anxiety among elementary school teachers in post-COVID Taiwan and explore the tension between embracing the new normal and clinging to the past. This study focuses on Hsinchu, known for its high concentration of tech industries, providing a unique lens through which to examine the intersection of technology and education during such crisis transitions.

Literature Review

Emergency Remote Teaching

UNESCO (2021) reported that countries reacted to school closure differently and a wide spectrum from traditional television and radio to modern learning platforms were implemented during the pandemic as educational supplements. At the period of “suspending classes without suspending learning”, researchers (e.g. García & Weiss, 2020; Kennedy et al., 2022; OECD, 2020; Sintema, 2020) had recognized global educators’ effort in closing learning losses in the past months. Unlike traditional, well-planned online teaching which may involve careful instructional design and course production (Singh & Thurman, 2019), Emergency Remote Teaching (ERT) in the past three years mostly relied on teachers improvised, adapted, and reacted to changes (Fuchs, 2022; Fuchs & Karrila, 2021). Moreover, despite being digital natives, twenty-first century students' literacies, such as general academic literacy (Salmerón et al., 2021) and digital reading literacy (Delgado et al., 2018), remain underdeveloped and require demonstration and hands-on practice from teachers. Primary school students, in particular, require the most intensive assistance from teachers (Lau & Lee, 2020; Misirli & Ergulec, 2021) and even parents (Dong et al., 2020) to learn online.

While technology-enhanced classrooms have positive aspects such as engaging and motivating students, and modern AI technologies can even share some teaching responsibilities, teachers may feel stressful when teaching with such advanced tools (Othman & Sivasubramaniam, 2019). Therefore, a line of research has focused on teachers’ capability in terms of technology integration in both online and face-to-face teaching. Areas such as infrastructure (Darab & Montazer, 2011), technology affordance (Yusuf et al., 2021), closing digital divide (Demirbilek, 2014), instructional technology and school supports (Diaz Lema et al., 2023; Ng & Nicholas, 2013), student-centered learning model (Chan, 2010), diagnostic instructional design using technologies (Pokhrel & Chhetri, 2021), professional development program for pre- and in-service teachers (Aslan & Zhu, 2017; Minihan et al., 2022), instructional technology consultations (Samifanni & Gumanit, 2021), or simply put, Technological Pedagogical Content Knowledge (TPACK, see Mishra & Koehler, 2006), were studied. Fernández-Batanero et al. (2021) found that insufficient professional development support for online teaching caused panic and stress among teachers.

Teacher Wellbeing During the COVID

With respect to teachers' wellbeing, another line of research investigated teachers' working conditions and factors regarding psychological illness. The literature revealed that teachers possess less job satisfaction both physically and psychologically because they perform professional, emotional, and physical labors intensively in schools and after schools, and the irregular working conditions due to educational and technology reforms (Johnson et al., 2005; MacIntyre et al., 2019).

Tech-savvy teachers in the twenty-first century may soon create highly interactive and engaging virtual classrooms in which students would sense belongingness and the value of learning (Vesely et al., 2022). However, this was not always the case for all teachers. In the first year of the COVID-19 outbreaks in 2020, the World Health Organization (2022) concluded that people with psychological disorders like depression and anxiety increased by 25% globally. K-16 teachers, being part of the global epidemic, suffered from stress, anxiety, or even depression under social distancing, not only because they fought the virus like ordinary people but also because they needed to establish personal online teaching capabilities while supporting their students and parents (Blume, 2020; Silva et al., 2021). For example, working from home puts teachers in exhaustion due to multitasking family and job responsibilities simultaneously (Zhou & Zhou, 2022), unrealistic online teaching expectations (Hascher et al., 2021), less-disciplined distance students (Toropova et al., 2021), and tensions between parental-teacher relationships, all leading to teachers' burnout.

Digital anxiety occurred when users uncomfortably experienced using information and communication technologies (Joiner et al., 2012; Sigursteinsdottir & Rafnsdóttir, 2022). Studies have reported that a lack of proper training caused teachers' lower self-effectiveness about online and blended teaching than traditional face-to-face teaching (Bennett, 2014), and subsequently, digital stress and anxiety emerged (Fernández-Batanero et al., 2021). In addition to quantitative surveys, qualitative measures also revealed hidden factors such as the crashing of work-life balance (Petrakova et al., 2021), and UNESCO (2020) advocated scholarly research and concluded four areas that loaded teachers with additional pressures in online teaching: assessment, teaching model, parents, and isolation. To resolve the problem, some may argue that decent stress improves performance, whilst Nayak (2014) clarifies that anxiety is an unhealthy overreaction to stress and negatively influences daily and work life. During COVID, teachers were expected to care about students' social and emotional needs in addition to resume teaching and learning (Darling-Hammond & Hyler, 2020; Shin, 2022; Wilson et al., 2023). In this sense, teacher anxiety may initiate chain reactions such as students being negatively affected by teachers' behaviors due to stress, exhaustion, depression, and anxiety (Ramberg et al., 2020; Shen et al., 2015) and eventually threaten social stability (Cortés-Álvarez et al., 2022; Shillingford-Butler et al., 2012).

Current Study

The main goal of the present study is to understand how the end of COVID-19 pandemic affects elementary teachers. Variables of interest are experience of depression, anxiety and stress. Specifically, we examine whether these factors are associated with changing of instructions (i.e. Emergency Remote Teaching and face-to-face teaching after schools reopened). This work can then inform attempts to alleviate educational impacts of future pandemic in school contexts. We examined two research questions:

  • RQ1: Did elementary school teachers’ depression, anxiety, and stress change from ERT to FTF? We hypothesize that the reopen of schools lowered teachers’ digital pressures given they again worked normally.

  • RQ2: Could any interactions of depression, anxiety, and stress among demographic variables be found in the elementary school teachers of Hsinchu between ERT and FTF? We hypothesize that the means of negative emotional states for at least one pair of teachers groups are different.

Research Design

Instrument

Many instruments were used to measure teachers' wellbeing during the COVID-19 pandemic and ERT (e.g., Jakubowski & Sitko-Dominik, 2021). To measure elementary school teachers’ emotional states during and after the pandemic, the researchers applied the Depression, Anxiety and Stress Scale—21 Items (DASS-21, Lovibond & Lovibond, 1995). DASS-21 is a combination of three dimensional self-report scales designed to measure conception of psychological disorder: Depression, anxiety and stress (Psychology Foundation of Australia, 2022). Each of the three DASS-21 scales contains 7 items, divided into subscales with similar content. The advantage of DASS-21 is that it successfully discriminates against the three highly correlated negative emotions (Lovibond & Lovibond, 1995). According to the instrumentation document, the depression scale assesses dysphoria, hopelessness, devaluation of life, self-deprecation, lack of interest / involvement, anhedonia and inertia. Moreover, the anxiety scale assesses autonomic arousal, skeletal muscle effects, situational anxiety, and subjective experience of anxious affect. Finally, the stress scale assesses difficulty relaxing, nervous arousal, and being easily upset / agitated, irritable / over-reactive and impatient. DASS-21 possesses high between-group invariance (see Norton, 2007) and internal consistency, the Cronbach’s α for depression, anxiety, and stress are 0.94, 0.87, and 0.91 respectively (Antony et al., 1998). Sizeable studies have used DASS-21 in evaluating teachers’ negative emotions (e.g. Arias-Flores et al., 2022; Kamath et al., 2022). For example, Ozamiz-Etxebarria et al. (2021) in Basque, Spain surveyed university and k-12 school teachers during ERT; 32.2% developed symptoms of depression, 49.4% anxiety, and 50.6% stress. Datta and Rej (2020) found 30–50 years old Indian female teachers expressed a significantly higher anxiety, depression and stress over other age groups due to burnout among family duties and online teaching during COVID-19 lockdown.

Scores for depression, anxiety and stress are calculated by summing the scores for the relevant items (from 0 to 3). Given that DASS-21 is the abbreviated-form of instrument, the obtained score needs to be doubled to align with original DASS-42. Lovibond and Lovibond (1995) then recommended cut-off scores for severity labels as shown in Table 1.

Table 1 Severity cut-off scores for DASS-21
Full size table

Context and Sample

Hsinchu City, often heralded as the “Silicon Valley of Taiwan,” stands as a premier example of technological immersion influencing educational practices. This city is not only home to Taiwan Semiconductor Manufacturing Company (TSMC) and over 400 high-tech firms but also boasts some of the country's highest levels of technological readiness. Such an environment has naturally facilitated the integration of digital technologies in education, making Hsinchu an ideal setting to study the impact of such integration on teacher well-being. The demographic characteristics of Hsinchu City — including the highest average income in Taiwan, a young population, and the presence of two top-tier research universities — contribute significantly to its educational dynamics. These factors collectively enhance the city’s capability to leverage digital technologies effectively within its educational system. The pervasive tech culture in Hsinchu supports an educational ecosystem where continuous professional development in digital tools is the norm, not the exception.

Hsinchu City had 31 elementary schools and 2,295 full time teachers in 2021. They were typically more adept at integrating advanced technological tools into their teaching methods, which not only enhances educational delivery but also mitigates the anxiety often associated with the forced adoption of technology seen in less prepared regions. We decided on the two-sided confidence level at 95% with a margin of error of ± 5%, and the required sample size was 330. The online survey was conducted in November of 2021, and 358 valid responses were obtained. Table 2 shows the sample distribution.

Table 2 Descriptive Statistics (sample size, mean, and SD)
Full size table

Procedure and Data Analysis

Invitations were shared by school provosts and disseminated via social media. Teachers accessed the survey by clicking a link to the survey page. After providing informed consent, participants completed the DASS-21, recalling two distinct periods: FTF and then ERT. We structured our survey questions to prompt specific recall and marked the Level 3 epidemic alert event to help respondents accurately remember their experiences during the ERT period. After completing these questions, participants responded to demographic inquiries and were thanked for their participation. Descriptive statistics and repeated measures multivariate analysis of variance (RM-MANOVA) were utilized to address the research questions.

Results

Descriptive Statistics

The representative sample of Hsinchu elementary school teachers included 359 responses. Among the remaining 358 valid cases (see Table 2), the typical persona of the surveyed teachers were married (69.3%) female (84.6%) with kids (60.6%), and they were mostly homeroom teachers (52.8%) with more than 10 years of experience (60.1%). During the school lockdown, Hsinchu teachers expressed normal to slightly mild depression(M = 5.09, SD = 5.73), moderate to severe anxiety(M = 10.91, SD = 8.80), and normal to mild stress(M = 7.16, SD = 8.11). When returning to face-to-face teaching in September, 2021, teachers’ depression(M = 2.54, SD = 4.42), anxiety(M = 6.35, SD = 7.32), and stress(M = 3.22, SD = 5.09) all returned to normal.

RQ1: Overall Changes in DASS-21

An RM-MANOVA test was conducted to test the effect on Emergency Remote Teaching. The results showed there was a difference between the ERT and FTF group on DASS21 over time, F(3, 355) = 156.66, p = 0.000, η2 = 0.57. Univariate tests also indicated there were effects on individual sub-scales, F(1, 357) = 91.60, p = 0.000, η2 = 0.204 for depression, F(1, 357) = 0.257.77, p = 0.000, η2 = 0.419 for anxiety, and F(1, 357) = 196.94, p = 000, η2 = 0.356 for stress.

RQ2: Inter-group Differences in Negative Emotional States

A two-way repeated MANOVA test was performed to analyze the effect of ERT-to-FTF (within subjects) and demographic variables of teachers (between subjects) on depression, anxiety and stress. Pillai’s trace multivariate test results are reported because we don’t assume equal variance across the groups.

By Gender

Firstly, there is a non-significant effect of an interaction effect between gender and type of instruction (i.e. ERT to FTF) on depression, anxiety, and stress, F(3, 354) = 0.95, p = 0.415; Pillai’s trace = 0.008 (see Fig. 1). Secondly, within-subject main effects analysis showed that type of instruction has a statistically significant effect on DASS-21, F(3, 354) = 57.83, p = 0.000; Pillai’s trace = 0.329; between-subject main effects analysis showed that gender has a statistically significant effect on DASS-21, F(3, 354) = 3.77, p = 0.011; Pillai’s trace = 0.031. Thirdly, univariate tests revealed significant within-subject effects, F(1, 356) = 38.56, p = 0.000, η2 = 0.098 for depression, F(1, 356) = 116.24, p = 0.000, η2 = 0.246 for anxiety, and F(1, 356) = 86.70, p = 0.000, η2 = 0.1.96 for stress, and the tests of within-subjects contrasts still show no significant interactions between gender and type of instruction on depression (p = 0.318), anxiety (p = 0.247), and stress (p = 0.240). Finally, univariate tests revealed between-subject effects, F(1, 356) = 1.49, p = 0.224, η2 = 0.004 for depression, F(1, 356) = 8.08, p = 0.005, η2 = 0.022 for anxiety, and F(1, 356) = 3.45, p = 0.064, η2 = 0.010 for stress.

Fig. 1
figure 1

Comparing the Marginal Means DASS-21 by Gender. Note: Blue: male; Green: female

Full size image

By Marriage Status

There is also a non-significant effect of an interaction effect between marriage status and type of instruction (i.e. ERT to FTF) on depression, anxiety, and stress, F(3, 354) = 2.55, p = 0.055; Pillai’s trace = 0.021 (see Fig. 2). Secondly, within-subject main effects analysis showed that type of instruction has a statistically significant effect on DASS-21, F(3, 354) = 116.71, p = 0.000; Pillai’s trace = 0.497; between-subject main effects analysis showed that marriage status has a non-significant effect on DASS-21, F(3, 354) = 1.81, p = 0.144; Pillai’s trace = 0.015. Thirdly, Univariate tests revealed significant within-subject effects, F(1, 356) = 68.78, p = 0.000, η2 = 0.162 for depression, F(1, 356) = 228.38, p = 0.000, η2 = 0.391 for anxiety, and F(1, 356) = 196.13, p = 000, η2 = 0.355 for stress, and the tests of within-subjects contrasts still show no significant interactions between marriage and type of instruction on depression (p = 0.154), anxiety (p = 0.428), except stress, F(1, 356) = 6.22, p = 0.013. Finally, univariate tests revealed between-subject effects, F(1, 356) = 1.72, p = 0.190, η2 = 0.005 for depression, F(1, 356) = 0.36, p = 0.550, η2 = 0.001 for anxiety, and F(1, 356) = 4.51, p = 0.034, η2 = 0.013 for stress.

Fig. 2
figure 2

Comparing the Marginal Means DASS-21 by Marriage Status. Note: Blue: single; Green: married

Full size image

By Having Kids

Firstly, there is a statistically significant effect of an interaction effect between having kids and type of instruction (i.e. ERT to FTF) on depression, anxiety, and stress, F(3, 354) = 3.63, p = 0.013; Pillai’s trace = 0.030 (see Fig. 3). Secondly, within-subject main effects analysis showed that type of instruction has a statistically significant effect on DASS-21, F(3, 354) = 132.04, p = 0.000; Pillai’s trace = 0.528; between-subject main effects analysis showed that having kids has a statistically significant effect on DASS-21, F(3, 354) = 4.40, p = 0.005; Pillai’s trace = 0.029. Thirdly, univariate tests revealed significant within-subject effects, F(1, 356) = 83.60, p = 0.000, η2 = 0.190 for depression, F(1, 356) = 259.30, p = 0.000, η2 = 0.421 for anxiety, and F(1, 356) = 212.18, p = 000, η2 = 0.373 for stress, and the tests of within-subjects contrasts show no significant interactions between kids and type of instruction on depression (p = 0.325), anxiety (p = 0.083), except stress, F(1, 356) = 279.89, p = 0.002. Finally, univariate tests revealed between-subject effects, F(1, 356) = 4.68, p = 0.031, η2 = 0.013 for depression, F(1, 356) = 1.99, p = 0.159, η2 = 0.006 for anxiety, and F(1, 356) = 12.32, p = 0.001, η2 = 0.033 for stress.

Fig. 3
figure 3

Comparing the Marginal Means DASS-21 by Having Kids. Note: Blue: Yes; Green: No

Full size image

By Years of Teaching Experience

A non-significant effect of an interaction effect between years of teaching experience and type of instruction (i.e. ERT to FTF) on depression, anxiety, and stress, are found, F(12, 1059) = 1.42, p = 0.147; Pillai’s trace = 0.048 (see Fig. 4). Secondly, within-subject main effects analysis showed that type of instruction has a statistically significant effect on DASS-21, F(3, 351) = 75.96, p = 0.000; Pillai’s trace = 0.394; between-subject main effects analysis showed that years of teaching experience has a non-significant effect on DASS-21, F(12, 1059) = 2.16, p = 0.012; Pillai’s trace = 0.024. Thirdly, univariate tests revealed significant within-subject effects, F(1, 353) = 57.63, p = 0.000, η2 = 0.140 for depression, F(1, 353) = 143.22, p = 0.000, η2 = 0.289 for anxiety, and F(1, 353) = 116.89, p = 0.000, η2 = 0.249 for stress, and the tests of within-subjects contrasts still show no significant interactions between years of teaching experience and type of instruction on depression (p = 0.080), anxiety (p = 0.797), and stress (p = 0.089). Finally, univariate tests revealed between-subject effects, F(4, 353) = 0.73, p = 0.570, η2 = 0.008 for depression, F(4, 353) = 0.71, p = 0.587, η2 = 0.008 for anxiety, and F(4, 353) = 7.81, p = 0.127, η2 = 0.020 for stress.

Fig. 4
figure 4

Comparing the Marginal Means DASS-21 by Teaching Experiences. Note: Blue: < 5 yrs; Green: 6 ~ 10 yrs; Brown: 11 ~ 20 yrs; Purple: 21 ~ 30 yrs; Yellow: > 31 yrs

Full size image

By Teaching Roles

Firstly, there is a statistical significant effect of an interaction effect between roles and type of instruction (i.e. ERT to FTF) on depression, anxiety, and stress, F(9, 1062) = 2.30, p = 0.014; Pillai’s trace = 0.329 (see Fig. 5). Secondly, within-subject main effects analysis showed that type of instruction has a statistically significant effect on DASS-21, F(3, 352) = 57.53, p = 0.000; Pillai’s trace = 0.329; between-subject main effects analysis showed that role has a non-significant effect on DASS-21, F(9, 1062) = 3.58, p = 0.000; Pillai’s trace = 0.029. Thirdly, univariate tests revealed significant within-subject effects, F(1, 354) = 51.12, p = 0.000, η2 = 0.126 for depression, F(1, 354) = 105.31, p = 0.000, η2 = 0.229 for anxiety, and F(1, 354) = 74.35, p = 0.000, η2 = 0.174 for stress, and the tests of within-subjects contrasts still show no significant interactions between roles in addition to teaching and type of instruction on depression (p = 0.105) and anxiety (p = 0.064), except stress, F(3, 354) = 4.12, p = 0.007. Finally, univariate tests revealed between-subject effects, F(3, 354) = 5.26, p = 0.001, η2 = 0.043 for depression, F(3, 354) = 10.00, p = 0.000, η2 = 0.078 for anxiety, and F(3, 354) = 3.97, p = 0.008, η2 = 0.033 for stress.

Fig. 5
figure 5

Comparing the Marginal Means DASS-21 by Roles. Note: Blue: homeroom teacher; Green: subject teacher; Yellow: teacher with administrative roles; Purple: special education teacher and school counselor

Full size image

Overall, all interaction effects between demographic variables and type of instruction on DASS-21 remain insignificant. Therefore, we fail to reject the null hypothesis. There is no interaction effect between gender, marriage, experience, and type of instruction on depression, anxiety, and stress scores. However, teaching roles and kids exist in moderation effects on depression, anxiety, and stress scores.

Discussion

This paper explores the impact of the COVID-19 pandemic on elementary school teachers' digital anxiety in Taiwan. Due to the fast spread of the virus and social distancing policies, Taiwan closed schools and moved to Emergency Remote Teaching in the middle of 2021. Although proper use of technology-enhanced online learning can motivate and engage students, teachers may experience anxiety when teaching with advanced tools due to a lack of professional development support for online teaching (Wong et al., 2021). Elementary school teachers in Hsinchu, on the other hand, experienced only slightly mild depression, moderate to severe anxiety, and normal to mild stress. Once the ERT ended, their levels of depression, anxiety, and stress soon all significantly returned to normal (see figures and “E–F” columns in Table 3).

Table 2 Summary of significant differences among DASS-21 Comparisons
Full size table

Our analysis presents a nuanced perspective contrasting the findings of Pressley et al. (2021), who argued that changes in the mode of instruction to ERT impacted teachers’ digital stress and anxiety, but not demographic differences. In our study, some group comparisons during the ERT to FTF transition revealed significant demographic influences on these stressors among elementary school teachers in Hsinchu. Interestingly, we found that female teachers exhibited higher levels of anxiety compared to their male counterparts, regardless of the shift back to normal teaching settings. This gender difference highlights the persistent anxiety issues that are not solely contingent on teaching mode but perhaps also on broader gender-related challenges in the workplace. In contrast, teachers with children reported lower levels of depression, anxiety, and stress compared to those without, suggesting that the office culture among teachers may also play a role, particularly showing more empathy towards those needing to care for children during the pandemic. This cultural aspect could explain why teachers without children felt more burdened by ERT-related stress (Leo et al., 2022).

Furthermore, no significant differences were observed based on marital status or years of teaching experience, indicating that these factors might not be as influential in determining teachers' responses to ERT as previously thought. The unique technological readiness of Hsinchu City, known for its high integration of digital tools in educational settings, may have buffered some of the anticipated demographic impacts, contributing to a smoother transition back to normalcy for many teachers. However, the role played within the school had a pronounced effect on stress levels. Administrative teachers, such as principals and provosts, who typically have fewer teaching hours and do not serve as homeroom teachers, reported the lowest levels of depression, anxiety, and stress during both ERT and FTF periods, likely due to their ability to delegate tasks (Demirbilek, 2023). Conversely, special education teachers faced the highest levels of these stressors in both educational settings. Their continued high scores on the DASS-21 suggest an ongoing struggle, compounded by the lack of adequate technical support in resource classrooms as noted by Al-Amri (2022) and Trzcińska-Król (2020). The shift to post-COVID hybrid instruction is expected to maintain these elevated stress levels among special education teachers (Rice & Barbour, 2023).

With regard to interactions, a few interaction effects were found in two areas: Having kids and roles against stress when returning to normal (see “Inter.” columns in Table 2). The slope of stress dropped steeper for teachers without kids when returning to normal (see Fig. 3). Besides, homeroom teachers had the highest level of stress as special education teachers during ERT, and their slope dropped deeper and approached the stress of subject teachers when returning to normal. Childless teachers are often expected to take on greater responsibility within the teaching community, resulting in increased pressure on them during ERT. In an early commentary, Kraemer et al., (2009, p.70) noted: “Innovation does not depend only on the nature of the innovation itself. Often, more important is the social and cultural environment in which it will operate”. New technology turbulence (e.g. Generative AIs) often bring waves of hype and disruption, but the whirlwind eventually settles into a tranquil sea; similar to what Cuban (2001) described, the full potential of new technology requires the epistemological change of teachers and school. The discrepancy of changes in DASS-21 among teacher roles might hint that the epistemological change did not happen at a holistic scale.

Conclusion

This study provides insights into teachers' digital anxiety after COVID-19 and the challenges they face during ERT, contributing to the discourse on teachers' psychological wellbeing. Although we sampled teachers from a high-tech town whose DASS-21 scores were merely mild depression, anxiety, and stress, differences of change were found when taking both teacher demographics and ERT-FTF transition into account. Based on our findings, we recommend the development of comprehensive professional development programs tailored to enhance digital competencies among teachers while concurrently offering support mechanisms to address digital anxiety. Additionally, creating a community of practice within schools can provide ongoing peer support and foster a collaborative environment for sharing effective digital teaching strategies. Further investigations could explore longitudinal effects of digital anxiety and the effectiveness of specific interventions designed to mitigate these impacts. Comparative studies across different educational levels and subject specializations could also provide deeper insights into the varying needs of teachers and the efficacy of tailored support systems in diverse educational contexts (e.g. Lizana, & Lera, 2022).

This study utilizes retrospective data to examine the change between ERT and FTF on the digital anxiety of teachers in Hsinchu, acknowledging inherent limitations such as recall bias and selective memory. To address these issues, the study employed structured survey techniques designed to aid accurate recall. Despite these efforts, the retrospective nature of data collection remains a limitation. Furthermore, the study focuses exclusively on elementary school teachers in Hsinchu, a region with high technological readiness, which may not be representative of other educational contexts, particularly those with less technological infrastructure.

Taiwan Premier's recent US$719.6 million investment in mobile devices, large-screen displays, and one-on-one digital learning for k-12 schools in 2022–2025 may bring another digital turbulent change for teachers (Executive Yuan, 2021). As the government invests in digital technology for k-12 schools, our insights gained here should inform broader educational policy and practice, ensuring that teachers are not only equipped with the necessary technological tools but are also supported in adapting to these tools without undue stress or anxiety. Emphasizing teacher well-being is essential to fostering resilient educational systems capable of navigating future challenges.