Online teaching self-efficacy of Chinese university teachers amidst Covid-19: its changes and the moderation of adaptability and administration quality

ABSTRACT

Teacher self-efficacy is among the most valued teacher motivational constructs. However, little is known about university teachers’ self-efficacy and even less about changes to it throughout the Covid-19-related online teaching. This study applied a retrospective pre- and post-design to investigate changes in online teaching self-efficacy (OTSE) during Covid-19. Participants included 160 Chinese university teachers, who reported their OTSE before and after the COVID-19 lockdowns, adaptability and administration quality together with demographic information. The self-efficacy for online instruction failed to increase significantly over this period (β=.21, p = .083), whereas that for online technology applications increased significantly (β=.329, p < .01). Individual adaptability and administration quality significantly moderated the changes in OTSE. The implications and limitations of the study are discussed.

KEYWORDS:

• Teacher self-efficacy
• higher education
• adaptability
• online teaching

Online teaching has existed in various forms prior to Covid-19, such as massive open online courses (MOOCs) and distance education. More than half a million online courses in China have been provided to approximately 30 million university students (Ministry of Education of the People’s Republic of China [MEoPRC], 2022). All higher education institutions transitioned to online teaching in response to COVID-19 and, until May 2020, approximately 1.08 million university teachers taught 3.5 billion student attendees (MEoPRC, 2020). To effectively curb the spread of Covid-19, China executed its dynamic zero-case policy in December 2021 and districts with unexpected cases reported enacted social lockdowns, including transferring to university teaching online. For instance, the March to June outbreak in Shanghai in 2022 caused universities in Shanghai to return to online teaching (Shang Municipal Health Commission, [SMHC], 2022). The emergency online teaching experience elicited unprecedented challenges for both university teachers and students, and received much attention from researchers (Casacchia et al., 2021; Tao & Gao, 2022). However, limited is known about university teachers’ perceptions of their capability in conducting online teaching, especially its changes during this period (Myyry et al., 2022).

Teachers’ perception of their capability to complete specific tasks needed to be an effective teacher, termed as teacher self-efficacy (TSE), has been regarded as one of the most influential motivational constructs in teacher education (Perera et al., 2019). Teachers with higher TSE tend to be more persistent in challenging situations and more motivated to maintain high expectations for their students (Bruce et al., 2010). Higher TSE protects teachers from experiencing stressful feelings associated with unanticipated educational reforms (Putwain & von der Embse, 2019) and increases their intentions to experiment with innovative pedagogies, including initiating student collaboration and critical thinking (Boeve de Pauw et al., 2022). At the commencement of the online teaching period associated with Covid-19, many university teachers were concerned about their capability to maintain the learning quality of their students (Bao, 2020). Such a perception of online teaching capability might threaten university teachers’ mental health and cause them to hesitate when experimenting with online pedagogies (König et al., 2020). However, existing research on TSE has predominantly been based on physical classroom teaching of school teachers and has rarely has been investigated among university teachers (Ma et al., 2022; Matos et al., 2021). Although the call for more research on the self-efficacy of university teachers has long been made (Landino & Owen, 1988), the situation has not yet changed (Matos et al., 2021). Therefore, it is essential to investigate whether/how university teachers’ OTSE changes when adopting online teaching and moderators that influence it (Myyry et al., 2022). Research fulfilling such an aim will extend the current understanding on TSE, which is mainly based on physical school teaching, to university teachers’ online teaching and will benefit the design of professional development to assist them in developing a robust OTSE that, in turn, might enhance the online learning quality of tertiary students.

The construct of self-efficacy

Self-efficacy indicates the perceived capability of individuals to execute specific actions needed to achieve certain goals and it may elicit greater effects on a person’s behaviors and motivations to conduct such behaviors, than the particular skill sets they actually possess (Zimmerman, 2000). Less self-efficacious individuals tend to avoid innovative tasks, whereas those more likely to achieve success are inclined to pursue their goals (Locke & Latham, 2006). Furthermore, level of self-efficacy is associated with whether or not individuals feel stressed and depressed (Bandura, 1997). For instance, highly efficacious people tend not to become depressed due to their increased belief in their capability to overcome any difficulties that may arise.

Self-efficacy is domain-specific and being highly self-efficacious in completing certain tasks does not ensure individuals will feel capable in all other tasks (Bandura, 2019). Such a particularity also applies to different dimensions of completing one task. Individuals may feel capable of completing certain aspects of one task, but uncertain about other aspects of the same tasks (Perera et al., 2019). Self-efficacy tends to be malleable in its initial developmental stages, however, it resists change without extra effort once it becomes stabilized (Bandura, 1997).

The changeability of self-efficacy is possible with information from four sources, namely mastery experience, vicarious experience, verbal persuasion, and physiological and emotional states (Bandura, 1997). Mastery experience elicits the most powerful source and indicates individual capability through experiencing failures and successes in accomplishing similar tasks (Menon & Sadler, 2018). This source is especially effective when individuals are able to achieve their goals by making sufficient efforts to achievement them (Poulou, 2007). The experience of observing other individuals complete similar tasks also informs individuals about their capability to do the same, by considering the similarity shared with others. In addition, individuals can be persuaded that they are capable of completing tasks, or not, by others, especially those regarded as trustworthy, and people tend to make varying judgments about their capability in when experiencing different physical and emotional conditions (Bandura, 1997).

Teacher self-efficacy and online teaching

The construct of self-efficacy has been applied in multiple fields and in teacher education it has been operationalized as teacher self-efficacy (TSE). TSE has been applied to indicate individual teachers’ perceptions of their ability to complete the tasks required to be effective and is regarded as one of the predominant constructs in teacher education (Perera et al., 2019). Much research on TSE has been conducted among school teachers and it has been found to predict ‘teachers’ intentions to adopt innovative teaching methods (Boeve de Pauw et al., 2022), remain in the profession (Chesnut & Burley, 2015), and the ability to protect themself from stressful emotions (Cho et al., 2021). In contrast, limited research has been conducted with university teachers (Matos et al., 2021). Recent studies have demonstrated that highly efficacious university teachers are more likely to provide individualized instruction and assessments for their students (König et al., 2020) and report higher levels of occupational satisfaction (Matos et al., 2021). However, the existing TSE research has mainly focused on physical classroom teaching, and there has been little focus on online contexts and has relied on measurements such as the most frequently applied Teacher Sense of Efficacy Scale (TSES, Tschannen-Moran & Woolfolk Hoy, 2001).

The TSES scale covers three main domains, namely applying various instructional strategies, engaging students in teaching, and maintaining classroom discipline (Tschannen-Moran & Woolfolk Hoy, 2001). The TSES has been validated in multiple cultural backgrounds, either with adaptations, such as in Germany and New Zealand in Pfitzner-Eden et al. (2014) and China, Japan, and Korea in Ruan et al. (2015) or without, in Mexico in Salas-Rodríguez et al. (2021). However, solely measuring domains associated with classroom teaching is ‘undisclosed’ (Friedman & Kass, 2002) and, therefore, extra tasks, including socializing with school staff and students, have also been measured (Skaalvik & Skaalvik, 2007). Seven dimensions of TSE have been found using the TSES, including influencing school administrative decisions, acquiring school resources, instruction, discipline maintenance, engaging parents, involvement with communities, and building a positive school climate (Bandura, 2006). However, none of these subscales have received as much attention as the full TSES (Authors, 2020). A short form of this scale was also adapted for use with Chinese school teachers and received satisfactory model fits (Ma et al., 2021, 2022).

Online teaching requires teachers to develop more distinct competencies than physical classroom teaching (Robinia, 2008). One of the most commonly reported challenges was the escalation in teachers’ reliance on voice to maintain effective communication with students when teaching online, due to the reduced communication effects of facial expressions and physical gestures (Bao, 2020). It becomes necessary to engage students by using multiple technologies, including multimedia, which might, in turn, enhance students’ commitment to online learning (Albee, 2015; Verma et al., 2020). Also, teachers need to become even more sensitive to changes in their students’ emotions, due to the lack of interaction between peer students and teachers in online teaching contexts (Cardullo et al., 2021). Engaging students in such a context required teachers to refine their pedagogies that are differing with those in physical teaching context, for instance, designing online learning environment, ustilizing educational technology and interacting with students (Moser et al., 2021). Although much difference exists between competencies needed for physical and online teaching, there have been few attempts to validate TSE scales specifically for online teaching. The only validated self-efficacy scale for online teaching, the Michigan Nurse Educators’ Sense of Efficacy for Online Teaching Survey, was adapted from the TSES and contains two domains, namely TSE for online instruction and TSE for online technology application (Robinia, 2008).

The emergency transition to online teaching challenged university teachers’ self-efficacy for online teaching. They felt challenged to format course content and assess their students’ attention, and to utilize technologies to involve students in online group discussions (Casacchia et al., 2021). Lacking experience in the physical classroom or online teaching, and having insufficient knowledge of digital technology, were among the most frequently reported factors negatively impacting university teachers’ OTSE (Horvitz et al., 2015). The effects of prior experience with using relevant technologies were especially apparent in assisting teachers with developing higher TSE for designing online learning materials and integrating inclass activities to understand and engage students (Xu et al., 2021). In contrast, Watermeyer et al. (2021) found that most university teachers reported being efficacious in online teaching and conducting assessments, when provided with sufficient assistance from their faculty. Improvements in OTSE of university teachers were reported during a one-month emergency online teaching session, with no identified differences between those from varying disciplines, with teachers who did not have prior online teaching experience increasing their OTSE most noticeably (Myyry et al., 2022). The length of both physical and online teaching experience was found to predict variations in OSTE, which might be because of the communication skills developed during such an experience (Lei & Medwell, 2021). Male teachers who have taken online training related to online teaching outperformed female teachers in feeling prepared for transitioning into online teaching (Cardullo et al., 2021). However, university teachers also reported enduring low OTSE due to the less satisfactory perception of reduced teacher social presence and involving the participation of students of different types, including those maintaining silence and refusing open camera (Song, 2022).

Among the factors influencing OTSE, a supportive learning environment that includes a learning management system and technological resources, was found to predict teachers’ self-efficacy in assisting students’ online learning and keeping them engaged (Cardullo et al., 2021). Administrative support, by providing technical assistance and assuring students’ digital presence, could enhance teachers’ intention to conduct and improve their online teaching (Ma et al., 2021, 2022). Teachers also need more exposure to updated technology, to be guided to major online instructional design and pedagogical knowledge, and the provision of financial reimbursement and extra time for professional development (Kisanga, 2016). Therefore, it is essential to investigate the moderation effects of administration quality on the changes in OTSE during this emergency online teaching period.

Teacher adaptability associated with online teaching amidst Covid-19

Changes elicit challenges to teachers’ adaptability, including emergency transitions to online instruction due to COVID-19 (Stockinger et al., 2021), with a predominant challenge being how to ensure the quality of students’ online learning in higher education. Of note, students’ overall academic scores decreased significantly by 0.2 on average during COVID-19 disruptions to higher education (Orlov et al., 2021) and, although university students improved their professional knowledge, their social communication skills failed to improve (Ba˛czek et al., 2020). Teachers need to adjust their teaching to strengthen students’ online learning quality, by keeping learning activities aligned with course objectives, ascertaining the specific needs of students, continuously adjusting teaching strategies, enhancing the social connection of students, and fostering communication among students (Filius et al., 2018). They tried to adjust their teaching by restructuring the course content to maintain a positive perception of their student’s learning performance (Moser et al., 2021). Also, teachers need to make adjustments to maintain online teaching interactions with students, which is regarded as one of the most challenging domains by teachers (Tao & Gao, 2022). Teachers had to make extra emotional efforts to initiate the online connection with students and felt challenged to engage students at an indepth level as they were mainly able to introduce students to learning materials and conduct teacher-student interaction instead of organizing student-centered activities (Le et al., 2022). However, persistence with measures like making discussion compulsory even in problem-led instruction failed to enhance the higher-order cognition of students (Biggs, 2012). Unexpected challenges for teachers have also been caused by students’ unfamiliarity with technology, limited time frame, and inaccuracy in evaluating the level of engagement of students (Gosselin, 2009). Therefore, teachers must continue to make adjustments to maintain the in-depth engagement of students, to assist them to achieve the equivalent academic achievements, indicated by critical thinking capability and passing rate of examinations (Berthelon et al., 2019; Filius et al., 2018). In addition, all of these innovative challenges associated with students raised demands for teachers to regulate and maintain individual emotional balance by making extra emotional input and seeking collegial assistance (Liu et al., 2021).

Adaptability indicates the psycho-behavioral mechanism that assists individuals in regulating thinking, actions, and emotions in response to unprecedented variations in their daily lives (Martin et al., 2012). This construct differs from resilience and other defense mechanisms because the latter focuses on challenges, whereas adaptability addresses the unpredictable characteristics of daily life (Collie & Martin, 2016). Adaptability has been well explored among students across different levels and is associated with various domains of both personal development and academic performance (Collie & Martin, 2016). It has been found to be significantly related to their enjoyment at school, satisfaction with life, and sense of meaningfulness (Burns et al., 2018). It is also associated with university students’ Covid-19-associated negative emotions, including anxiety and hopelessness (Stockinger et al., 2021). However, research on the adaptability of teachers is limited (Martin et al., 2021). Addressing this lack of knowledge becomes more essential considering the nature of the teaching profession, which requires teachers to respond to students’ instant actions, satisfy their diversified needs, and make constant reforms in education (Parsons et al., 2016).

Much attention has been paid to investigating the cultivation of teacher adaptability, considering its positive effects on teachers’ professional development, including TSE (Martin et al., 2021). Among the existing research, teacher assistants who reported being adaptable tended to have high levels of occupational satisfaction, be involved in the workplace, develop a positive attitude toward individual capability, and become more resilient in the face of challenges (Martin et al., 2021). Teachers who coped with stress associated with the emergency online teaching transition more flexibly tended to report fewer negative emotions, including anxiety (MacIntyre et al., 2020). Teachers who can adjust their thoughts, actions, and emotions could become more resilient by disengaging from their feelings in their occupation (Collie et al., 2018). Highly adaptive teachers are more likely to fulfill working demands and achieve their occupational goals (Collie & Martin, 2016), which might, in turn, enhance their positivity about their capabilities. For instance, school teachers’ adaptability has been found to predict both domains of OTSE, namely online instruction and technology application, during COVID-19 (Ma et al., 2021, 2022). However, few studies have been conducted that examine the moderation effects of adaptability on OTSE changes among university teachers.

Study aims

The present study utilized a retrospective pre- and post-design to explore two research questions: a) whether and how the OTSE of university teachers changed during roughly an 18-month period, ranging from the commencement of Covid-19-related online teaching and the time when they were surveyed in June 2021; b) whether and to what extent teachers’ adaptability and administration quality in university moderate changes in OTSE.

Method

Instruments

Online teaching self-efficacy

The present study applied the Scale for Online Teaching Self-Efficacy (Ma et al., 2021), adapted from the Michigan Nurse Educators’ Sense of Efficacy for Online Teaching Survey (Robinia, 2008), namely. This 10-item scale has two dimensions, self-efficacy for online instruction (e.g., gauging student comprehension of what you have taught) and self-efficacy for technology application (e.g., using asynchronous discussions, such as in-time chat rooms, to maximize interaction between students in an online course). The items were rated based on a 7-point scale, with 1 and 7 labeled as extremely uncertain and absolutely certain, respectively. In the current study, the two-dimension structure was confirmed, and the model fits being χ²/df = 2.34, CFI =.968, TFI =.954, RMSEA =.088, SRMR =.0411 at Time 1 and being χ²/df = 2.22, CFI =.966, TFI =.948, RMSEA =.088, SRMR =.0454 at Time 2. In addition, the Cronbach’s alphas for self-efficacy for online instruction were .915 and .905, and for self-efficacy for technology application were .913 and .906 at Time 1 and 2, respectively.

Adaptability

The present study used the adaptability scale developed by Martin et al. (2012), which covers individual adaptability in cognition (e.g., revise the way I think about a new situation to help me through it), behavior (e.g., change the way I do things if necessary), and emotion (e.g., minimize frustration or irritation so I can deal with it best). This scale has been validated among Chinese school teachers and received satisfactory internal consistency (α = 0.775). In the present study, the model fits for the second-order factor structure with two subscales, namely behavior-cognition and emotion, were satisfactory, χ²/df = 1.72, CFI =.990, TLI =.985, RMSEA =.067, and SRMR =.016 and the Cronbach’s alpha was .962. This scale has nine items and is rated using a 7-point scale, with 1 and 7 being labeled as extremely disagree and completely agree, respectively.

Administration quality amidst Covid-19 online teaching

Four items were created to investigate the administration quality of the participants’ institutions. They were a) whether your university (faculty/department) provided sufficient online teaching equipment and resources, b) whether your university(faculty/department) executed effective online teaching supervision (e.g., compulsory online sign-in, classroom discipline, and engagement assessment), c) whether your university(faculty/department) assisted its staff and students in solving technical issues, and d) whether your university(faculty/department) conducted professional development projects on online teaching. These items were rated based on a 5-point scale, ranging from (1) completely disagree to 5 (extremely agree) and had a Cronbach’s alpha of .734 in the current study.

Procedure and sample

Permission was received from the institution and all participants provided informed consent prior to participation. An online survey was utilized to collect data for the present study using snowballing recruitment. The survey contained four sections: demographic information, the Scale for Online Teaching Self-Efficacy, the adaptability scale, and the four items on school administration. Specifically, participants were instructed to reflect on their OTSE at two time points, namely at the beginning of Covid-19, in early January 2020, and when they were surveyed at the end of June 2021. Although doubts exist over recall accuracy, this retrospective method has been recommended to collect data at different time points due to its advantages over traditional pre/post-designs in reducing attrition (see Euser et al., 2009). Furthermore, this method outperforms traditional pre/post designs in allowing participants to make judgments based on a consistent understanding of the constructs and prohibiting them from making invalid preliminary evaluations, due to a lack of awareness as often witnessed in the latter (Little et al., 2020).

The sample size needed for the present study was estimated based on Nunnally’s (1967) suggestion of the 1/10 ratio between the number of measurement items and sample size in structural equation modeling. A minimum of 100 participants were needed to complete a confirmatory factor analysis of the scale with the largest number of items, namely the Scale for Online Teacher Self-Efficacy (n = 10). A total of 160 participants warrants a sufficient sample size in the study.

Data analysis

A four-step procedure was utilized to analyze the quantitative data. First, the confirmatory factor analysis (CFA) was used to examine the factorial structure of the scale for online TSE at both times, the scale for adaptability, and the administration quality items. A combination of model fits, including χ2/df, comparative fit index (CFI), Tucker-Lewis index (TLI), the root mean square error of approximation (RMSEA), and the standardized root means square residual (SRMR), were applied with χ2/df < 2, CFI and TLI≥0.95, RMSEA ≤.08, and SRMR ≤.06 identifying good models in line with previous literature (Hooper et al., 2008). Second, a two-step multilevel modeling was conducted in SPSS to investigate OTSE changes (Harrison et al., 2018; Heck et al., 2013). Time was operationalized as the only explanatory factor of the within-individual variation to check how TSE changed over time in Model 1. This model was regarded as the baseline model. Demographic variables were added as explanatory factors to adjust the initial difference in TSE levels attributable to the participants’ background information in Model 2. The model fits were judged using −2 times log-likelihood (−2LL) and Akaike’s information criterion (AIC), with smaller values indicating better model fit. The significance of −2LL was tested based on the chi-square distribution (Heck et al., 2013), and a reduction of not less than 6 in AIC refers to a significant improvement in the model fit (Harrison et al., 2018). Results are reported based on the best-fit model. Thirdly, the moderation effects of adaptability and administration quality on changes in TSE for online teaching were tested using PROCESS in SPSS. TSE scores at Time 1 and Time 2 were included as independent and dependent variables, with adaptability and school administration quality included separately as moderators. Fourth, linear regression was applied to detect predictability.

Results

Descriptive findings

A total of 169 responses were included in the final analysis. There were 91 (53.8%) male and 78 female (46.2%) teachers. Among these, 37 (21.9%) had teaching experience of fewer than five years, 20 (11.8%) taught for more than five but less than ten years, 31 (18.3%) taught for more than ten but less than 15 years, 33 (19.5%) taught for more than 15 but less than 20 years, and 48 (28.4%) taught for more than 20 years. Ninety-five (56.2%) were from the science department, 38 (22.5%) from engineering, 13 (7.7%) from management, 13 (7.7%) from education, 4 (2.4%) from medicine, 4 (2.4%) from literacy, and 2 (1.2%) from law. Twenty-nine (17.2%) were from state key universities, 34 (20.1%) from provincial key universities, 66 (39.1%) from regular provincial universities, and 29 (17.2%) from local universities. Regarding the locations of the universities, 30 (16.9%) were from metropolitan cities, 66 (39.1%) from second-tier cities, 32 (18.9%) from third-tier cities, 25 (14.8%) from fourth-tier cities, and 16 (9.5%) not specified. The majority of teachers (n = 115, 68%) taught synchronously, 12 (7.1%) taught asynchronously, 9 (5.3%) used remote supervision, and 5 (3.1%) used MOOC to teach.

Means, standard deviations, and correlations between different variables are listed in Table 1. The two domains of OTSE correlated significantly across two time points, with r ranging from .288 to .709. The correlations between adaptability and the two domains of OTSE were significant, with r ranging between .189 and .662.

Table 1. Correlation coefficients between OTSE, adaptability, and school administration.

Variables	OTSE for instruction1	OTSE for instruction2	OTSE for technology1	OTSE for technology2	Adaptability	School administration
OTSE for instruction1^a	1
OTSE for instruction2^b	.288**	1
OTSE for technology1^c	.628**	.402**	1
OTSE for technology2^d	.363**	.709**	.500**	1
Adaptability	.579**	.582**	.588**	.580**	1
School administration	.053	.088	.084	.072	.102	1
Mean (SD)	5.52(.860)	5.25(.961)	5.32(.951)	5.61(.964)	5.58(.921)	4.22(.671)

Note. ^a,b OTSE for instruction1 and OTSE for instruction2 indicate self-efficacy for online instruction at Time 1 and Time 2, respectively; ^c,d OTSE for technology1 and OTSE for technology2 indicate self-efficacy for technology application at Time 1 and Time 2, respectively.

** p < .01, * p < .05.

Changes in OTSE

The model fits of the second model improved significantly for both domains of OSTE (see Table 2), namely Δ AIC>6 and Δ −2LL > 3.84 with degree of freedom being 1. However, the two domains showed different change patterns. Self-efficacy for online instruction had an initial level of 5.072, se =.505, t(104.926) = 10.051, p < .001, 95% CI [4.071, 6.072] and failed to increase significantly (β = .210, se =.120, t(119) = 1.748, p = .083, 95% CI [−.028, .447]), whereas that for technology application had a lower initial level of 4.92, se =.60, t(103.478) = 8.197, p < .001, 95% CI [3.731, 6.112], but improved significantly (β = −.329, se =.102, t(119) = −3.233, p < .01, 95% CI [−.531, −.128]).

Table 2. Parameters and model fits for multilevel modeling for OTSE.

Parameters	OTSE for instruction^a		OTSE for technology^b
	Model1	Model2	Model1	Model2
Fixed effects
Intercepts	5.20**	5.07**	5.564**	4.931**
Time 1	.217*	.210	−.362**	−.329**
Time 2^c
Female		.133		.372
Male^d
Less than 5 years		.128		.374
5 to 10 years		−.326		−.226
Less than 15 years		−.338*		.067
15 to 20 years		−.225		−.053
More than20 years^e
Engineering		−.697		−.604
Education		−2.01**		−.1.777**
Science		−.626		−.845*
Management^f
State Key university		.393		.230
Provincial Key university		.504		.658*
Regular provincial university		−.018		.284
Local university^g
First tier city		−.001		.178
Second tier city		−.106		−.017
Third tier city		.153		−.098
Fourth tier city^h
Synchronous teaching		.321		.364
Asynchronous teaching		.361		.361
Other formats^i
School administration		.110		.139
Random effects
Residual	.861**	.863**	.710**	.622**
Intercept	.253**	.168	.531**	.544**
Model summary
AIC	994.939	698.814	1006.368	695.946
−2LL	990.939	694.814	1002.368	691.946
Number of parameters	4	21	4	21

Note. ^a indicates self-efficacy for online instruction; ^b indicates self-efficacy for technology application; ^{c,d,e,f,g,h,i.} the parameters were redundant.

** p < .01; * p < .05.

Moderation effects of adaptability and administration quality on OTSE changes

The moderation effects of adaptability and administration quality on self-efficacy for online instruction and technology, are reported separately (See Table 3). First, regarding the moderation of adaptability on self-efficacy for online instruction, the overall model explained 37.8% of the variance in self-efficacy for online instruction at Time 2, R² = .378, F(3,165) = 33.40, p < .001. Specifically, self-efficacy for online instruction at Time 1 significantly predicted that at Time 2, β = −.89, SE =.334, t(165) = −2.662, p = .009, 95% CI [−1.546, −.229]. However, adaptability failed to predict self-efficacy for online instruction at Time 2, β = −.442, SE =.315, t(165) = −1.403, p = 1.625, 95% CI [−1.065, .18]. The interaction effect was significant, β = −.188, SE =.059, t(165) = 3.172, p = .002, 95% CI [.071, .305]. Second, regarding the moderation of adaptability on changes in self-efficacy for technology application, the overall model explained approximately 40% of the variance in self-efficacy for technology application, R² = .401, F(3,165) = 36.759, p < .001. Self-efficacy for technology application at Time 1 significantly predicted that at Time 2, β = −1.223, SE =.329, t(165) = −3.717, p < .001, 95% CI [−1.873, −.573]. However, adaptability failed to predict self-efficacy for technology application at Time 2, β = −.379, SE =.332, t(165) = −1.141, p = .226, 95% CI [−1.034, −.277]. The interaction effect was significant, β = .203, SE =.06, t(165) = 3.396, p < .001, 95% CI [.085, .321].

Table 3. Parameters for the moderation modeling.

Variables	β	SE	t	p	95%CI
OTSE for instruction2^b
Constant	7.71	1.743	4.424	<.001	[4.27, 11.156]
OTSE for instruction1^a	−1.223	.329	−3.717	<.001	[−1.873, −.573]
Adaptability	−.379	.332	−1.141	.256	[−1.034, .277]
Adaptability*OTSE for instruction1	.203	.06	3.396	.<.001	[.085, .321]
Constant	7.80	1.326	5.881	<.001	[5.18, 10.417]
OTSE for instruction1	−.55	.242	−2.27	.025	[−1.028,-.072]
School administration	−1.093	.364	−3.003	.003	[−1.812, −.374]
School administration*OTSE for instruction1	.219	.066	3.346	.001	[.09, .349]
OTSE for technology2^d
Constant	7.10	1.686	4.211	<.001	[3.771, 10.428]
OTSE for technology1^c	−.888	.334	−2.662	.009	[−1.546, −.229]
Adaptability	−.442	.315	−1.403	.163	[−1.065, .18]
Adaptability*OTSE for Technology1	.189	.059	3.172	.002	[.071, .305]
Constant	6.608	1.103	5.993	<.001	[4.431, 8.784]
OTSE for technology1	−.212	.215	−.987	.325	[−.636, −.212]
School administration	−.866	.293	−2.957	=.004	[−1.443, −.288]
School administration* OTSE for Technology1	.168	.056	3.002	.003	[.057278]

Note. ^a,b OTSE for instruction1 and OTSE for instruction2 indicate self-efficacy for online instruction at Time 1 and Time 2; ^c,d OTSE for technology1 and OTSE for technology2 indicate self-efficacy for technology application at Time 1 and Time 2.

Regarding the moderation of school administration quality on self-efficacy for online instruction changes, the overall model explained 12% of the variance in self-efficacy for online instruction at Time 2, R² = .12, F(3,165) = 7.496, p < .001. Specifically, both self-efficacy for online instruction at Time 1, β = −.55, SE =.242, t(165) = −2.27, p = .025, 95% CI [−1.028, −.072] and administration quality, β = −1.093, SE =.364, t(165) = −3.003, p = .003, 95% CI [−1.812, −.374], significantly predicted self-efficacy for online instruction at Time 2. The interaction effects were significant, β = −.219, SE =.066, t(165) = 3.346, p = .001, 95% CI [.090, −.349]. For self-efficacy for technology application changes, the overall model explained 22.7% of the variance in self-efficacy for technology application at Time 2, R² = .227, F (3,165) = 16.115, p < .001. Specifically, self-efficacy for technology application at Time 1, β = −.212, SE =.215, t(165) = −.987, p = .325, 95% CI [−.636, .212] and administration quality, β = −.866, SE =.293, t(165) = −2.957, p = .004, 95% CI [−1.443, −.288], both significantly predicted self-efficacy for technology application at Time 2. The interaction effects were significant, β = −.866, SE =.056, t(165) = 3.002, p = .003, 95% CI [.057, .278].

Discussion

The present study provided initial evidence of changes in OTSE of a group of Chinese university teachers before and after their Covid-19-related online teaching. Also, it investigated the moderation effects of individual adaptability and school administration on changes in OTSE.

The self-efficacy for online instruction of these university teachers failed to increase significantly, in contrast with that before Covid-19, whereas self-efficacy for technology application increased significantly after the Covid-19-related online teaching period. This finding is aligned with the theoretical assumption on the domain specificity of self-efficacy (Bandura, 2019) and extends it by adding evidence on varying trajectories in different domains of online teaching. It also indicates the necessity to include and examine the correspondent self-efficacy for extra domains associated with online teaching, as indicated in prior studies, for example, course restructuralization in Casacchia et al. (2021), conducting student assessments in Watermeyer et al. (2021), and responding to ‘students’ psychological conditions in Cardullo et al. (2021). However, self-efficacy in these potential domains was not explored in the current study due to the two-domain structure of the OTSE, namely online instruction and technology application. The two-factor structure may result from this scale being based on the TSES, which was solely contextualized within classroom teaching in school contexts (Friedman & Kass, 2002; Salas-Rodríguez et al., 2021). This is especially important considering the declinations reported in cultivating students’ communicating capability, compared with their gains in professional knowledge (Ba˛czek et al., 2020).

The insignificant growth in self-efficacy for online instruction in the present study contradicts the findings of Putri et al. (2020), reporting that teachers might have been concerned with the generalizability of their current teaching capabilities in this innovative teaching context. The concern is due to the need to develop innovative pedagogies for online teaching, which differ from traditional physical classroom teaching. However, in this study, teachers tended to rely on traditional instructional strategies used in physical classroom teaching and failed to develop instructional strategies applicable to online teaching contexts. Filius et al. (2018) reported that the lack of online teaching professional capability is the main challenge for teachers to transfer from physical teaching to online teaching successfully. This failure to perceive competency in online teaching capability can be explained by the findings of the present study, reporting that an increase in online teaching experience did not witness OTSE growth automatically.

Self-efficacy for technology application increased significantly, although this was reported to be lower than that for online instruction, findings that are only partially aligned with those of Verma et al. (2020), who reported that technology integrating challenges were among the predominant threats for teachers when they transferred to online teaching. This indicates that technology application could elicit challenges for teachers at the beginning of utilization before their self-efficacy increases with the accumulation of experiences, demonstrating the positive change in self-efficacy with mastery experience (Bandura, 1997).

Caution should be taken to not assume that university teachers improve their capability to integrate technology, despite improvements in their self-efficacy for technology application, as found in this study. In contrast, the insignificant increase in self-efficacy for online instruction might not imply a negative outcome. Two important points need to be considered. First, self-efficacy is a subjective judgment of an individual’s capability rather than an evaluation based on objective criteria, so professional development is still needed to improve the effectiveness of online teaching, especially regarding the declinations in students’ academic performance (Orlov et al., 2021). Second, low levels of self-efficacy have been reported to inspire individuals to be more reflective and looking for strategies to improve their capabilities, and this is especially important considering the vulnerability of self-efficacy in its early stages (Bandura, 1997).

This study confirmed the moderation effects of adaptability on changes in self-efficacy for both online instruction and technology applications. This finding resonates with the suggestion of Martin et al. (2012) to investigate the effects of adaptability in online teaching contexts but contradicts the incapability of adaptability with moderating OTSE changes among school teachers (Ma et al., 2021, 2022). This disparity might be because school teachers tended to rely on physical teaching, considering students’ limited online learning experience and easiness of losing attention at school, compared to university students, which reduces the tendency to increase their OTSE even if teachers could adapt to online teaching. In addition, the study confirmed the significant positive moderation of administration quality amidst COVID-19 on changes in both OTSE domains. This finding might be explained, as reported by Loughland and Alonzo (2018), by the level of support provided by schools to assist their teachers in adjusting to the constant challenges associated with online teaching, by providing essential resources and support. It is also aligned with Ma et al. (2021, 2022), who found that school teachers reported the school administration as one of the influential factors in their teaching self-efficacy. Also, it is possible that high-quality administration, including providing technical support and ensuring online teaching disciplines, as pointed out by Verma et al. (2020), could reduce the challenges perceived by teachers, which improved their self-efficacy for teaching.

Conclusion

The current study reported that Chinese university teachers failed to increase their self-efficacy for online instruction when teaching online due to Covid-19-related restrictions, but an improved self-efficacy for technology application during this same period. Adaptability only moderated change in self-efficacy for online instruction, and administration quality moderated the changes in both domains of OSTE.

Certain implications could be drawn for university teachers’ professional development and future research. First, universities need to pay more attention to improving the online instructional strategies of their staff and guiding them to improve online learning outcomes amidst dramatic transmissions of teaching formats associated with Covid-19. Second, purposely designed professional developments on university teachers’ adaptability are needed because adaptability tends not to increase without such, which could prohibit the growth of OTSE. Third, the university should ensure its administrative effectiveness in the face of unanticipated challenges like COVID-19 to assist teachers in developing a robust OSTE. Also, future research could further advance knowledge on this topic by collecting online teaching-related variables based on samples that have larger sizes and are more representative, by examining the effects of professional development projects that were designed to improve online teaching capability on OTSE by applying more advanced models based on latent variables, and by including observational data to improve the accuracy of the results.

Certain limitations exist in the study. First, the snowball data collection method could not ensure sample representativeness. Second, the measurement of OTSE at the beginning of COVID-19 relied on participants’ memory, which might have caused the inaccuracy of the present findings. Third, the multilevel modeling in the present study is based on observed variables and failed to utilize more advanced latent growth models, which might have increased the possibility of measurement errors. Four, only self-reported data was collected, which calls for data including class observation to triangulate its trustworthiness. Despite this, this research provided innovative evidence on OTSE changes in university teachers and its moderators amidst Covid-19.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

https://osf.io/7zuny/?view_only=fdb6fa741104427089885576fecde150

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The work was supported by the Jiangsu Provincial funding for educational science (C/2022/01/08), Funding for social sciences in Jiangsu Higher Education (2022SJYB2047).