Implementers, designers, and disseminators of integrated STEM activities: self-efficacy and commitment

ABSTRACT

Background

Previous research has revealed that teaching self-efficacy could play a critical role in engaging students in integrated STEM education; however, teachers’ multiple identities in STEM education (i.e. implementers, disseminators, and designers) and their commitment with respect to the different identities have not been considered and examined.

Purpose

This study aimed to investigate STEM teachers’ self-efficacy and commitment as implementers, disseminators, and designers, and to explore the relationships between teachers’ self-efficacy and commitment with respect to the three identities.

Sample

629 secondary STEM teachers completed a questionnaire that included the items of teachers’ background information and 46 items to measure their self-efficacy and commitment with respect to the three identities.

Methods

To compare the differences in teacher self-efficacy and commitment with respect to the three identities, repeated measures analyses of variance were used. A mixed-model analysis was conducted to examine the effects of both identity and experience with self-efficacy as a covariate on teacher commitment. Also, the structural equation modelling (SEM) method was employed to investigate the structural relationships within and between teacher self-efficacy and commitment with respect to the three identities.

Results

The results indicated that STEM teachers’ self-efficacy and commitment were influenced by their identities as well as their STEM teaching experience. STEM teachers’ self-efficacy and commitment to being implementers and designers were significantly higher than their self-efficacy and commitment to being disseminators. Additionally, the results of structural equation modelling indicated that teachers’ self-efficacy for the three identities was highly related, as was their commitment to the three identities. Self-efficacy of being disseminators had the largest impact on teacher commitment to being disseminators, designers and implementers.

Conclusion

The results highlight the importance of researching STEM teachers’ multiple identities, and suggest that STEM teachers’ different identities lead to different levels of self-efficacy and commitment.

KEYWORDS:

• Teacher identity
• commitment
• self-efficacy
• STEM education
• teacher education

Introduction

The integration of science, technology, engineering, and mathematics (STEM) has been gaining prominence with a view to preparing more students to meet the needs of the 21^st century. Teachers are one key factor influencing the success of students’ learning in STEM education (Kim 2019). However, teaching STEM using an integrated approach could be challenging to teachers (Margot and Kettler 2019). Rather than treating STEM as individual disciplines, using an integrated approach, or so-called integrated STEM (abbreviated as iSTEM), requires interdisciplinary or cross-disciplinary combinations of the individual disciplines (Kelley and Geoff Knowles 2016). For example, as suggested by Sanders (2009), iSTEM education involves learning and teaching between/among any two or more of the STEM disciplines. As most teachers were trained to teach only one discipline, they may experience difficulties in building connections across the STEM disciplines and be reluctant to provide students with opportunities for interdisciplinary learning (Kelley and Geoff Knowles 2016).

To further explain and understand their acceptance of and resistance to adopting iSTEM in classrooms, teachers’ beliefs and perceptions have been investigated (Margot and Kettler 2019; DeCoito and Myszkal 2018; Wang et al. 2011). Among these beliefs and perceptions, teachers’ self-efficacy has been identified as one of the key predictors of teaching practices in educational research (e.g. Cantrell, Young, and Moore 2003; Labone 2004; Tschannen-Moran, Hoy, and Hoy 1998; Tschannen-Moran and Hoy 2007; Naidoo and Naidoo 2021). According to Bandura’s theory, self-efficacy is defined as people’s belief in their capabilities to successfully accomplish a task under specific conditions (Bandura 1977) and is described as the ‘foundation of human agency’ (Bandura 2002, 3) due to its effect on behaviour and perseverance in the face of challenge (Bandura 1986). Hence, there have been studies exploring teachers’ self-efficacy for teaching robotics (Jaipal-Jamani and Angeli 2017), for teaching to execute the goals of STEM education (DeCoito and Myszkal 2018), and for knowledge required to teach STEM (Lee, Hsu, and Chang 2019). The common concerns of these studies on teachers’ self-efficacy are their teaching practices or knowledge required for teaching.

Although previous studies have investigated teachers’ self-efficacy for teaching iSTEM, teachers’ multiple roles and identities in STEM education (e.g. implementers or designers of iSTEM curricula) have not been considered or examined. Self-efficacy varies with task requirements in different identities, which refers to ‘persons hav[ing] as many identities [selves] as distinct sets of social relations in which they occupy a position and play a role’ (Stryker 2001, 227). That is, a teacher may hold different sets of efficacy beliefs as an implementer, disseminator, and designer of iSTEM activities. Additionally, people’s behaviours are in accord with the expectations attached to the identities and shaped by commitments that refer to ‘the degree to which persons’ relationships to others in their networks depend on possessing a particular identity and role’ (Stryker and Burke 2000, 286). El Nagdi, Leammukda, and Roehrig (2018) suggested a close connection between identities of STEM teachers and their commitments, and indicated that while teachers confront challenges in implementing iSTEM activities, their commitments to the identities are essential to sustain their teaching practices. However, there has been relatively little research on the relationships between STEM teachers’ self-efficacy, identities, and commitments.

This study highlights the multiple identities of STEM teachers. STEM teachers not only implement the ready-made curricula but also design and disseminate STEM materials. In order to foster teachers’ ownership and involvement in STEM education, the identities of being disseminators and designers, in addition to implementers, of integrated STEM activities could be crucial. Identities are associated with commitments and are assumed to generate primordial outcomes, such as self-efficacy (Jasso 2002). Teachers’ self-efficacy has been justified as a significant predictor of their commitment, regardless of the fact that the measures of self-efficacy are different (Chan et al. 2008). Teachers’ self-efficacy could also make an impact on commitment to teaching (Coladarci 1992) and their willingness to learn about innovative teaching approaches and strategies (Hoy and Spero 2005). Yet, what remains unanswered in the previous studies is the possible structural relationships between STEM teachers’ self-efficacy and commitment with respect to their different identities. Thus, the purpose of this study was to investigate STEM teachers’ self-efficacy and commitment as implementers, disseminators, and designers, and to explore the relationships between teachers’ self-efficacy and commitment with respect to the three identities by using the structural equation modelling method. The results of this study could advance current understanding of the evolution of identities of STEM teachers and contribute to the development of STEM teachers’ professional development programs.

Theoretical background

Teacher self-efficacy

Teacher self-efficacy can be defined as an ‘individual’s beliefs in their capabilities to perform specific teaching tasks at a specified level of quality in a specified situation’ (Dellinger et al. 2008, 752). Earlier research on teacher efficacy was based on the perspective of locus of control (Rotter 1966), such as teacher-perceived influence on student learning and outcomes (Gibson and Dembo 1984). Recent research, following Bandura’s (1977) social cognitive theory, has paid more attention to the teaching behaviours that lead to changes in student outcomes, such as ‘planning activities that accommodate the range of individual differences among my students’ (Dellinger et al. 2008, 764).

Zee and Koomen (2016) synthesized 40 years of research on teacher self-efficacy and concluded that teacher self-efficacy ‘shows positive links with students’ academic adjustment, patterns of teacher behaviour and practices related to classroom quality, and factors underlying teachers’ psychological well-being, including personal accomplishment, job satisfaction, and commitment’ (981). In their meta-analysis study, Chesnut and Burley (2015) further found that teachers’ self-efficacy beliefs significantly influenced their commitment to teaching with a moderate effect size (0.32). These findings imply that enhancing teacher self-efficacy could be an approach to increase teacher commitment to innovative teaching behaviours. Yet, little is understood about whether the relationships between teachers’ self-efficacy and commitment differ with respect to their different identities.

Teacher commitment

Drawing upon theories of organizational psychology, Buchanan (1974) defined commitment as a complex attitude including a sense of identification as one’s own goals and values, of involvement in one’s duty, and of loyalty and affection for identification and involvement. That is, organizational commitment is ‘a partisan, affective attachment to the goals and values of an organization, to one’s role in relation to goals and values, and to the organization for its own sake, apart from its instrumental worth’ (Buchanan 1974, 533). As the object of commitment is changed from organization to occupation, occupational commitment can be similarly conceptualized as ‘a psychological link between a person and his or her occupation that is based on an affective reaction to that occupation’ (Lee, Carswell, and Allen 2000, 800). However, organizational commitments may differ from or conflict with occupational commitments. That is, teachers’ commitment to their schools and professional communities (organization) may be different from their commitment to their students and subject areas (occupation). These different commitments result in various patterns of teaching behaviors (Firestone and Rosenblum 1988). The moral nature of occupational commitment links itself to the level of intrinsic motivation to work in a particular career role (Hackett, Lapierre, and Hausdorf 2001). Teaching as an occupation is ‘an identifiable and specific line of work’ (Lee, Carswell, and Allen 2000, 800). In this study, we focus on teachers’ commitment to iSTEM teaching.

Following Buchanan’s ideas on commitment, we define teacher commitment as a desire to teach and value this profession. This definition also echoes the affective and normative attributes of commitment to an organization. Affective commitment is referred to as ‘the desire to maintain membership in an organization’ (Meyer and Allen 1991, 74), whereas normative commitment can be viewed as ‘a moral obligation to engage in a mode of conduct reflecting loyalty and duty in all social situations in which he has a significant personal involvement’ (Wiener 1982, 423). Day, Elliot, and Kington (2005) further found that teacher commitment is a core and stable element of teachers’ professional identity and involves a cluster of values which drive commitment. Together these previous studies on commitment suggested a close connection between teacher commitment and their identity.

Teacher identities: implementer, designer, and disseminator

Identity theory developed by Stryker, Burke, and their colleagues have focused on two aspects of identity (Stryker and Burke 2000): the linkages of social structures with identities and the internal process of self-verification. While ‘persons are seen as living their lives in relatively small and specialized networks of social relationships, through roles that support their participation in such networks’ (Stryker and Burke 2000, 285), identities can be referred to as ‘the meanings that individuals hold for themselves-what it means to be who they are’ (Burke 2003, 196) in a network or group where individuals can play different social roles when interacting with other participants. On the other hand, identities can also be considered as internalized role expectations attached to positions occupied in a network, and a sense of identity reflects commitment to those role expectations (Stryker and Burke 2000). In schools, teachers participate in a variety of activities such as enacting a curriculum, developing curriculum materials, transforming the materials into practice, and introducing new curricula or lesson plans to their colleagues. These activities allow teachers to build different social networks and engage in various interactions with students, parents, fellow teachers, and school administrators. By playing different roles in these networks, teachers could possess multiple identities, and their commitment to these identities may vary.

When teachers participate in curriculum reform and adopt innovative pedagogies, such as iSTEM education, how do teachers design, implement, and disseminate curriculum materials? Two approaches have been identified to conceptualize the processes: remote control and mutual adaptation (Brown 2009; Remillard 2005). On one hand, the remote-control approach suggests that curriculum materials are designed to be implemented by teachers in the same way and disseminated in line with reform goals and principles. In this approach, teachers may take the roles of implementers and disseminators, either voluntarily or by appointment, and follow the reform goals without transforming the materials or adapting their practices. On the other hand, the mutual-adaptation approach proposes that curriculum materials are adapted by teachers for the needs of particular contexts. This approach highlights a dynamic interplay between teacher and curriculum materials; curriculum implementation could involve re-designing curriculum materials based on teachers’ interpretations of the curriculum (Taylor 2013; Brown et al. 2009) and teachers’ roles of implementer and designer may be inseparable.

The two approaches suggest the multiple identities teachers could develop while adopting curriculum innovations. In practice, when teachers enact a reform-based curriculum, their enactment may or may not be fully aligned with the intended curriculum and the reform vision. In addition to being implementers, teachers may create, revise, and transform curriculum activities as curriculum designers in order to realize their interpretation of the curriculum. Additionally, as a reform-based curriculum is promoted, experienced teachers may be invited or volunteer to be seed teachers or workshop instructors to spread new ideas about the curriculum reform, share their interpretation of the curriculum, and distribute the instructional materials. Therefore, in STEM education, teachers as designers can individually or collectively design innovative curricula and activities for their own use or for use by others (Pepin et al. 2019), while teachers as implementers can enact an interdisciplinary curriculum in classrooms, and teachers as disseminators can act as facilitators or leaders to help to scale up the implementation of iSTEM curricula (Goos, Bennison, and Proffitt-White 2018). Implementers, designers, and disseminators are important identities for teachers in STEM education. There exists a research need to understand how teachers view these three identities and how the identities interact with their self-efficacy and commitment.

The current study

In this study, we aimed at investigating STEM teachers’ self-efficacy and commitment as implementers, disseminators, and designers of iSTEM activities, and at examining the structural relationships between teachers’ self-efficacy and commitment with respect to the three identities. The study was guided by the following questions.

1. Do secondary school STEM teachers show different levels of self-efficacy with respect to the three identities? Does their experience of iSTEM teaching influence their self-efficacy?

2. Do secondary school STEM teachers show different levels of commitment with respect to the three identities? Does their experience of iSTEM teaching influence their commitment?

3. Do the relationships between secondary school STEM teacher self-efficacy and commitment differ with respect to the three identities?

4. What are the structural relationships within and between secondary school STEM teachers’ self-efficacy of and commitment with respect to the three identities?

Methods

Context of STEM education in Taiwan

This study was conducted in Taiwan where STEM education has been highly promoted by the government to improve the science literacy of all citizens, to train STEM talents, and to enhance national competitiveness. For example, the High-Scope Project funded by the Ministry of Education (2017) has encouraged science, mathematics, and technology teachers at the secondary school level to collaborate with university researchers to develop inquiry-based or project-based STEM curricula. Additionally, according to the latest curriculum guidelines (Ministry of Education 2018), all senior high schools should offer an interdisciplinary science course of ‘Inquiry and Practice’ that is a two-credit-hour course for two semesters and mandatory for students in the science track. STEM teachers are expected to plan, design, and implement the interdisciplinary science course based on the needs and interests of their students. To enhance teachers’ professional competences in integrated STEM education, professional development programs and workshops have been conducted nationwide, and participating teachers can receive credentials after finishing the programs or workshops. Some teachers who had experience of iSTEM education or who had performed well during professional development activities were recruited as workshop facilitators or instructors and helped disseminate and scale up the curriculum.

Participants

This study employed a stratified sampling method to collect data from secondary school teachers of science, technology, and mathematics-related subjects (grades 7 to 12) in Taiwan. After teachers of special schools or vocational schools were excluded, the population was divided into four groups (northern, central, southern, as well as eastern and outer Islands) according to the Taiwan region as the first stratum. In addition, the type of school was applied as the secondary stratum resulting in three subgroups, namely, junior high school (grades 7 to 9), senior high school (grades 10–12), and high school (grades 7 to 12).

A total of 705 questionnaires were collected, among which 52 were invalid and thus deleted, resulting in an aggregate of 653 valid questionnaires. After discarding 24 valid cases because of missing data in one of the observed items, we had 629 left in the sample. That is, data from 96.3% of the complete cohort were used. Among these valid questionnaires, 201 teachers reported that they had iSTEM teaching experience and either offered STEM courses or instructed STEM-related learning activities, and 428 teachers had no experience of iSTEM teaching.

Prior to analysing the data to achieve the study purpose, we split the data into two subsets. One third of the sample (n = 208) was used to validate the questionnaire and examine the structure of the factors, while two thirds of the sample (n = 421) was used to examine the structural relationships between factors. The participants were randomly assigned to one of two subsets for different analytical methods.

Measures

The questionnaire consisted of two parts. The first asked for the background information of the STEM teachers, such as the teacher’s demographics, their total years of teaching, and their iSTEM teaching experience. In the second part, the participants were asked to respond to 46 items of teacher self-efficacy and commitment on 5-point Likert scales ranging from 1 (strongly disagree) to 5 (strongly agree).

As shown in Tables 1, 28 items were used to measure teacher self-efficacy (SE) of being implementers, designers, and disseminators, which were hypothesized as the three latent factors of SE-implementer, SE-designer, and SE-disseminator. The items were adopted and revised from Riggs and Enochs (1990). Additionally, to measure the affective and normative attributes of commitment, we revised items from Meyer, Allen, and Smith (1993) and developed 18 items to measure teacher commitment (TC) to being implementers, designers, and disseminators, which were hypothesized as three latent factors of TC-implementer, TC-designer, and TC-disseminator.

Table 1. Items measuring the latent factors of teachers’ self-efficacy (SE) and commitment (TC).

Items	Content
SE-Implementer
SE_Q01	I can evaluate students’ performance in STEM teaching activities
SE_Q02	I can guide students to engage in STEM teaching activities
SE_Q03	I can instruct students in how to properly participate in STEM activities
SE_Q04	I can provide feedback to students based on their performances in STEM teaching activities
SE_Q05	I can analyse students’ STEM learning outcomes
SE_Q06	I can choose suitable STEM teaching activities according to the course objectives
SE_Q07	I can apply different teaching strategies to motivate students’ engagement in STEM activities
SE_Q08	I can choose appropriate teaching media and tools to present STEM content
SE_Q09	I can clearly explain the learning goals of STEM activities to students
SE_Q10	I can properly arrange learning contexts for STEM activities
SE-Designer
SE_Q11	I can design STEM teaching activities according to the learning goals
SE_Q12	I can design different STEM teaching activities based on students’ diverse abilities
SE_Q13	I can design STEM teaching activities by utilizing different teaching models
SE_Q14	I can design assessments to evaluate students’ performance after STEM teaching activities
SE_Q15	I can learn how to design STEM teaching activities
SE_Q16	When planning STEM teaching activities, I can anticipate the difficulties that may be encountered in the implementation process
SE_Q17	I can generate some ideas about STEM teaching activities
SE_Q18	I can design STEM teaching activities to promote students’ learning
SE-Disseminator
SE_Q19	I can explain the design principles of STEM teaching activities
SE_Q20	I can serve as a lecturer for STEM professional development programs
SE_Q21	I can demonstrate STEM teaching activities for other teachers to observe
SE_Q22	I can help other teachers understand the design of STEM teaching activities
SE_Q23	I can assist other teachers in teaching STEM activities
SE_Q24	I can explain the teaching rationale of STEM education to other teachers
SE_Q25	I can explain the benefits of STEM education to other teachers
SE_Q26	I can explain how STEM teaching activities can help students’ learning
SE_Q27	I can demonstrate STEM teaching activities to other teachers
SE_Q28	I can help other teachers plan their STEM teaching
TC-Implementer
TC_Q01	I would like to be a professional in the implementation of STEM teaching activities
TC_Q02	I feel a responsibility to the profession in the implementation of STEM activities
TC_Q03	I want to be able to implement STEM teaching activities
TC_Q04	I think it’s important to be able to implement STEM teaching activities
TC_Q05	I am proud to be in the profession of implementing STEM teaching activities
TC_Q06	I identify with the profession of implementing STEM teaching activities
TC-Designer
TC_Q07	I would like to be a professional in the design of STEM teaching activities
TC_Q08	I feel a responsibility to the profession in the design of STEM teaching activities
TC_Q09	I want to be able to design STEM teaching activities
TC_Q10	I think it’s important to be able to design STEM teaching activities
TC_Q11	I am proud to be in the profession of designing STEM teaching activities
TC_Q12	I identify with the profession of designing STEM teaching activities
TC-Disseminator
TC_Q13	I would like to be a professional in the dissemination of STEM teaching activities
TC_Q14	I feel a responsibility to the profession in the dissemination of STEM teaching activities
TC_Q15	I want to be able to disseminate STEM teaching activities
TC_Q16	I think it’s important to be able to disseminate STEM teaching activities
TC_Q17	I am proud to be in the profession of disseminating STEM teaching activities
TC_Q18	I identify with the profession of disseminating STEM teaching activities

Data analysis

To establish the construct validity of the latent factors, the first subset of the sample (n = 208) was used for exploratory factor analyses in order to extract interpretable scales from the Likert-type responses and to obtain internal-consistency reliability estimates. The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were performed to determine whether the sample was appropriate for such analysis. The decision regarding the number of factors to retain was based on the criteria of eigenvalue greater than one.

To address research questions 1 to 3, the whole sample was used to generate descriptive statistics and to compare the differences in teacher self-efficacy and commitment with respect to the three identities. The data were analysed using SPSS 25. For the first and second research questions, repeated measures analyses of variance (ANOVAs) were employed to compare the differences. Regarding the third research question, a mixed-model analysis method was conducted to examine the effects of both identity and experience with self-efficacy as a covariate on teacher commitment. The significant level was set as .01. All reported effect sizes were partial eta squared and have been interpreted as follows: η2 < .06 small effect, η2 .06–.14 medium effect, and η2 ≥ .14 large effect (Cohen 1973).

To answer the fourth research question, the first subset of the sample (n = 208) was examined by the structural equation modelling (SEM) method to determine whether a one-order or two-order structure of factors was appropriate for the data (Figure 1). After the structure was selected, another subset (n = 421) was used for confirmatory factor analysis with the SEM method to test the structural relationships within the three latent variables of either teacher self-efficacy or teacher commitment, and the structural relationships between the latent variables of teacher self-efficacy and commitment. All analyses were conducted using AMOS (version 26.0).

Figure 1. One-order factor structure (a) and two-order factor structure (b).

Regarding the SEM methods used, because some of the observed variables were non-symmetrical, unweighted least squares factor extraction was used to estimate the parameters (Joreskog and Sorbom 1984). We utilized the goodness-of-fit index (GFI), the adjusted goodness-of-fit index (AGFI), the normed fit index (NFI), and the root mean square residual (RMR) to compensate for the sensitivity. By looking at the variances and covariances accounted for by the model, GFI showed how closely the model replicated the observed covariance matrix. GFI was not sensitive to sample size, and AGFI was adjusted to take into account the degrees of freedom in the model. NFI assessed the model by comparing the χ ² value of the model to the χ ² value of the null model. RMR was an index of the degree of discrepancy between elements in the sample and the hypothesized covariance matrix. The values of GFI, AGFI, and NFI larger than .9 are indicative of a good fit, and RMR values (values range from 0 to 1.00) as low as 0.08 are deemed acceptable (Schumacker and Lomax 2016).

Exploratory factor analysis for latent factors

For the three factors of self-efficacy (i.e. SE-implementer, SE-designer, and SE-disseminator), the KMO measures of sampling adequacy index were found to be .95, .91, and .92 respectively. The results of Bartlett’s test of sphericity were all significant (Approx. Chi-Square (45) = 2664.26, p < .01 for SE-implementer; Approx. Chi-Square (28) = 1488.53, p < .01 for SE-designer; Approx. Chi-Square (45) = 2607.97, p < .01 for SE-disseminator), indicating that the sample was, indeed, appropriate for such an analysis. Additionally, the analysis of the 10 items for SE-implementer yielded one factor which explained 81.42% of the total variance. Similarly, the analyses showed that the eight items of SE-designer and the 10 items of SE-disseminator explained 71.51% and 76.81% of the total variance respectively. Finally, the factor loadings of the items were greater than .78.

In terms of the three factors of teacher commitment (i.e. TC-implementer, TC-designer, and TC-disseminator), the KMO measures of sampling adequacy index were .88, .86, and .87 respectively. The results of Bartlett’s test of sphericity were all significant (Approx. Chi-Square (15) = 1104.69, p < .01 for TC-implementer; Approx. Chi-Square (15) = 988.25, p < .01 for TC-designer; Approx. Chi-Square (15) = 1086.48, p < .01 for TC-disseminator). These test results revealed that the sample was appropriate for such an analysis. Also, the analysis of the six items for TC-implementer yielded one factor which explained 75.75% of the total variance. Similarly, the analyses showed that the six items of TC-designer and the six items of TC-disseminator explained 72.40% and 73.24% of the total variance respectively. Furthermore, the factor loadings of the items were greater than .65.

Results

Differences in teachers’ self-efficacy

The first research question was to compare STEM teachers’ self-efficacy with respect to the three identities. As shown in Table 2, the means of STEM teachers’ self-efficacy of being implementers and designers of integrated STEM activities with and without iSTEM teaching experience were generally greater than the neutral value (= 3). The results of the one-sample t-test showed that the mean differences of STEM teachers’ self-efficacy for the two identities of implementers and designers with and without experience were significantly larger than the neutral value at 0.01 level of significance. The mean difference for the identity of disseminators with experience (t (200) = .793, p = .429) did not significantly differ from the neutral value while the mean difference for the identity of disseminators without experience (t (427) = −13.694, p < .001) was significantly smaller than the neutral value.

Table 2. Means and standard deviations of self-efficacy and commitment for the three identities.

Variables	Implementer		Designer		Disseminator
	M	SD	M	SD	M	SD
With experience (n = 201)
Self-efficacy	3.62	0.65	3.58	0.65	3.04	0.83
Commitment	3.31	0.75	3.31	0.76	3.12	0.76
Without experience (n = 428)
Self-efficacy	3.14	0.79	3.12	0.73	2.49	0.76
Commitment	2.93	0.82	2.94	0.79	2.73	0.78

To further examine the influence of their iSTEM teaching experience in their self-efficacy for the three identities, we used repeated measures ANOVAs. Because the assumption of sphericity was violated (Approx. Chi-Square (2) = 220.631, p < .001), the corresponding Greenhouse-Geisser F value and degrees of freedom were used. The test result showed no significant effect for the interaction between experience and identity (F(1.542, 966.822) = 2.473, p = .099). The main effects of identity and experience were both significant (F (1.542, 966.822) = 496.699, p < .001, η2 = .442; F(1, 627) = 71.879, p < .001, η2 = .103). The results indicated that STEM teachers’ self-efficacy was influenced by their identities with a large effect size and by their experience with a medium effect size. According to the results of Bonferroni post hoc tests at a significant level of .01, STEM teachers’ self-efficacy of being implementers and designers was significantly better than their self-efficacy of being disseminators, and there was no significant difference between STEM teachers’ self-efficacy of being implementers and designers. STEM teachers who had experience of iSTEM activities perceived significantly more self-efficacy for implementing, designing, and disseminating STEM teaching than those who had not.

Differences in teachers’ commitment

To address the second research question, STEM teachers’ commitment with respect to the three identities and the influence of their experience of iSTEM teaching on their commitment were examined. The means of STEM teachers’ commitment to being implementers, designers and disseminators of integrated STEM activities with iSTEM teaching experience were generally greater than the neutral value (= 3) (Table 2). The results of the one-sample t test showed that the mean differences of experienced STEM teachers’ self-efficacy for the two identities of implementers and designers were significantly larger than 3 at the 0.01 level of significance, and the mean difference for the identity of disseminators with experience (t (200) = 2.137, p = .034) did not significantly differ from the neutral value. The mean difference for the identities of implementers and designers (ts (427) = −1.850, −1.404; ps = .063, .161) without experience did not significantly differ from the neutral value, while the mean difference for the identity of disseminators without experience (t (427) = −7.241, p < .001) was significantly smaller than 3.

To further investigate whether STEM teachers with and without experience were differently committed to being implementers, designers as well as disseminators, we used repeated measures ANOVAs. Because the assumption of sphericity was violated (Approx. Chi-Square (2) = 176.259, p < .001), the corresponding Greenhouse-Geisser F value and degrees of freedom were used. The test result showed no significant effect for the interaction between experience and identity (F (1.606, 1006.907) = .309, p = .685). The main effects of identity and experience were both significant (F (1.606, 1006.907) = 102.693, p < .001, η2 = .141; F (1, 627) = 34.391, p < .001, η2 = .052). The results revealed that STEM teachers’ commitment was influenced by their identities with a medium effect size and by their experience with a small effect size. Similar to the results of self-efficacy, the Bonferroni post hoc tests’ results showed that STEM teachers’ commitment to being implementers and designers was significantly higher than their commitment to being disseminators, and there was no significant difference between STEM teachers’ commitment to being implementers and designers. STEM teachers who had experience of integrated STEM activities significantly committed more to implementing, designing, and disseminating STEM teaching than those who did not have such experience.

Differences in the relationships between teachers’ self-efficacy and commitment

To investigate whether the relationships between self-efficacy and commitment differed in the three identities of being implementers, designers, and disseminators of integrated STEM activities, a mixed-model analysis was conducted. The analysis used commitment as the dependent variable, self-efficacy as a covariate, and identity, experience, and their interactions with self-efficacy as the independent variables. The results found no significant interactions among identity, experience, and self-efficacy (F(2, 689.848) = 2.869, p = .057) or between identity and experience (F(2, 722.459) = 2.193, p = .112). There was also no significant main effect of experience (F(1, 1806.085) = 0.472, p = .492) on commitment. These results suggested that the factor of experience could be deleted to test the relationships between self-efficacy and commitment. Then another mixed-model analysis was employed with commitment as the dependent variable, self-efficacy as a covariate, and identity and its interaction with self-efficacy as the independent variables. The analysis results showed a significant interaction between identity and self-efficacy (F(2, 673.592) = 20.562, p < .001) and indicated that STEM teachers’ self-efficacy had different effects on their commitment with respect to the three identities. The coefficients of self-efficacy for being implementers (b = .154) and designers (b = .160) were significantly less than the coefficient of self-efficacy for being disseminators (b = .287). The results revealed that STEM teachers’ self-efficacy could predict their commitment more for being disseminators than for being implementers and designers.

Structural relationships within and between self-efficacy and commitment

Before modelling the structural relationships within and between STEM teachers’ self-efficacy of and commitment to being implementers, designers, and disseminators, we used a subset of data (n = 208) to determine whether a one-order or two-order structure of factors was appropriate for the data (Figure 1). We specifically compared the fit of a one-order factor structure (with three correlated factors) to the fit of a two-order factor structure (with three sub-factors and one overarching higher order factor).

The results of the structural equation modelling analyses of teacher self-efficacy can be found in Table 3. As can be seen, both the correlated one-order factor structure and the two-order factor structure showed a good fit on all fit indices. Thus, the two structures could be considered to justify the structural relationships within the three factors of teacher self-efficacy. Similarly, the results of teacher commitment in Table 4 also suggested that both structures were appropriate for the data and could be used to examine the structural relationships within the three factors of commitment.

Table 3. Structural equation modelling analyses of teacher self-efficacy: Overall model fit.

Fit Index (n = 208)	One-order Factor Structure	Two-order Factor Structure
Goodness-of-fit index (GFI)	.996	.996
Normed fit index (NFI)	.996	.995
Root mean square residual (RMR)	.032	.033
Adjusted goodness-of-fit index (AGFI)	.996	.995

Table 4. Structural equation modelling analyses of teacher commitment: Overall model fit.

Fit Index (n = 208)	One-order Factor Structure	Two-order Factor Structure
Goodness-of-fit index (GFI)	.990	.990
Normed fit index (NFI)	.988	.988
Root mean square residual (RMR)	.060	.060
Adjusted goodness-of-fit index (AGFI)	.987	.987

Since the two structures showed a good fit, both were established to examine the relationships within teacher self-efficacy and commitment by using another subset of data (n = 421). Table 5 displays the modelling results and suggests satisfactory model fits. Considering the simplicity of the model, we chose the one-order factor structure and added paths from the factors of self-efficacy to the factors of commitment, as shown in Figure 2, to further examine the relationships between teacher self-efficacy and commitment. Additionally, based on the correlations among the six factors and the results of the mixed model analyses, the paths between self-efficacy of being disseminators to the commitment to being designers and implementers were added.

Figure 2. Modified structural relationships within and between self-efficacy and commitment. SE-Di = self-efficacy of being a disseminator, SE-De = self-efficacy of being a designer, SE-Im = self-efficacy of being an implementer, TC-Di = teacher commitment to being a disseminator, TC-De = teacher commitment to being a designer, TC-Im = teacher commitment to being an implementer.

Table 5. Structural equation modelling analyses of the structural relationships within and between teacher self-efficacy and commitment: Overall model fit.

Fit Index (n = 421)	One-order Factor Structure	Two-order Factor Structure	One-order Factor Structure with added paths
Goodness-of-fit index (GFI)	.988	.988	.993
Normed fit index (NFI)	.987	.987	.992
Root mean square residual (RMR)	.050	.051	.039
Adjusted goodness-of-fit index (AGFI)	.987	.987	.992

The actual factor loadings of self-efficacy and commitment items on the latent factors are shown in Table 6. All factor loadings generated by the covariance matrix exceeded 0.50. The estimated parameters of the overall model are displayed in Figure 2. The correlations between any two factors of self-efficacy and between any two factors of teacher commitment were all strong. The results in Figure 2 indicated that teachers’ self-efficacy for the three identities was highly related, as was their commitment to the three identities. Self-efficacy of being disseminators had the largest impact on teacher commitment to being disseminators, designers and implementers.

Table 6. Factor loadings of the confirmatory factor analysis (n = 421).

Self-efficacy						Commitment
Implementer		Designer		Disseminator		Implementer		Designer		Disseminator
Q01	.89	Q11	.86	Q19	.83	Q01	.87	Q07	.83	Q13	.85
Q02	.90	Q12	.81	Q20	.76	Q02	.86	Q08	.84	Q14	.87
Q03	.90	Q13	.84	Q21	.80	Q03	.86	Q09	.87	Q15	.88
Q04	.87	Q14	.87	Q22	.86	Q04	.86	Q10	.82	Q16	.84
Q05	.90	Q15	.80	Q23	.85	Q05	.87	Q11	.83	Q17	.89
Q06	.89	Q16	.81	Q24	.90	Q06	.76	Q12	.73	Q18	.56
Q07	.88	Q17	.85	Q25	.91
Q08	.89	Q18	.88	Q26	.88
Q09	.88			Q27	.83
Q10	.86			Q28	.84

Discussion

This study shows that both secondary STEM teachers’ self-efficacy and their commitment are influenced by their identities with large and medium effect sizes respectively. The results imply that different identities of STEM teachers can be related to different tasks which influence their self-efficacy (Dellinger et al. 2008) as well as to different role expectations which could reflect their commitment (Stryker and Burke 2000). This reminds teacher educators and educational reformers to notice the differences in teacher self-efficacy and commitment among the different identities.

This study found that STEM teachers’ self-efficacy of and commitment to being implementers and designers were significantly better than their self-efficacy of and commitment to being disseminators. There may be reasons for the results. First, in Taiwan, not all teachers participated in the dissemination of the curriculum materials, and only some experienced teachers were invited to be the workshop facilitators or took a leading role in accelerating the scaling-up of the new curriculum. Thus, most teachers might not feel qualified to be disseminators. Secondly, even for teachers who had experience of the three roles, being a disseminator might be challenging. It requires teachers to be familiar with the innovative instructional materials and understand the theories behind the instructional design so they could explain the design principles and rationale to others. Also, to convince their colleagues to use the materials, disseminators may need to identify the barriers to implementation and provide possible solutions for teachers. These dissemination tasks could involve a set of pedagogical knowledge and communication skills so the STEM teachers in this study appeared to feel less capable of performing and committing to these tasks.

Additionally, there was no significant difference between STEM teachers’ self-efficacy of being implementers and designers or between their commitment to being implementers and designers. The results imply that Taiwanese secondary STEM teachers could adapt reform-based curriculum materials according to the needs of their teaching contexts as they implement them. It also suggests that a mutual adaptation approach may be taken (Brown et al. 2009) when the participating STEM teachers in this study enacted the curriculum. However, designing curricula could involve revising, adapting, and creating iSTEM activities, and teachers may hold different efficacy beliefs about these design tasks. Further investigations of teacher self-efficacy of and commitment to being designers can explore teachers’ beliefs and perceptions of the complicated process of curriculum design.

This study also showed that STEM teachers’ self-efficacy could predict their commitment more for being disseminators than for being implementers and designers. The structural equation modelling analyses found that teacher self-efficacy of being disseminators had the largest impact on teacher commitment to being disseminators, designers and implementers. The results highlight the importance of researching teachers’ multiple identities, and the topic of how teachers play a role of disseminators deserves more research attention. The results also imply that encouraging teachers to engage in disseminating iSTEM curriculum materials might be effective in terms of increasing their commitment to the implementation and design of STEM curricula. Although it is not necessary for all teachers to be disseminators, as mentioned previously, some disseminating tasks might support teachers to develop pedagogical knowledge and communication skills, which in turn might enrich their understanding of the design and implementation of a curriculum. Future research may integrate some disseminating tasks or techniques into teacher education programs (e.g. explaining the design principles to other teachers and identifying possible challenges of implementation) to promote teachers’ professional development. Furthermore, although this study explored STEM teachers’ multiple identities, a question of how teachers may shift or progress from one identity to another remains unanswered. The progression of teachers’ identities could also be explored in future research.

Conclusion

This study provides a validated instrument to measure teachers’ self-efficacy and commitment for the three identities of being implementers, designers, and disseminators, and explored the relationships between teachers’ self-efficacy and commitment with respect to the three identities. The evidence reported here supports that (1) STEM teaching experience and identities influenced teachers’ self-efficacy and commitment; (2) secondary school STEM teachers perceived more self-efficacy and were more committed to the identities of being implementers and designers than to the identity of being disseminators; (3) STEM teachers’ self-efficacy could predict their commitment more for being disseminators than for being implementers and designers; (4) the correlations between any two factors of self-efficacy and between any two factors of teacher commitment were all strong, and teachers’ self-efficacy of being disseminators had larger impacts on their commitment to being designers and implementers than their self-efficacy of being designers and implementers.

This study not only contributes to the extension of studies on teacher self-efficacy and commitment by relating to their identity, but also sheds light on teacher professional development for iSTEM teaching. The recognition of the three identities for STEM teachers also extends current research from the relationships between self-efficacy and commitment (Chesnut and Burley 2015) to the relationships within the two constructs. While teachers may mobilize ‘occasional identities’ in response to new challenges and changing circumstances (Stronach et al. 2002), this research revealed the correlated relationships within teacher self-efficacy and commitment for the three identities and the hierarchical relationships between teacher self-efficacy and commitment for the three identities.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

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

Additional information

Funding

This study was supported by the Ministry of Science and Technology in Taiwan under MOST [107-2511-H-003-004-MY3, MOST 109-2511-H-003-015-MY3], and the “Institute for Research Excellence in Learning Sciences” of National Taiwan Normal University from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education in Taiwan.