What factors mediate the relationship between leadership for learning and teacher professional development? Evidence from meta-analytic structural equation modelling

Abstract

Leadership and teacher professional development are two of the most significant factors influencing student outcomes. This study adopted leadership for learning as its theoretical framework to investigate the effect of leadership on teachers’ engagement with teacher professional development, and the mediating roles of teacher self-efficacy and collective teacher efficacy in this relationship. With this purpose, the study employed meta-analytical structural equation modelling on data from 102 studies with a sample size of 100,211 participants from 23 countries. The results indicated that leadership for learning had a ‘large’ effect on teacher professional development, teacher self-efficacy and collective teacher efficacy. Similarly, a significant relationship was found between teacher professional development, teacher self-efficacy and collective teacher efficacy as well as between teacher self-efficacy and collective teacher efficacy. These results supported both the study hypothesis and the results evident in the literature.

Keywords

Leadership for learning teacher professional development teacher self-efficacy collective teacher efficacy MASEM

Introduction

Both leadership and teacher professional development (TPD) have been central to discussions regarding school improvement and effectiveness (Hallinger and Heck, 2010; Robinson et al., 2008; Tran et al., 2022).

Considering teachers are key to enhancing student outcomes and school effectiveness (Coppe et al., 2024), TPD is considered one of the most significant means of enhancing student outcomes (Darling Hammond et al., 2017; Yoon et al., 2023), and TPD impact is enhanced by school leadership (Patton et al., 2015).

Leadership also has a significant impact on student outcomes (Robinson et al., 2008), and, for example, it has been shown that the differences between a below-average principal compared to an above-average principal can be as much as three months of learning in mathematics and reading, with this being comparable in size to teacher effects on student learning (Grissom et al., 2021).

Instructional (IL), transformational (TL) and distributed/shared (DL) forms of leadership have been prominent in the educational leadership field (Hallinger et al., 2020). However, rather than relying on one conception, principals often employ a mix of leadership practices instead of a single leadership style (Ahn et al., 2021; Bowers, 2020; Day et al., 2016; Hallinger, 2011). Further, many argue that school leadership is so complex that it requires the shared practice of leadership rather than principal leadership alone, acknowledging, in the case of one distributed leadership view, that leadership emerges from the interaction of the leader, the followers and the context (Spillane et al., 2001). Amidst these debates, leadership for learning (LL) has emerged as an integrated perspective encompassing IL, TL and DL characteristics (Daniëls et al., 2019; Hallinger, 2011; Gurr, 2019).

The current study adopted the LL approach as its theoretical framework to investigate the leadership effect on TPD. We preferred to employ the LL framework because it allows for testing the effect of a comprehensive range of leadership practices on teachers’ continuous learning and development and fits with contemporary views of educational leadership (Tan et al., 2022). Also, including single models in leadership studies as if they were separate and distinct bodies of literature may limit our ability to understand how leaders actually improve teaching/learning (Liu et al., 2022). Additionally, the LL framework focuses on not only student learning but also teachers’ learning to enhance the overall IL capacity of the school (MacBeath and Townsend, 2011).

The current study also employs a mediation effect model in which the mediation of teacher self-efficacy (TSE) and collective teacher efficacy (CTE) in the LL–TPD relationship are investigated. The effect of leadership on learning occurs mostly indirectly over multiple interconnected variables (Day et al., 2016; Robinson et al., 2008). Therefore, the inclusion of possible mediators into the analysis could help build a better understanding of the paths from LL to TPD (Ahn and Bowers, 2024). In addition, adult learning literature proposes that TPD activities do not only contribute to teachers’ particular knowledge and skills but also produce added benefits regarding teachers’ psychological states such as increased efficacy beliefs (Yang, 2020). Therefore, including TSE and CTE as mediators could offer better insights into the relationship between LL and TPD.

Consequently, the current study aims to reveal a more global understanding of the LL–TPD relationship as well as the mediating roles of TSE and CTE as two significant components related to both leadership and teachers’ development (Goddard et al., 2015; Thien et al., 2021). With this purpose, the study employs meta-analytical structural equation modelling, which tests these relationships by combining correlation coefficients collected from the relevant literature. Synthesising results from a wide array of studies, the findings can contribute significantly to both the theory and practice of LL for enabling effective TPD.

Theoretical framework and the hypotheses

Leadership for learning and teacher professional development

A reaction to the limitations of single-model leadership has been the development of LL views (Marks and Printy, 2003; Bush, 2013). Typically, the LL views encompass IL, TL and DL practices which collectively help build school capacity for learning and achievement (Ahn and Bowers, 2024; Daniëls et al., 2019), and they focus on the ‘means or paths through which leadership achieves improvement in teaching and learning’ (Hallinger and Heck, 2010: 657).

LL is not an easily defined concept, and there have been three major approaches to its development (Author A). The first has been to combine existing approaches. The LL view of Marks and Printy (2003) combined IL and TL They suggested that IL was more effective when practiced as a shared endeavour, and that TL was essential to support this collaborative process. The second is to build on an existing approach. Hallinger (2011, 2018) took his earlier IL view (Hallinger and Murphy, 1985) and expanded it, adding context influences and leader characteristics, emphasising people development and providing a more detailed explanation of the direction and academic structures and processes of the earlier IL model. Another model like this is that of Murphy et al. (2007) who described LL under eight major dimensions: a vision for learning, IL programmes, curricular programmes, assessment programmes, communities of learning, resource acquisition and use, organisational culture and advocacy. Daniëls et al. (2019) showed how these dimensions mapped to key conceptions of IL, TL, DL and shared leadership, and to characteristics of effective school principals. The third approach to LL is to develop a new view from the ground up. The LL project (www.educ.cam.ac.uk/networks/lfl) is an example. Formed in 2001, and involving seven country contexts in an empirical research program, it developed a model that had four common framing values (LL; democratic values, critical friendship and moral purpose), and four tiers to represent leadership from students, teachers, senior managers and communities of learners. The LfL model views leadership as an activity that can be exercised by anyone, and learning applies to all. The leadership actions are guided by five principles at the top of the model: focusing on learning; sharing leadership, engaging in dialogue, sharing accountability and creating favourable learning conditions.

MacBeath and Townsend (2011) suggest that the difference between IL leadership and LL, is that the focus of IL leadership is on leadership, whilst the focus of LL is on learning. Combining the goal setting, role modelling, supervising and evaluation components of IL, the school-capacity-building tenets of TL through promoting teacher commitment and growth for achieving shared goals, and enabling collaborative decision-making and agency within the school, LL has strong potential for building a professional community of teachers who systematically dedicate time and energy to their professional development (Ahn et al., 2021; Harris and Jones, 2017; Nawab and Quraishi, 2024; Townsend, et al., 2020). With its learning-centered nature, LL builds a learning vision at school, provides learning support, manages the learning program and models the way for effective learning (Liu et al., 2016). LL also aims to facilitate teachers’ engagement with professional learning and turn schools into communities of professional learning to eventually improve student learning (Er, 2024; Hallinger et al., 2018). Therefore, LL closely aligns with TPD.

TPD refers to experiences that enhance teacher knowledge and skill (Patton et al., 2015). It is ‘both an obligation and an opportunity, serving as a forum for change and for confirmation of current practice’ (Langelaan et al., 2024). Current approaches to TPD emphasise a job-embedded, dynamic and interactive process that is guided by teachers’ individual needs and aspirations (Bergmark, 2023) and which can help teachers to adapt their teaching to rapid societal change (Coppe et al., 2024). High-quality TPD is teacher-driven, of sustained duration, collaborative and content-specific (Asterhan and Lefstein, 2024; Darling-Hammond et al., 2017; Desimone and Garet, 2015). Effective TPD not only provides teachers with updated ‘pedagogical strategies, an understanding of how to use educational resources and knowledge of the content they teach’ (Morris et al., 2017: 825) but also facilitates teachers’ awareness of their various responsibilities in and out of the classroom (Powers et al., 2016). Importantly, for this paper, the benefits of TPD are enhanced by leadership, such as LL, which enables both formal (e.g. mentoring programmes, seminars or workshops) and informal learning mechanisms (e.g. peer teaching, classroom observations or shared planning) within the school (Hallinger and Liu, 2016; Shal et al., 2024; Tran et al., 2022).

Building on this prior evidence, we promote the following hypothesis:

H1. LL has a positive direct effect on TPD.

Teacher self-efficacy and collective teacher efficacy

TSE refers to ‘individual teachers’ beliefs in their own abilities to plan, organise, andcarry out activities required to attain given educational goals’ whilst collective teacher efficacy refers to ‘teachers’ beliefs about the ability of …the faculty of teachers …to execute courses of action required to produce given attainments’ (Skaalvik and Skaalvik, 2007: 612–613). Teachers’ efficacy beliefs are crucially important in attaining educational goals and improving student learning (Dilekçi and Limon, 2022; Goddard et al., 2015; Stephanou and Oikonomou, 2018). Prior evidence indicated a significant link between TSE/CTE, and other variables such as teacher commitment (Tschannen-Moran and Hoy, 2001), job satisfaction (Skaalvik and Skaalvik, 2014), enthusiasm (Burić et al., 2020) and performance (Klassen and Tze, 2014).

As suggested by social cognitive theory, ‘efficacy beliefs determine how environmental opportunities and impediments are viewed’ by teachers (Bandura, 2006: 4), and thus influence their motivation and attitudes towards teaching. TSE and CTE are informed by four sources of information (Goddard et al., 2004): enactive mastery experiences (i.e. the experiences of their own), vicarious experiences (i.e. the observation of others’ experiences), verbal persuasion and affective/physiological states, and can manifest in three domains (Tschannen-Moran and Hoy, 2001): instruction, student engagement and classroom management.

Despite prior evidence that TSE and CTE are positively correlated (Skaalvik and Skaalvik, 2007; Stephanou and Oikonomou, 2018), there is research that suggests they have important impact differences. For instance, Skaalvik and Skaalvik (2007) found that TSE mediated the CEF–burnout relationship, whilst Guidetti et al. (2018) identified that TSE mediated the effect of CTE on perceived work ability. Similar studies were conducted which showed that their correlational and mediation analysis produced different results (Malinen and Savolainen, 2016; Stephanou and Oikonomou, 2018; Viel-Ruma et al., 2010). Based on these findings, including both TSE and CTE in the analysis was suggested to understand their combined and separate influence on teacher behaviour (Skaalvik and Skaalvik, 2019).

Research has also identified that the supervisory support provided by the principal and others through effective leadership has been associated with increased TSE and CTE (Bellibaş and Liu, 2017; Cansoy and Parlar, 2018; Liu et al., 2021) whilst the lack of supervisory or collegial support was found to have a negative effect on both (Skaalvik and Skaalvik, 2016). Leadership can influence TSE and CTE by various mechanisms; for instance, by clarifying performance standards, setting goals and inspiring a compelling vision (Hallinger et al., 2018), developing a common understanding of educational goals and values (Skaalvik and Skaalvik, 2019), modelling IL expectations, building healthy communications (Walker and Slear, 2011), creating a supportive learning environment and providing professional development opportunities (Ninković and Knežević Florić, 2024), coaching instruction and providing constructive feedback (Zheng et al., 2019). In brief, LL can contribute significantly to TSE and CTE ensuring that schools operate as a collective body to improve teaching and learning (Ahn et al., 2021). Building on this prior evidence, we hypothesise that:

H2. LL has a positive direct effect on TSE.

H3. LL has a positive direct effect on CTE.

TSE and CTE are also considered to have a close relationship with TPD. According to some researchers, TPD activities can strengthen TSE and CTE (Liu and Liao, 2019), particularly by providing enactive mastery experiences (Goddard, 2001; Tschannen-Moran and McMaster, 2009). On the other hand, teachers engage in professional development ‘to enhance the professional knowledge, skills and attitudes of educators so they might, in turn, improve the learning of students’ (Guskey, 2000: 16). Teachers’ belief in their individual (TSE) and collective (CTE) ability to change the course of instruction in support of student achievement could also influence their willingness to engage in TPD activities. For instance, Yoon et al. (2023) showed that TSE could be a significant factor in teachers’ willingness to transfer new knowledge and skills from TPD into actual classroom instruction Similarly, CTE helps develop social norms that could support teachers’ beliefs and attitudes towards professional learning as well as empower them to control the course of their TPD (Coleman, 1985). Furthermore, the more the teachers relate student achievement to IL quality, they tend to put effort into finding ways of improving their instruction (Skaalvik and Skaalvik, 2007). As such, TSE and CTE can determine whether and how teachers engage in TPD.

Building on this empirical ground, we propose the following four hypotheses:

H4. TSE has a positive direct effect on TPD.

H5. CTE has a positive direct effect on TPD.

H6. TSE mediates the relationship between LL and TPD.

H7. CTE mediates the relationship between LL and TPD.

Figure 1 shows the hypothesised model of the current study to test the direct, indirect and mediating relationships between LL, TPD, TSE and CTE.

Figure 1.

The hypothesised model.

Method

This study investigates the direct/indirect relationships between LL and TPD, with the mediating role of TSE and CTE using data from published studies. Therefore, we utilised meta-analytical structural equation modelling (MASEM) to measure the overall fit of hypothesised (Figure 1) and alternative models likely to emerge from the analysis of meta-analytic data (Jack, 2015). As this method combines the strengths of meta-analysis and structural equation modelling, it is considered to allow a powerful analysis to test casual relationships among the study variables (Jak and Cheung, 2020).

Data collection

We collected data for this study from studies indexed on Scopus and Web of Science. We used the following search string to collect data on 17 January 2024:

(“leader*” OR “leadership” OR “principal*”) AND (“teacher* professional development” OR “professional development” OR “teacher* development” OR “staff* development” OR “teacher* professional learning” OR “professional learning” OR “teacher* learning”)

The search terms were deliberately broad to capture a wide range of research about leadership in schools (e.g. IL, TL, DL, LL) that is focussed on teacher professional learning. The wildcard symbol, *, broadens the search to include terms that start with the preceding letters.

Our preliminary search yielded 916 studies: 515 from Scopus and 401 from WoS. We first removed the duplicates (397 studies) using the R program. From the remaining 519 studies, we eliminated 62 because they were not related to the relationship between the study variables. We skimmed the titles and abstracts of the remaining 457 studies with the following inclusion–exclusion criteria in mind:

Include – quantitative studies providing eligible data (i.e. the correlations between variables) for MASEM

Exclude – qualitative studies; conceptual/theoretical papers

During this stage, 355 studies were excluded: 127 for not providing eligible data; 124 for using qualitative methodology, and 104 for being conceptual/theoretical papers. Eventually, 102 studies were left to collect data for the current analysis. Whilst there are both limitations and delimitations to this process, and there may be worthwhile studies that have not been used because they are not reported in the journal databases used or have been eliminated because of the absence of suitable data for the MASEM process, the resulting 102 studies form a substantial knowledge corpus of studies that have results that can inform our hypotheses. The full list of studies used can be accessed from the Supplemental file associated with the article, or by contacting the authors. The 102 studies, whilst a partial knowledge representation of leadership in schools and teacher professional learning, is a list that is sufficiently robust in terms of quantity and quality to allow the hypotheses to be adequately tested.

Following the PRISMA protocol (Moher et al., 2010), the data extraction process is shown in Figure 2.

Figure 2.

The process of data collection and extraction.

Data analysis

Correlation coefficients from studies were used to calculate the overall effect size between LL, TPD, TSE and CTE. Fisher's Z was used as the effect size index, but it was later converted to Pearson's r to interpret the effect size. The random effects model was used to calculate overall effect sizes, and the results were interpreted according to Gignac and Szodorai (2016) scale. Accordingly, an r coefficient up to 0.1 indicates a ‘small’ effect, up to 0.2 a ‘medium’ effect, 0.3 and above a large’ effect.

The existence of variance between the effect sizes was measured using the heterogeneity test. The significant results (p < 0.05) from this test indicate the presence of heterogeneity. The size of the variance was determined using the I² index. According to Higgins et al. (2003), for the I² index up to 25% indicates low heterogeneity, whilst up to 50% indicates medium and up to 75% high heterogeneity. Bias analysis was also performed, and contour funnel charts were used to test whether any publication bias contributed to the calculated overall effect sizes (see Appendix). The significance of the asymmetry in the funnel plot was assessed using Egger's regression test (Egger et al., 1997) and rank correlation test (Begg and Mazumdar, 1994). The R platform (R Core Team, 2021) meta package (Schwarzer, 2022) was used to perform these analyses.

The model fits of the hypothesised and alternative models were assessed using Cheung's (2015a) two-stage structural equation modelling (TSSEM) method. First, the random effects model was used to create an overall covariance matrix using the correlation coefficients collected from the studies. This covariant matrix was then used to test the goodness of fit by establishing a structural equation model. TSSEM was performed using the metaSEM package (Cheung, 2015b) and lavaan (Rosseel, 2012).

In addition to the direct effects between variables, indirect effects were also calculated in the models. The significance of the indirect effects was tested using the likelihood-based confidence intervals method (Jack, 2015: 51). The fit indices of the models were evaluated based on various cut-off points. RMSEA ≤ 0.6 and SRMR ≤ 0.8 are interpreted as a ‘good’ fit (Hu and Bentler, 1999) whilst RMSEA and SRMR ≤0.10 indicate a poor fit (Kline, 2005; Tabachnick and Fidell, 2007). TLI and CFI ≥0.90 indicate ‘good fit’ whilst TLI and CFI ≥0.95 are interpreted as ‘excellent fit’ (Hu and Bentler, 1999; Tabachnick and Fidell, 2007). In addition, AIC and BIC indices were used to determine which model fits the data best. Accordingly, the model with smaller AIC and BIC values is better suited to the data compared to the competing model (Schermelleh-Engel et al., 2003).

Results

Overall effect size between LL, TPD, TSE and CTE

Overall effect sizes calculated according to the random effects model regarding the relationships between LL, TPD, TSE and CTE are given in Table 1.

Table 1.

Summary of the effects and heterogeneity measures.

Relations	K	N	Overall ES	95% CI		Heterogeneity		I ²
Relations	K	N	Overall ES	Lower	Upper	χ _(df)	p<	I ²
LL&TPD	56	63,416	0.456	0.395	0.512	4257.46(55)	0.0001	98.7%
LL&TSE	38	24,090	0.389	0.332	0.443	891.39(37)	0.0001	95.8%
LL&CTE	23	20,726	0.530	0.469	0.585	649.16(22)	0.0001	96.6%
TPD&TSE	16	16,281	0.537	0.437	0.624	743.69(15)	0.0001	98%
TPD&CTE	8	13,887	0.647	0.408	0.803	2715.49(7)	0.0001	99.7%
TSE&CTE	15	10,003	0.562	0.444	0.661	1580.13(14)	0.0001	99.1%

The results in Table 1 show that there is a significant ‘large’ relationship between LL and TSE (r = 0.389, 95% [0.332, 0.443]), TPD (r = 0.456, 95%[0.395, 0.512]) and CTE (r = 0.530, 95%[0.469, 0.585]). The relationships between TPD and both TSE (r = 0.537, 95%[0.437, 0.624]) and CTE (r = 0.647, 95%[0.408, 0.803]) are also ‘large.’ A ‘large’ relationship was also found between TSE and CTE (r = 0.562, 95%[0.444, 0.661]).

The heterogeneity tests confirmed that the heterogeneity for these six relationships was significant (p < 0.05) and ‘large’, which indicates that the variance between studies does not result from sampling error. For the funnel plots (see Appendix) Egger's regression test was significant for the relationship between TSE and CTE (p < 0.05), but not for other relationships (p > 0.05). On the other hand, the rank correlation test results were not significant for all binary relationships (p > 0.05). These results altogether indicate that the calculated overall effect sizes are not the product of publication bias.

Model-data fit of the models

To test the fit of the structural equation models, correlation matrices were created using the correlation coefficients collected from the studies. The analysis performed to determine the heterogeneity of the correlation matrices showed that these matrices were heterogeneous (χ²(145) = 7245.813, p < 0.0001). Therefore, in the first stage of MASEM, the random effects model was used to combine the correlation matrices The overall correlation matrices calculated in the first stage were also used in the second stage of MASEM to test the fit of the alternative models (model 1, model 2, model 3, model 4). The fit indices of the established models are given in Table 2.

Table 2.

Summary of the indices for the goodness-of-fit of the models.

Model	χ ² _(df)	p	RMSEA [95% CI]	SRMR	TLI	CFI	AIC	BIC
Model 1	42.426 (2)	0.000	0.015 [0.011, 0.019]	0.122	0.895	0.965	38.426	19.549
A-Model 1	4.538 (1)	0.033	0.006 [0.001, 0.012]	0.069	0.982	0.997	2.538	−6.901
A-Model 2	11.516 (2)	0.003	0.007 [0.004, 0.011]	0.080	0.975	0.992	7.516	−11.361
A-Model 3	2.227(1)	0.136	0.004 [0.000, 0.010]	0.034	0.994	0.999	0.227	−9.213
A-Model 4	6.978(1)	0.008	0.008 [0.003, 0.014]	0.079	0.969	0.995	4.978	−4.460

Regarding the hypothesised model in Figure 3, in which TSE and CTE mediate the relationship between LL and TPD, the chi-square test result (Table 2) was significant with two degrees of freedom (χ²= 42.426, p < 0.05). RMSEA showed a ‘good fit’ with a value of 0.015, 95% [0.011, 0.019] whilst SRMR indicated an ‘acceptable fit’ with a value of 0.122. Similarly, the CFI value of 0.965 showed an ‘excellent’ fit whilst the TLI value showed an almost ‘good’ fit. Whilst these results show that the model shows an acceptable fit, the SRMR value above the acceptable range shows that the model can be improved.

Figure 3.

The path diagram of the hypothesised model with parameter estimates and 95% CI.

As shown in Figure 3, LL had a significant effect on TPD (β = 0.16, 95% [0.013, 0.300]), whilst it had a large effect on TSE (β = 0.42, 95% [0.377, 0.473]) and CTE (β = 0.55, 95% [0.505, 0.605]). LL explains 31% of the variance in CTE and 18% of the variance in TSE. CTE has a significant (p < 0.05) ‘large’ effect (β = 0.49, 95% [0.368, 0.615]) on TPD. Similarly, TSE has a significant ‘large’ effect (β = 0.40, 95% [0.296, 0.499]) on TPD. TSE and CTE together explain 49% of the change in TPD. In this model, the indirect effect of LL on TPD was significant over both CTE (0.273, 95% [0.205, 0.338]) and TSE (0.169, 95% [0.125, 0.220]). These results show that both CTE and TSE are partial mediators in the LL–TPD relationship. The ANOVA test conducted to determine whether there is a significant difference between the path coefficients of these two mediator variables revealed no significant difference between them (p > 0.05).

Alternative models will now be explored and these are indicated in Table 2.

In Alternative Model 1 (A-Model 1) shown in Figure 4, a covariance relationship is built between TSE and CTE. As shown in Table 2, for this model the chi-square test result was significant with one degree of freedom (χ² = 4.538, p < 0.05). RMSEA showed a good fit with the value of 0.006, 95% [0.001, 0.012] and SRMR with the value of 0.069. Similarly, CFI indicated a perfect fit with a value of 0.997 and TLI with a value of 0.982. These results indicate that the model fits the data well.

Figure 4.

The path diagram of the alternative model 1 (A-Model 1) with parameter estimates and 95% CI.

Figure 4 shows that LL had a significant ‘large’ effect on CTE (β = 0.53, 95% [0.486, 0.583]) and TSE (β = 0.38, 95% [0.334, 0.435]), explaining the 29% of the variance in CTE, and 15% of the variance in TSE. CTE had a significant (p < 0.05) ‘large’ effect on TPD (β = 0.64, 95% [0.470, 0.831]) and a significant ‘medium’ effect on TSE (β = 0.20, 95% [0.001, 0.359]). TSE and CTE together explain 48% of the variation in TPD. The covariance established between TSE and CTE shows that there is a strong relationship between these two variables with a value of 0.307, 95% [0.208, 0.410]. The fact that the covariance is positive, significant (p < 0.05) and strong, shows that these variables tend to increase and decrease together. In this model, the indirect effect of LL on TPD over both CTE (0.342, 95% [0.252, 0.438]) and TSE (0.075, 95% [0.001, 0.143]) was significant (p < 0.05). These results indicate that both CTE and TSE are partial mediators in the LL–TPD relationship. The ANOVA test performed to determine whether there was a significant difference between the path coefficients of these two mediator variables revealed a significant difference (p < 0.05), which indicates that the indirect effect of LL on TPD over CTE is significantly higher than it is over TSE.

In Alternative Model 2 (A-Model 2) shown in Figure 5, LL and TPD together affect TSE over CTE. As seen in Table 2, the chi-square test result for this model was significant with one degree of freedom (χ² = 11.516, p < 0.05). RMSEA showed a ‘good fit’ with the value of 0.007, 95% [0.004, 0.011] and SRMR with the value of 0.08. Similarly, CFI indicated a perfect fit with a value of 0.992 and TLI with a value of 0.975. These results show that this model also fits the data well.

Figure 5.

The path diagram of the alternative model 2 (A-Model 2) with parameter estimates and 95% CI.

As shown in Figure 5, LL (β = 0.273, 95% [0.181, 0.356]) and TPD (β = 0.61, 95% [0.733, 0.901]) had a significant ‘large’ effect on CTE. Similarly, CTE had a ‘large’ effect on TSE (β = 0.64, 95% [0.392, 0.493]). Whilst LL and TPD together explain 49% of the variation in CTE, CTE explains 41% of the variation in TSE. The indirect effect of LL (β = 0.17, 95% [0.114, 0.233]) and TPD (β = 0.39, 95% [0.298, 0.476]) on TSE over CTE was ‘moderate’. These results indicate that CTE is a partial mediator variable for both LL and TPD. The ANOVA test performed to determine whether there was a significant difference between these two path coefficients revealed a significant (p < 0.05) difference, showing that the indirect effect of TPD on TSE was significantly higher than that of LL.

In Alternative Model 3 (A-Model 3) shown in Figure 6, a direct relationship was built from TPD to TSE. As seen in Table 2, the chi-square test result for this model was significant with one degree of freedom (χ² = 2.227, p < 0.05). Both RMSEA (0.004, 95% [0.000, 0.010]) and SRMR (0.036) showed a ‘good fit.’ Similarly, CFI (0.999) and TLI (0.994) indicated a ‘perfect fit’. These results show that the model fits the data very well.

Figure 6.

The path diagram of the alternative model 3 (A-Model 3) with parameter estimates and 95% CI.

As shown in Figure 6, both LL (β = 0.38, 95% [0.266, 0.480]) and TPD (β = 0.33, 95% [0.132, 0.542]) have a significant effect on CTE. Similarly, both CTE (β = 0.40, 95% [0.261, 0.534]) and TPD (β = 0.40, 95% [0.261, 0.534]) have a ‘large’ effect on TSE. Whilst LL and TPD together explain 36% of the variation in CTE, CTE and TPD explain 40% of the variation in TSE. The indirect effect of TPD on TSE over CTE (β = 0.13, 95% [0.052, 0.140]) is ‘moderate’. The ANOVA test performed to determine whether there was a significant difference between the direct and indirect effects of TPD revealed no significant (p > 0.05) difference, which indicates that TPD affected TSE both directly and indirectly.

The Alternative Model 4 (A-Model 4) in Figure 7 draws a direct relationship between LL to TSE. As presented in Table 2, the chi-square test result for this model was significant with one degree of freedom (χ² = 6.978, p < 0.05). RMSEA (0.008, 95% [0.003, 0.014]) and SRMR (0.079) showed a ‘good’ fit whilst CFI (0.995) and TLI (0.969) indicated a ‘perfect’ fit. These results show that the model fits the data well.

Figure 7.

The path diagram of the alternative model 4 (A-Model 4) with parameter estimates and 95% CI.

As shown in Figure 7, both LL (β = 0.22, 95% [0.117, 0.322]) and TPD(β = 0.65, 95% [0.487, 0.820]) has a significant large effect on CTE. Similarly, CTE has a large effect on TSE (β = 0.53, 95% [0.418, 0.649]). The direct effect of LL on TSE (β = 0.11, 95% [0.09, 0.200]) is ‘moderate’. LL and TPD together explain 60% of the variation in CTE whilst CTE and LL explain 35% of the variation in TSE. The indirect effect of LL on TSE over CTE (β = 0.344, 95% [0.251, 0.441]) was ‘large’. The ANOVA test performed to determine whether there was a significant difference between the direct and indirect effects of LL revealed a significant (p < 0.05) difference, which indicates that the indirect effect of LL on TSE is significantly higher than its direct effect.

We also assessed which model best fits the data by comparing the AIC and BIC indexes in Table 2. For the hypothesised model, AIC was 38.426 and BIC was 19.549. The AIC for A-Model 1 was 2.538 and the BIC was −6.901 whilst the AIC for A-Model 2 was 7.516 and the BIC was −11.361. For the A-Model 3, the AIC was 0.227 and the BIC was −9.213 whilst the AIC for the A-Model 4 was 4.978 and the BIC was −4.460. The model with lower AIC and BIC values is considered to fit the data better (Schermelleh-Engel et al., 2003). Accordingly, the model showing the best model-data fit was the A-Model 3, in which LL and TPD together affect TSE over CTE and TPD had a direct relationship with TSE. The second model with the best model-data fit was A-Model 1, in which TSE and CTE together mediated the relationship between LL and TPD.

Discussion

The current study investigated the relationships between LL, TPD, TSE and CTE. The results supported the first five hypotheses and indicated that LL had a ‘large’ effect on TPD, TSE and CTE. Similarly, a significant relationship was found between TPD, TSE and CTE as well as between TSE and CTE. The mediating role of TSE and CTE in the LL–TPD relationship was explored using a two-stage meta-analytical structural equation modelling (MASEM), confirming the last two hypotheses.

Interpretation and implications of findings

The current study suggested several alternative answers regarding the relationships among LL, TPD, TSE and CTE.

A reciprocal causality between TPD and TSE/CTE was found, suggesting that TPD enhances teachers’ efficacy beliefs both at individual and group-level, whilst these beliefs in turn facilitate their engagement in professional development. This finding is consistent with the reciprocal causality assumption of social cognitive theory, which also underpins the efficacy theory of Bandura (1989). The theory proposes that human behaviour results from the reciprocal interaction of socio-cognitive, behavioural and environmental conditions which influence each other bidirectionally (Wood and Bandura, 1989). This might suggest that TPD could enhance TSE and CTE by positively influencing the quality of teachers’ instruction, which in return facilitates teachers’ belief in successfully completing future tasks (Goddard, 2001; Tschannen-Moran and McMaster, 2009; Yoon et al., 2023). Furthermore, the TPD process enables teachers to learn together ‘by exchanging knowledge, sharing experiences and collectively searching for solutions to problems of practice’ (Moolenaar et al., 2012: 253), and this learning experience can enhance teachers’ ‘perceptions of their conjoint capability to foster student learning through concerted actions’ (Windlinger et al., 2020: 67)

Efficacy beliefs determine how teachers view environmental conditions and impediments (Bandura, 2006), and have a significant influence on teachers’ emotional and cognitive processes, choices and actions (Bandura, 1989; Caprara et al., 2006). Therefore, it is expected that higher levels of TSE and CTE facilitate teachers’ engagement in TPD because teachers believe that they can exercise influence on their work, and the TPD process enhances this. TPD also encourages emotional, cognitive and metacognitive reflection of teachers about their practice, and facilitates their resilience and expectation about student outcomes (Yang, 2020; Woodcock et al., 2022). In brief, the higher the TSE/CTE, the higher the teachers’ outcome expectancy estimates, so the stronger the effort they put into performing better (Bandura, 2006; Woodcock and Hardy, 2023; Yoon et al., 2023).

Our findings pinpoint the significant role of LL in helping teachers engage with TPD whilst also highlighting that school leaders could support TPD better by strengthening TSE and CTE through forming a shared vision and establishing challenging goals for student learning, supervising instruction, reinforcing collaboration and collegial leadership, building teacher capacity for development, and enforcing teachers’ identification with the group (Bellibaş and Liu, 2017; Daing and Mustapha, 2023; Liu and Hallinger, 2018). This line of research also indicates that LL can support teachers’ efficacy beliefs by creating a culture of collaboration and collegiality that allows for discretionary and shared decision-making (Daniëls et al., 2019; Leithwood et al., 2020; Liu et al., 2022; Thien et al., 2021). As suggested by Goddard et al. (2015), ‘teacher collaboration is key to the pathway from leadership to collective efficacy beliefs because it is the shared interactions among group members that serve as the building blocks of collective efficacy’ (504). They also underline that ‘from the perspective of social cognitive theory, leadership that establishes such norms also serves as a form of social persuasion that can positively influence collective efficacy beliefs’ (Goddard et al., 2015: 504).

Since LL conceptualises leadership as a reciprocal and shared process (Hallinger and Heck, 2010), it may create a professional community at schools that supports teachers’ agency to learn from each other and support each other's development (Ahn et al., 2021; Anderson and Oliver, 2022).

Although LL was positively associated with both TSE and CTE (as is also evident in reaserch on TL; Kurt et al., 2012), our findings indicated that the role of CTE was more significant in building a path from LL to TPD. As suggested by Goddard (2001), when teachers believe that the staff as a whole can achieve desired results, they also persist in their own personal efforts to achieve them. Thus, CTE could be a useful path to support both TSE and motivation to seek TPD (Kaya and Demir, 2022; Skaalvik and Skaalvik, 2019).

Our study also contributed to the discussions in the literature on whether TSE and CTE are closely linked or not (Cansoy and Parlar, 2018; Hayward and Ohlson, 2023; Stephanou and Oikonomou, 2018). Prior research regarding the relationship between TSE and CTE suggests contradictory results. Some indicated that TSE can predict CTE (Ninković and Knežević Florić, 2018) building on Bandura's statement that TSE ‘extends the conception of agent causality to people's beliefs in their collective efficacy to produce desired outcomes’ (p. 51). On the other hand, some found that CTE is an antecedent to TSE (Goddard, 2001; Skaalvik and Skaalvik, 2019), whilst some only indicated reciprocal causality between TSE and CTE (Cansoy and Parlar, 2018; Stephanou and Oikonomou, 2018). Our results revealed a strong one-way relationship between TSE and CTE, indicating that CTE has a stronger influence on TSE rather than vice versa. Furthermore, our findings suggested that the combination of these two constructs could help predict TPD better. Therefore, we support Skaalvik and Skaalvik (2019) assertion that studies investigating the role of efficacy beliefs on leader and teacher-level variables should consider including both constructs in their analysis.

The current study has some significant implications for policy and practice, particularly about LL and TPD. Leaders who facilitate CTE and TSE, perhaps by building a collaborative and collegial culture (Liu et al., 2021), will enhance the impact of TPD. They can also improve teachers’ motivation for development by acknowledging successful performance, setting inspiring goals and providing support (Goddard et al., 2015). Building on prior research that revealed a strong relationship between CTE, TSE and student achievement (Ross and Gray, 2006; Sun and Leithwood, 2017) as well as a close relationship between leadership and CTE/TSE (Leithwood et al., 2020; Ninković and Knežević Florić, 2018; Thien et al., 2021), our findings suggest that school leaders should put special effort in facilitating both TSE and CTE, acknowledging that teachers’ belief in their ability to improve their practices is key for teachers to ultimately transfer the knowledge, skills or insights into actual classroom instruction (Patton et al., 2015). This can be achieved, to a larger extent, by building a schoolwide capacity that can foster sustained learning and development of all school members (Liu et al., 2016).

The same suggestion goes for the policymakers who have a unique and powerful position to design an educational environment that supports TSE, CTE and TPD. As suggested by other scholars, productive TPD benefits from the active participation of teachers in determining their needs, the courses of action they can take, and the autonomy to transfer their new knowledge and skills into actual classroom teaching so they develop both personal and collective efficacy (Liu and Liao, 2019; Yang, 2020). In designing professional development activities, both policymakers and school leaders should seek more collaborative and reflective means of learning such as collaborative action research, problem-solving groups and team teaching, rather than one-size-fits-all designs such as standard workshops and seminars (Donohoo, 2017).

Limitations

As the current study employs the meta-analysis of correlational values provided by existing casual studies on the study variables, the results are bound up by the limitations of measures of these studies to a large extent. All included studies were from fully refereed journals, and this provides a good level of quality assurance. In addition, since LL is a broad concept of leadership, we included studies of IL, TL and DL in the analysis as they are regarded as the building blocks of LL framework. Future studies could consider evaluating the moderation effects of these leadership types in the LL–TSE/CTE–TPD relationship.

Supplemental Material

sj-xlsx-1-ema-10.1177_17411432241308461 - Supplemental material for What factors mediate the relationship between leadership for learning and teacher professional development? Evidence from meta-analytic structural equation modelling

Supplemental material, sj-xlsx-1-ema-10.1177_17411432241308461 for What factors mediate the relationship between leadership for learning and teacher professional development? Evidence from meta-analytic structural equation modelling by Turgut Karakose, David Gurr, Tijen Tülübaş and Sedat Kanadlı in Educational Management Administration & Leadership

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Turgut Karakose

David Gurr

Supplemental material

Supplemental material for this article is available online.

Author biographies

Turgut Karakose is a Professor of educational management and currently works at Kütahya Dumlupinar University, Türkiye. He is also the editor-in-chief of Educational Process International Journal (EDUPIJ). His main research interests include educational leadership and administration, school leadership, higher education, psychology and human behaviour. He has published extensively in leading international journals and also authored books and chapters on education/management.

David Gurr is a Professor of educational management and currently works at the University of Melbourne, Australia. He is also the senior associate editor of The Journal of Educational Administration. His main research interests include school leadership, school improvement, leadership development, school supervision, school review and educational accountability.

Tijen Tülübaş is an Associate Professor of educational management and currently works at Kütahya Dumlupınar University, Türkiye. She is also the arbitration and triage editor of the Educational Process International Journal (EDUPIJ). Her research interests include organisational behaviour, educational policy, leadership and higher education.

Sedat Kanadli is an Associate Professor of curriculum and instructional studies and currently works at Mersin University, Türkiye. His research interests include teacher education, teacher behaviour, student achievement, meta-analysis and structural equational modelling.

Appendix

References

Ahn

Bowers

(2024) Do teacher beliefs mediate leadership and teacher behaviors? Testing teacher self-efficacy's mediation role between leadership for learning and teacher outcomes. Journal of Educational Administration, 62(2), 197–222. DOI: https://doi.org/10.1108/JEA-12-2022-0227.

Ahn

Bowers

Welton

(2021) Leadership for learning as an organization-wide practice: Evidence on its multilevel structure and implications for educational leadership practice and research. International Journal of Leadership in Education 27(6): 1300–1351. DOI: https://doi.org/10.1080/13603124.2021.1972162.

Anderson

Olivier

(2022) A quantitative study of schools as learning organizations: An examination of professional learning communities, teacher self-efficacy, and collective efficacy. Research Issues in Contemporary Education 7(1): 26–51. Available at: https://eric.ed.gov/?id=EJ1344350.

Asterhan

CSC

Lefstein

(2024) The search for evidence-based features of effective teacher professional development: A critical analysis of the literature. Professional Development in Education 50(1): 11–23. DOI: https://doi.org/10.1080/19415257.2023.2283437.

Bandura

(1989) Human agency in social cognitive theory. American Psychologist 44(9): 1175–1184. DOI: 10.1037/0003-066X.44.9.1175.

Bandura

(2006) Adolescent development from an agentic perspective. In: Pajares

Urdan

(eds) Self-Efficacy Beliefs of Adolescents. Greenwich, CT: Information Age Publishing, 1–43.

Begg

Mazumdar

(1994) Operating characteristics of a rank correlation test for publication bias. Biometrics 50(4): 1088–1101. DOI: https://doi.org/10.2307/2533446.

Bellibaş

MŞ

Liu

(2017) Multilevel analysis of the relationship between principals’ perceived practices of instructional leadership and teachers’ self-efficacy perceptions. Journal of Educational Administration 55(1): 49–69. DOI: https://doi.org/10.1108/JEA-12-2015-0116.

Bergmark

(2023) Teachers’ professional learning when building a research-based education: Context-specific, collaborative and teacher-driven professional development. Professional Development in Education 49(2): 210–224. DOI: https://doi.org/10.1080/19415257.2020.1827011.

10.

Bowers

(2020) Examining a congruency-typology model of leadership for learning using two-level latent class analysis with TALIS 2018. OECD Education Working Papers No. 219. DOI: https://doi.org/10.1787/c963073b-en.

11.

Burić

Slišković

Sorić

(2020) Teachers’ emotions and self-efficacy: A test of reciprocal relations. Frontiers in Psychology 11: 530990. DOI: https://doi.org/10.3389/fpsyg.2020.01650.

12.

Bush

(2013) Instructional leadership and leadership for learning: Global and South African perspectives. Education as Change 17(1): 5–20. DOI: https://doi.org/10.1080/16823206.2014.865986.

13.

Cansoy

Parlar

(2018) Examining the relationship between school principals’ instructional leadership behaviors, teacher self-efficacy, and collective teacher efficacy. International Journal of Educational Management 32(4): 550–567. DOI: https://doi.org/10.1108/IJEM-04-2017-0089.

14.

Caprara

Barbaranelli

Steca

, et al. (2006) Teachers’ self-efficacy beliefs as determinants of job satisfaction and students’ academic achievement: A study at the school level. Journal of School Psychology 44(6): 473–490. DOI: https://doi.org/10.1016/j.jsp.2006.09.001.

15.

Cheung

MW-L

(2015a) Meta-Analysis: A Structural Equation Modeling Approach. John Wiley&Sons.

16.

Cheung

MW-L

(2015b) metaSEM: An R package for meta-analysis using structural equation modelling. Frontiers in Psychology 5: Article 1512, 7. DOI: https://doi.org/10.3389/fpsyg.2014.01521.

17.

Coleman

(1985) Schools and the communities they serve. The Phi Delta Kappan 66(8): 527–532. Available at: https://www.jstor.org/stable/20387425.

18.

Coppe

Parmentier

Kelchtermans

, et al. (2024) Beyond traditional narratives about teacher professional development: A critical perspective on teachers’ working life. Teaching and Teacher Education 139, Article 104436, 15. DOI: https://doi.org/10.1016/j.tate.2023.104436.

19.

Daing

Mustapha

(2023) School administrators’ instructional leadership skills and teachers’ performance and efficacy in senior high schools in the national capital region, Philippines. International Journal of Educational Policy Research and Review 10(1): 1–18. DOI: https://doi.org/10.15739/IJEPRR.23.001.

20.

Daniëls

Hondeghem

Dochy

(2019) A review on leadership and leadership development in educational settings. Educational Research Review 27: 110–125. DOI: https://doi.org/10.1016/j.edurev.2019.02.003.

21.

Darling-Hammond

Hyler

Gardner

(2017) Effective Teacher Professional Development. Palo Alto, CA: Learning Policy Institute.

22.

Day

Sammons

(2016) The impact of leadership on student outcomes: How successful school leaders use transformational and instructional strategies to make a difference. Educational Administration Quarterly 52(2): 221–258. DOI: https://doi.org/10.1177/0013161X15616863.

23.

Desimone

Garet

(2015) Best practices in teachers’ professional development in the United States. Psychology, Society, & Education 7(3): 252–263.

24.

Dilekçi

Limon

(2022) The relationship between principals’ instructional leadership and teachers’ positive instructional emotions: Self-efficacy as a mediator. Journal of Educational Leadership and Policy Studies 6(1): 1–20.

25.

Donohoo

(2017) Collective teacher efficacy research: Implications for professional learning. Journal of Professional Capital and Community 2(2): 101–116. DOI: https://doi.org/10.1108/JPCC-10-2016-0027.

26.

Egger

Davey Smith

Schneider

, et al. (1997) Bias in meta-analysis detected by a simple, graphical test. BMJ 315: 629–634. DOI: https://doi.org/10.1136/bmj.315.7109.629.

27.

(2024) The relationship between principal leadership and teacher practice: Exploring the mediating effect of teachers’ beliefs and professional learning. Educational Studies 50(2): 166–185. DOI: https://doi.org/10.1080/03055698.2021.1936458.

28.

Gignac

Szodorai

(2016) Effect size guidelines for individual differences researchers. Personality and Individual Differences 102: 74–78. DOI: https://doi.org/10.1016/j.paid.2016.06.069.

29.

Goddard

(2001) Collective efficacy: A neglected construct in the study of schools and student achievement. Journal of Educational Psychology 93(3): 467–476.

30.

Goddard

Sook Kim

, et al. (2015) A theoretical and empirical analysis of the roles of instructional leadership, teacher collaboration, and collective efficacy beliefs in support of student learning. American Journal of Education 121(4): 501–530.

31.

Goddard

Hoy

(2004) Collective efficacy beliefs: Theoretical developments, empirical evidence, and future directions. Educational Researcher 33(3): 3–13. DOI: https://doi.org/10.3102/0013189X033003003.

32.

Grissom

Egalite

Lindsay

(2021) How Principals Affect Students and Schools. A Systematic Synthesis of Two Decades of Research. NY, NY: The Wallace Foundation.

33.

Guidetti

Viotti

Bruno

, et al. (2018) Teachers’ work ability: A study of relationships between collective efficacy and self-efficacy beliefs. Psychology Research and Behavior Management 11: 197–206. DOI: https://doi.org/10.2147/PRBM.S157850.

34.

Gurr

(2019) Australian Considerations in relation to instructional leadership and leadership for learning. In: Townsend

(ed) Instructional Leadership and Leadership for Learning in Schools: Understanding Theories of Leading. Heidelberg, Germany: Palgrave Macmillan, 77–104.

35.

Guskey

(2000) Evaluating Professional Development. Thousand Oaks, CA: Corwin Press.

36.

Hallinger

(2011) Leadership for learning: Lessons from 40 years of empirical research. Journal of Educational Administration 49(2): 125–142. DOI: https://doi.org/10.1108/09578231111116699.

37.

Hallinger

Gümüş

Bellibaş

MŞ

(2020) ‘Are principals instructional leaders yet?’ A science map of the knowledge base on instructional leadership, 1940–2018. Scientometrics 122(3): 1629–1650. DOI: https://doi.org/10.1007/s11192-020-03360-5.

38.

Hallinger

Heck

(2010) Leadership for learning: Does collaborative leadership make a difference in school improvement? Educational Management Administration & Leadership 38(6): 654–678. DOI: https://doi.org/10.1177/1741143210379060.

39.

Hallinger

Hosseingholizadeh

Hashemi

, et al. (2018) Do beliefs make a difference? Exploring how principal self-efficacy and instructional leadership impact teacher efficacy and commitment in Iran. Educational Management Administration & Leadership 46(5): 800–819. DOI: https://doi.org/10.1177/1741143217700283.

40.

Hallinger

Liu

(2016) Leadership and teacher learning in urban and rural schools in China: Meeting the dual challenges of equity and effectiveness. International Journal of Educational Development 51: 163–173. DOI: https://doi.org/10.1016/j.ijedudev.2016.10.001.

41.

Hallinger

Murphy

(1985) Assessing the instructional leadership behavior of principals. Elementary School Journal 86(2): 217–248.

42.

Harris

Jones

(2017) Leading professional learning: Putting teachers at the centre. School Leadership & Management 37(4): 331–333. DOI: https://doi.org/10.1080/13632434.2017.1343705.

43.

Hayward

Ohlson

(2023) Teachers’ perspectives on teacher self-efficacy and principal leadership characteristics. International Journal of Education Policy and Leadership 19(2): 1–26. https://doi.org/10.22230/ijepl.2023v19n2a1291 .

44.

Higgins

JPT

Thompson

Deeks

, et al. (2003) Measuring inconsistency in meta-analyses. BMJ 327(7414): 557–560.

45.

Bentler

(1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal 6(1): 1–55. DOI: https://doi.org/10.1080/10705519909540118.

46.

Jack

(2015) Meta-Analytic Structural Equation Modelling. AG Switzerland: Springer.

47.

Jak

Cheung

MW-L

(2020) Meta-analytic structural equation modeling with moderating effects on SEM parameters. Psychological Methods 25(4): 430–455. DOI: https://doi.org/10.1037/met0000245.

48.

Kaya

Demir

(2022) Analysis of the relationship between school leadership and collective teacher efficacy: A cultural comparison. International Journal of Leadership in Education 0(0): 1–18. DOI: https://doi.org/10.1080/13603124.2022.2128431.

49.

Klassen

Tze

VMC

(2014) Teachers’ self-efficacy, personality, and teaching effectiveness: A meta-analysis. Educational Research Review 12: 59–76.

50.

Kline

(2005) Principles and Practice of Structural Equation Modeling (2nd ed). New York, NY: Guilford Press.

51.

Kurt

Duyar

Çalik

(2012) Are we legitimate yet? A closer look at the casual relationship mechanisms among principal leadership, teacher self-efficacy and collective efficacy. Journal of Management Development 31(1): 71–86. DOI: https://doi.org/10.1108/02621711211191014.

52.

Langelaan

Gaikhorst

Smets

, et al. (2024) Differentiating instruction: Understanding the key elements for successful teacher preparation and development. Teaching and Teacher Education 140, Article 104464, 14. DOI: https://doi.org/10.1016/j.tate.2023.104464.

53.

Leithwood

Sun

Schumacker

(2020) How school leadership influences student learning: A test of “the four paths model”. Educational Administration Quarterly 56(4): 570–599. DOI: https://doi.org/10.1177/0013161X1987877.

54.

Liu

Hallinger

(2018) Principal instructional leadership, teacher self-efficacy, and teacher professional learning in China: Testing a mediated-effects model. Educational Administration Quarterly 54(4): 501–528. DOI: https://doi.org/10.1177/0013161X18769048.

55.

Liu

Hallinger

Feng

(2016) Supporting the professional learning of teachers in China: does principal leadership make a difference?. Teaching and Teacher Education 59: 79–91. DOI: https://doi.org/10.1016/j.tate.2016.05.023.

56.

Liu

Bellibaş

MŞ

Gümüş

(2021) The effect of instructional leadership and distributed leadership on teacher self-efficacy and job satisfaction: Mediating roles of supportive school culture and teacher collaboration. Educational Management Administration & Leadership 49(3): 430–453. DOI: https://doi.org/10.1177/1741143220910438.

57.

Liu

Huang

(2022) To what extent is shared instructional leadership related to teacher self-efficacy and student academic performance in China? School Effectiveness and School Improvement 33(3): 381–402. DOI: https://doi.org/10.1080/09243453.2022.2029746.

58.

Liu

Liao

(2019) Professional development and teacher efficacy: Evidence from the 2013 TALIS. School Effectiveness and School Improvement 30(4): 487–509. DOI: https://doi.org/10.1080/09243453.2019.1612454.

59.

MacBeath

Townsend

(2011) Thinking and acting both locally and globally: What do we know now and how do we continue to improve? In: Townsend

MacBeath

(eds) International Handbook of Leadership for Learning: Part 1. Dordrecht, The Netherlands: Springer, 1237–1254.

60.

Malinen

Savolainen

(2016) The effect of perceived school climate and teacher efficacy in behavior management on job satisfaction and burnout: A longitudinal study. Teaching and Teacher Education 60: 144–152. DOI: https://doi.org/10.1016/j.tate.2016.08.012.

61.

Marks

Printy

(2003) Principal leadership and school performance: An integration of transformational and instructional leadership. Educational Administration Quarterly 39(3): 370–397. DOI: https://doi.org/10.1177/0013161X03253412.

62.

Moher

Liberati

Tetzlaff

, et al. (2010) Preferred reporting items for systematic reviews and me-ta-analyses: the PRISMA statement. International Journal of Surgery 8(5): 336–341. DOI: https://doi.org/10.1016/j.ijsu.2010.02.007.

63.

Moolenaar

Sleegers

PJC

Daly

(2012) Teaming up: Linking collaboration networks, collective efficacy, and student achievement. Teaching and Teacher Education 28(2): 251–262. DOI: https://doi.org/10.1016/j.tate.2011.10.001.

64.

Morris

Usher

Chen

(2017) Reconceptualizing the sources of teaching self-efficacy: A critical review of emerging literature. Educational Psychology Review 29(4): 795–833. DOI: https://doi.org/10.1007/s10648-016-9378-y.

65.

Murphy

Elliott

Goldring

, et al. (2007) Leadership for learning: A research-based model and taxonomy of behaviors. School Leadership & Management 27(2): 179–201.

66.

Nawab

Quraishi

(2024) ‘Teachers value PD if leaders value PD’: Role of school leaders in professional development of teachers in rural Pakistan. Leadership and Policy in Schools, 0(0), 1–16. DOI: https://doi.org/10.1080/15700763.2024.2303688.

67.

Ninković

Knežević Florić

OČ

(2018) Transformational school leadership and teacher self-efficacy as predictors of perceived collective teacher efficacy. Educational Management Administration & Leadership 46(1): 49–64. DOI: https://doi.org/10.1177/1741143216665.

68.

Ninković

Knežević Florić

OČ

(2024) School leadership and teaching practice: A systematic review of studies of the indirect effects. Journal of Educational Administration 62(4): 357–372. DOI: https://doi.org/10.1108/JEA-08-2023-0190.

69.

Patton

Parker

Tannehill

(2015) Helping teachers help themselves: Professional development that makes a difference. NASSP Bulletin 99(1): 26–42. DOI: https://doi.org/10.1177/0192636515576040.

70.

Powers

Kaniuka

Phillips

, et al. (2016) The impact of teacher-lead professional development programs on the self-efficacy of veteran teachers. Journal of Research Initiatives 2(1): 1–21. Available at: https://digitalcommons.uncfsu.edu/jri/vol2/iss1/16.

71.

R Core Team (2021) R: A language and environment for statistical computing. R Foundation for Statistical Computing (4.1.2) [Computer software]. Available at: https://www.R-project.org/.

72.

Robinson

VMJ

Lloyd

Rowe

(2008) The impact of leadership on student outcomes: An analysis of the differential effects of leadership types. Educational Administration Quarterly 44(5): 635–674. DOI: https://doi.org/10.1177/0013161X08321509.

73.

Ross

Gray

(2006) Transformational leadership and teacher commitment to organizational values: The mediating effects of collective teacher efficacy. School Effectiveness and School Improvement 17(2): 179–199. DOI: https://doi.org/10.1080/09243450600565795.

74.

Rosseel

(2012) Lavaan: An R package for structural equation modeling. Journal of Statistical Software 48: 1–36. DOI: https://doi.org/10.18637/jss.v048.i02.

75.

Schermelleh-Engel

Moosbrugger

Müller

(2003) Evaluating the fit of structural equation models: tests of significance and descriptive goodness-of-fit measures. Methods of Psychological Research 8(2): 23–74. Available at: https://www.stats.ox.ac.uk/~snijders/mpr_Schermelleh.pdf.

76.

Schwarzer

(2022) Package ‘meta’ (5.2-0) [Computer software]. CRAN. Available at: https://github.com/guido-s/meta/.

77.

Shal

Ghamrawi

Abu-Tineh

, et al. (2024) Teacher leadership and virtual communities: Unpacking teacher agency and distributed leadership. Education and Information Technologies 29: 15025–15042 DOI: https://doi.org/10.1007/s10639-023-12446-5.

78.

Skaalvik

(2007) Dimensions of teacher self-efficacy and relations with strain factors, perceived collective teacher efficacy, and teacher burnout. Journal of Educational Psychology 99(3): 611–625. DOI: https://doi.org/10.1037/0022-0663.99.3.611.

79.

Skaalvik

(2014) Teacher self-efficacy and perceived autonomy: Relations with teacher engagement, job satisfaction, and emotional exhaustion. Psychological Reports 114(1): 68–77. DOI: https://doi.org/10.2466/14.02.PR0.114k14w0.

80.

Skaalvik

(2016) Teacher stress and teacher self-efficacy as predictors of engagement, emotional exhaustion, and motivation to leave the teaching profession. Creative Education 7(13): 1785–1800. DOI: https://doi.org/10.4236/ce.2016.713182.

81.

Skaalvik

(2019) Teacher self-efficacy and collective teacher efficacy: Relations with perceived job resources and job demands, feeling of belonging, and teacher engagement. Creative Education 10(7): 1400–1424. DOI: https://doi.org/10.4236/ce.2019.107104.

82.

Spillane

Halverson

Diamond

(2001) Investigating school leadership practice: A distributed perspective. Educational Researcher 30(3): 23–28. DOI: https://doi.org/10.3102/0013189X030003023.

83.

Stephanou

Oikonomou

(2018) Teacher emotions in primary and secondary education: Effects of self-efficacy and collective-efficacy, and problem-solving appraisal as a moderating mechanism. Psychology 9(04): 820–875. DOI: https://doi.org/10.4236/psych.2018.94053.

84.

Sun

Leithwood

(2017) Leadership effects on student learning mediated by teacher emotions. In K. Leithwood, J. Sun & K. Pollock (Eds.), How School Leaders Contribute to Student Success: The Four Paths Framework. Dordrecht, The Netherlands: Springer, 137–152.

85.

Tabachnick

Fidell

(2007) Using Multivariate Statistics. (5th ed.). New York, NY: Allyn and Bacon.

86.

Tan

Gao

Shi

(2022) Second-order meta-analysis: Synthesising the evidence on associations between school leadership and different school outcomes. Educational Management, Administration & Leadership 50(3): 469–490. DOI: https://doi.org/10.1177/1741143220935456.

87.

Thien

Lim

Adams

(2021) The evolving dynamics between instructional leadership, collective teacher efficacy, and dimensions of teacher commitment: What can Chinese independent high schools tell us? International Journal of Leadership in Education 27(4): 739–761DOI: https://doi.org/10.1080/13603124.2021.1913236.

88.

Townsend

Berryman

Gurr

, et al. (2020) Leadership for learning in diverse settings: School leaders setting the agenda. In: Hall

Sammons

Lindorff

(eds) International Perspectives in Educational Effectiveness Research. Springer Nature, 327–360.

89.

Tran

Nguyen

, et al. (2022) Workplace conditions created by principals for their teachers’ professional development in Vietnam. International Journal of Leadership in Education 25(2): 238–257. DOI: https://doi.org/10.1080/13603124.2019.1708472.

90.

Tschannen-Moran

Hoy

(2001) Teacher efficacy: Capturing an elusive construct. Teaching and Teacher Education 17(7): 783–805. DOI: https://doi.org/10.1016/S0742-051X(01)00036-1.

91.

Tschannen-Moran

McMaster

(2009) Sources of self-efficacy: Four professional development formats and their relationship to self-efficacy and implementation of a new teaching strategy. The Elementary School Journal 110(2): 228–245.

92.

Viel-Ruma

Houchins

Jolivette

, et al. (2010) Efficacy beliefs of special educators: The relationships among collective efficacy, teacher self-efficacy, and job satisfaction. Teacher Education and Special Education 33(3): 225–233. DOI: https://doi.org/10.1177/0888406409360129.

93.

Walker

Slear

(2011) The impact of principal leadership behaviors on the efficacy of new and experienced middle school teachers. NASSP Bulletin 95(1): 46–64. DOI: https://doi.org/10.1177/0192636511406530.

94.

Windlinger

Warwas

Hostettler

(2020) Dual effects of transformational leadership on teacher efficacy in close and distant leadership situations. School Leadership & Management 40(1): 64–87. DOI: https://doi.org/10.1080/13632434.2019.1585339.

95.

Wood

Bandura

(1989) Social cognitive theory of organisational management. The Academy of Management Review 14(3): 361–384. DOI: https://doi.org/10.5465/amr.1989.4279067.

96.

Woodcock

Hardy

(2023) Teacher self-efficacy, inclusion and professional development practices: Cultivating a learning environment for all. Professional Development in Education 0(0): 1–15. DOI: https://doi.org/10.1080/19415257.2023.2267058.

97.

Woodcock

Sharma

Subban

, et al. (2022) Teacher self-efficacy and inclusive education practices: Rethinking teachers’ engagement with inclusive practices. Teaching and Teacher Education, 117, Article 103802, 12. DOI: https://doi.org/10.1016/j.tate.2022.103802.

98.

Yang

(2020) The effects of professional development experience on teacher self-efficacy: Analysis of an international dataset using Bayesian multilevel model. Professional Development in Education 46(5): 797–811. DOI: https://doi.org/10.1080/19415257.2019.1643393.

99.

Yoon

Park

Jung

(2023) What makes teachers’ professional learning more or less effective?: An evolution of community of practice for physical education teachers. Journal of Teaching in Physical Education 1: 1–10. DOI: https://doi.org/10.1123/jtpe.2023-0005.

100.

Zheng

Yin

(2019) Exploring the relationships among instructional leadership professional learning communities and teacher self-efficacy in China. Educational Management Administration & Leadership 47(6): 843–859. DOI: https://doi.org/10.1177/1741143218764176.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.02 MB

0.00 MB