Principal leadership typologies and their relationship with teacher self-efficacy and commitment: A latent profile mediation analysis

Abstract

This study investigated the heterogeneous nature of teachers’ perceptions of distinct leadership practices and explored the predictors of teacher- and school-level leadership typologies on teacher self-efficacy and commitment by using a multilevel latent profile mediation analysis of a dataset of 890 teachers from K-12 schools in Turkey. The results revealed three distinct leadership profiles (limiting, transitioning, and integrating) at the teacher level and two profiles (transitioned and integrated) at the school level, providing evidence that integrating teachers and integrated school profiles were associated with greater teacher self-efficacy and commitment. Additionally, both levels of integrated leadership profiles were indirectly related to teacher commitment through teacher self-efficacy vis-a-vis other profile groups. The study concluded that different leadership styles are related and that the impact of leadership on teachers is maximized when multiple leadership types are integrated.

Keywords

Integrated leadership school leadership profiles shared instructional leadership transformational leadership teacher self-efficacy teacher commitment

Introduction

Growing research on school improvement has affirmed that effective school leadership is crucial for increased student learning outcomes (Grissom et al., 2021). Principals can improve outcomes through multiple mechanisms that impact the school, but particularly effective are factors related to teachers. For instance, teachers’ self-efficacy and commitment have garnered increased attention since research evidence suggests that they are significantly linked with pupils’ achievement (Park, 2005; Zee et al., 2018). Therefore, educational leadership scholars have developed a particular interest in investigating whether and how school leaders might influence teacher self-efficacy (TSE) (Liu et al., 2021; Liu and Hallinger, 2018) and teacher commitment (TC) (Al-Mahdy et al., 2018; Cansoy et al., 2022).

However, when linking leadership to teacher behaviors and beliefs, most educational scholars have relied upon a single leadership model approach, investigating the relationship between one leadership type—instructional leadership, transformational leadership (TL), or shared leadership—and TSE and/or TC (e.g. Al-Mahdy et al., 2018; Cansoy et al., 2022; Liu and Hallinger, 2018). However, this approach falls far short of providing a comprehensive understanding of the nature of school leadership in practice (Bush and Glover, 2014). Consistently, the prominent contingency theory emerged as a reaction against normative features of different leadership models and the single-leader tendency in school leadership (e.g. Lambert, 2002; Leithwood et al., 1999), calling for a new understanding of leadership that combines different theories and models to suggest the optimal leadership practices contingent upon the internal and external context (Fernandez et al., 2010). Thus, recent research has moved from the single leadership model approach to combining multiple leadership practices to create leadership configurations (Bellibaş et al., 2021a; Urick, 2016; Urick and Bowers, 2014). This is based on the premise that leaders enact multiple leadership styles simultaneously to exert their influence on teachers and students, requiring researchers to employ a profiling approach to discover how multiple leadership types are integrated (e.g. Urick and Bowers, 2014; Veletić and Olsen, 2021).

The profiling approach to school leadership has received increasing attention during the past decade, much of the literature has focused only on identifying leadership profiles and cannot link each profile to school processes, structures, or people, thus creating a new venue for further research. To address this gap in the literature, the present study aims to identify the leadership profiles in schools and examine how each profile might be related to TSE and TC. In addition, building on strong theoretical and empirical evidence that TSE beliefs can shape teachers’ commitment to the profession (Chesnut and Burley, 2015; Hallinger et al., 2018), we take into account that leadership profiles can be linked to TC both directly and indirectly through TSE. Therefore, in this research, we employed a latent class profile analysis approach to discover school leadership profiles, alongside structural equation modeling to examine how each profile is associated directly with TC and indirectly through TSE, by utilizing a dataset of 890 teachers collected from Turkish elementary, middle, and secondary schools. The following research questions guided the study:

RQ1.

What are the different profiles of principals across Turkish schools based on teachers’ perceptions?

RQ2.

How are these leadership profiles related to TSE and TC?

RQ3.

Does TSE mediate the relationship between leadership profiles and TC?

Theoretical background

The extensive school leadership literature focused on school improvement has revealed several complementary and alternative leadership models. Gümüş et al. (2018) noted that traditional leadership studies attributed to a single person dominated the educational leadership literature until the beginning of the millennium. Since this time, conceptualizations of leadership have shifted to regard it as the product of collective performance, with a greater emphasis on educational accountability and decentralization initiatives. However, contingency theory criticizes the notion of one single ideal leadership presented as a panacea to all institutional problems (Hanson, 1979). The complex nature of the school context requires leadership to adapt to the situation and context, rather than adopting a uniform leadership style. In this way, leaders must select the most appropriate leadership mode to successfully assess contingencies and intervene accordingly (Bush and Glover, 2014).

Recent theoretical and empirical studies in educational leadership suggest that leadership types are not separate and multiple leadership models coexist and interact in the daily work of school leaders (Bowers, 2020; Marks and Printy, 2003; Urick, 2016; Urick and Bowers, 2014). This understanding supported the development of new leadership models (e.g. learning-centered leadership) that support the integration of multiple leadership types (e.g. TL, distributed leadership, and instructional leadership). Researchers concluded that the integration of multiple leadership practices can bolster the impact of leadership on teacher practices and school outcomes (Bellibaş et al., 2021a; Kwan, 2020; Marks and Printy, 2003).

Following the blueprint created by previous educational leadership researchers (Bowers, 2020; Urick, 2016; Urick and Bowers, 2014; Veletić and Olsen, 2021), the present study adopted an integrated leadership approach by employing the person-centered principal leadership trajectories to identify leadership profiles at both the individual teacher and school level. The researchers then examined how each leadership profile might be related to TSE and TC, to fill a critical gap in the educational leadership literature. Based on the literature (Chesnut and Burley, 2015; Hallinger et al., 2018), we propose that leadership profiles are related to TC both directly and indirectly through the mediating role of TSE (Figure 1).

In the following sections, we provide a discussion of each concept used in the theoretical model (shown in Figure 1).

Shared instructional leadership

Originating from the school effectiveness movement of the 1970s, instructional leadership has developed from the premise that principal leaders play a pivotal role in improving the quality of teaching and learning in K-12 school settings (Bellibaş et al., 2021b, 2022; Hallinger, 2018). The accumulated knowledge on instructional leadership points to two distinct perspectives regarding its purposes (e.g. Sheppard, 1996). While the narrow approach encompasses such instructional leadership practices as conducting classroom observations or monitoring student progress, which are intended to impact teaching endeavors directly, the broad approach attempts to improve the school culture, vision, and curriculum that indirectly affect student achievement (Marks and Printy, 2003).

Early conceptualizations of instructional leadership were based on the “superhero principals” perspective, which placed principals at the center of school management, placed all the leadership work on their shoulders, and ignored the sharing of responsibilities (Hallinger and Heck, 2010; Lambert, 2002). With the development and spread of distributed leadership in educational research as a result of policy advocacy for teacher empowerment and professional learning practices in the 1990s (Hallinger, 2003), scholars began promoting the idea of shared instructional leadership (SIL) to address the concern that principal-centered instructional leadership ignores the contributions of teachers and the school community to teaching (Murphy et al., 2016; Printy and Marks, 2006; Spillane et al., 2003). SIL emphasizes synergy arising from collaboration among members of the professional learning community rather than the individual endeavor of principals to improve teaching and learning (Blase and Blase, 2000; Lambert, 2002; Marks and Printy, 2003). It does not limit leadership solely to a role or position but rather underscores the interactions among members that flow through networks in the school community (Bolman and Deal, 2018).

SIL promotes a high quality of teaching through a vision in the school, mutual leadership of school staff, collective decision-making processes, shared expectations, and broad collaboration (Blase and Blase, 2000; Lambert, 2002; Newmann et al., 2001; Printy and Marks, 2006). Similarly, in the present study, SIL refers to all collective actions among the school community members to improve teaching and learning. It involves collaboration among the school community to develop the school vision and goals, supervise instruction, monitor student progress, and promote professional learning. Developing a school vision and goals involves school leaders and teachers working collaboratively to develop the school's purpose, aims, and mission (Lambert, 2002; Marks and Printy, 2003). Supervising instruction and feedback encompasses principals’ and teachers’ classroom observations, feedback on others’ classroom skills, and shared decision-making regarding the improvement of teaching (Blase and Blase, 2000). Monitoring student progress reflects the school principal and teachers working together to evaluate student performance and achievement (Hallinger, 2003). Professional learning points to the collaboration among the school community to organize and support professional learning activities at school (Blase and Blase, 2000; Printy et al., 2009).

Transformational leadership

TL, one of the popular leadership models in educational and instructional leadership (Bush, 2014), can be traced back to Burns’ work on political leadership (Burns, 1978), which was future extended in scope by Bass’ (1985) study on business management. TL is defined as uniting followers of leaders around a common vision and goals, being innovative to challenge problems, providing support by coaching and mentoring followers, and inspiring employees to develop their capacities (Bass and Avolio, 1990). In the early 1990s, scholars embraced TL as an important model in educational management, recognizing the importance of theory to counter changes in education (Leithwood, 1994; Leithwood and Jantzi, 1990). They suggested that the reconstruction of schools according to twenty-first-century requirements, regardless of transformational school leadership theory, would make no real change in existing education policies (Berkovich, 2016).

Transformational school leaders build a vision and promote common goals, communicate high-performance expectations, provide individualized support, deliver intellectual stimulation, and create a school culture conducive to change and improvement (Allen et al., 2015; Hallinger, 2003; Leithwood and Sleegers, 2006). In the present paper, we define TL as school principals’ efforts toward school vision building, intellectual stimulation, and individualized consideration (Leithwood and Jantzi, 2000; Moolenaar et al., 2010). Transformational leaders attempt to build a shared school vision for all teachers by empowering and providing support while setting high-performance expectations for all staff (Day et al., 2016). They have an idealized influence that teachers emulate, admire, and trust. In addition, TL highlights the individual efforts of principals toward matching the objectives and values of the school with the needs of teachers (Moolenaar et al., 2010).

Integrating SIL and TL: The profiling approach to differentiate school leadership configurations

Investigating school leadership calls for the use of new methodological approaches to understanding the interplay of leadership practice within the organizational and cultural context of schools (Hallinger and Huber, 2012). However, the majority of educational leadership studies to date have used common “variable-centered” prototypical analytical techniques, such as correlation, regression, and structural equation modeling (Tan et al., 2022). The variable-centered approach is based on the assumption that all individuals in a sample come from a single population for which a single set of “mean” parameters can be estimated; thus, it seeks to clarify the associations among variables of interest in a population (Howard and Hoffman, 2018). For instance, researchers have attempted to devise models to explain how distinct principal leadership practices (e.g. instructional, distributed, and transformational) directly and indirectly affect school organizations (Liu et al., 2021; Ryu et al., 2022), teachers’ beliefs (Liu and Hallinger, 2018; Thomas et al., 2020), teacher classroom practices (Bellibaş et al., 2022), and student learning (Leithwood et al., 2020). Additionally, scholars have suggested that when multiple leadership models coexist and interact, the impact on school outcomes is boosted (Bellibaş et al., 2021a; Kwan, 2020; Marks and Printy, 2003).

However, few studies have moved beyond the traditional variable-centered approach to investigate “person-centered” principal leadership trajectories by employing quantitative latent profile/class analysis (Bowers, 2020; Urick, 2016; Veletić and Olsen, 2021) or qualitative interview-based techniques (van Schaik et al., 2020). “Person-centered” approaches (e.g. latent class/profile analysis) help to classify individuals more accurately in distinct patterns, both qualitatively and quantitatively (Bauer and Shanahan, 2007; Bergman and Trost, 2006; Collins and Lanza, 2013). Rather than focusing on individual components of variables, the person-centered approach is based on the idea of identifying and comparing homogenous subpopulations of individuals who share similar profiles on the leadership model component under investigation. When similar individuals are grouped, researchers can predict whether different profiles theoretically differ on one or more criteria (Bauer and Shanahan, 2007). Thus, such person-centered approaches could determine how groups of principals that are classified based on similar leadership behaviors might differ from each other (Bergman and Trost, 2006).

In the present study, we adopted the person-centered approach to determine both ideal and non-ideal principal leadership configurations in schools and examine the impact of such configurations on school processes and outcomes. We used the latent profile analysis (LPA) approach to provide an understanding of how principals are grouped and how each group differs from others based on leadership behaviors (Printy et al., 2009; Urick and Bowers, 2014). More specifically, we conducted profile analysis to identify how categorical multidimensional TL and SIL practices (e.g. at low, medium, and high levels) are implemented by principals. Then, we compared the resulting leadership profiles relative to their direct influence on TC and indirect effect through TSE.

Teacher self-efficacy

Rooted in social-cognitive theory, self-efficacy is defined as “people's judgments of how well they can organize and execute, constituent cognitive, social, and behavioral skills in dealing with prospective situations” (Bandura, 1983, p. 467). This characteristic is influenced by various factors such as mastery experience (i.e. success and failure), vicarious experience (i.e. testimony to the successes and failures of others), verbal persuasion (i.e. family, friends, and colleagues), and physiological and affective states (i.e. excitement and fear) (Bandura, 1997). TSE is characterized as teachers’ belief in their capability to conduct effective classroom management, provide clear instruction, and ensure the learning of all students (Shengnan and Hallinger, 2021). Influenced by Tschannen-Moran and Woolfolk Hoy (2001), TSE in the present research refers to teachers’ belief in their ability to leverage effective instruction, classroom management, and student engagement. TSE in instruction encompasses a teacher's ability to use a variety of teaching approaches in their lessons. TSE in classroom management relates to precautions taken by teachers in classroom situations that impede effective teaching. Finally, TSE in student engagement refers to teachers’ belief in their ability to motivate students to ensure active learning (Tschannen-Moran and Woolfolk Hoy, 2001).

Teacher commitment

Organizational commitment, one of the key concerns of organizational effectiveness research, has been conceptualized in different ways (Cook and Wall, 1980). In the organizational behavior field, two substantial studies portray the concept in detail. First, Porter et al. (1974) described the three-component pattern of commitment as involving beliefs in organizational goals and values (identification), a willingness to strive on behalf of the organization (involvement), and a strong desire to maintain organizational membership (loyalty). Second, Allen and Meyer's (1990) multidimensional model comprises the emotional bonds developed with an organization through the positive work experiences of the individual (affective); the commitment to the individual based on their perceived costs, both economic and social (continuance); and the commitment to the norms of reciprocity and the perceived obligation to the organization (normative). Based on these explanations, TC is an internal force that drives teachers to desire to be good at their profession, as well as an external force that encourages them to fulfill their professional responsibilities at school (Park, 2005). In the present study, we employed the commitment model proposed by Cook and Wall (1980) and revised by Mathews and Shepherd (2002), which involves three dimensions: identification, involvement, and loyalty. Identification encompasses TCs to school goals and values, while involvement points to teachers’ willingness to strive to improve their school. Finally, loyalty corresponds to the strong emotional bonds that teachers have toward the school (Park, 2005).

Method

This study involved quantitative cross-sectional research. In this section, we describe the study participants, data collection instruments, and data analysis approach.

Participants

We conducted this study using survey data collected from public primary, secondary, and high schools in 25 districts of Turkey's Ankara province. Ankara is a cosmopolitan city that reflects the social, cultural, and economic diversity among different regions of Turkey. We followed a two-step procedure to select our research participants. First, we identified public schools in 25 districts of Ankara using a proportional stratified sampling design to ensure the representability of the sample. Second, we contacted randomly selected school principals in each district through WhatsApp, phone, and email. We prepared online questionnaire forms using Google Forms and distributed them to the schools’ teachers through WhatsApp and email. A total of 890 teachers from 107 schools participated in the study. While 22.5% (n = 200) of the respondents were male, 77.5% (n = 690) were female. In total, 87.4% (n = 778) of the teachers had a bachelor's degree and 12.6% (n = 112) had a postgraduate degree. In total, 37.7% (n = 335) of the participants taught in primary schools, 43.4% (n = 386) in secondary schools, and 19% (n = 169) in high schools. The mean age of the respondents was 40.2 (SD = 7.84), and the mean years of teaching experience were 7.1 (SD = 6.43).

Measures

SIL. The SIL scale was developed by the authors of this study. The scale comprises 16 items measuring joint work by the school community in the following areas: developing school vision and goals (e.g. principals and teachers work together to develop a school vision), supervising instruction and feedback (e.g. principals and teachers work together to conduct classroom observations), monitoring student progress (e.g. principals and teachers work collaboratively to analyze student achievement data), and professional learning for teachers (e.g. principals and teachers work together to organize professional learning activities at school). Respondents were asked to respond to these items on a five-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). We followed the proposed procedural steps suggested by DeVellis (2003). To begin, a pool of 25 items was created by the researchers, using previous theoretical and empirical studies (Lambert, 2002; Marks and Printy, 2003; Urick and Bowers, 2014) Then, the opinions of three education administration experts, a measurement and assessment expert, and a Turkish language expert were consulted for content validity, which reduced the number of items to 16. The final draft form was administered to 320 teachers for the pilot implementation, which was followed by exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) for the construct validity of the measurement. Finally, average variance extracted (AVE) and composite reliability (ω) values were calculated for the reliability analysis of the scale.

Validity and reliability analyses of the measurement were performed in two phases. In the first phase, the Kaiser–Meyer–Olkin (KMO = .95) and Bartlett’s sphericity test (χ² = 5587.44, p < 0.001) were estimated and the results showed that the data were sufficient for the consistency of the item values. As a result of applying the EFA varimax rotation, two items with low factor loads were excluded from the analysis; thus, the final scale was composed of a four-factor structure with 16 items. The factor loads of the scale explained 69% of the total variance explained, and the item factor loads ranged between .62 and .85. The Cronbach alpha coefficient was .96. In the second phase, CFA was performed on the final data for the construct validity of the scale. The CFA fit index values (χ²/df = 2.78; p < 0.001, Y = 0.043; SRMR = 0.016; CFI = 0.990; TLI = 0.987) were then calculated, along with AVE (0.717) and composite reliability (ω = 0.975). These results showed that the scale has distinctive items and is valid and reliable.

TL. Moolenaar et al. (2010) developed the second latent profile variable of the study. The dimensions of this scale include vision building, individualized consideration, and intellectual stimulation. Participants’ responses to the 5-point Likert-type scale ranged from 1 (strongly disagree) to 5 (strongly agree). The scale, which includes 18 items, was adapted to Turkish culture by Balyer and Özcan (2012). Sample items included “refers explicitly to our school's goals during decision-making processes” (vision building), “listens carefully to team members’ ideas and suggestions” (individualized consideration), and “encourages teachers to experiment with new strategies” (intellectual stimulation). The fit indices showed good fit (χ²/df = 2.76; p < 0.0001, RMSEA = 0.041, SRMR = 0.015, CFI = 0.983, TLI = 0.979). The distinctiveness of the scale's items (AVE = 0.838) was also found to be acceptable. The composite reliability coefficient (ω = .989) was excellent; thus, the scale was compatible with the Turkish school setting.

TSE. The TSE scale was developed by Tschannen-Moran and Woolfolk Hoy (2001) and adapted into Turkish by Çapa et al. (2005). TSE was measured with 24 items in the present study, according to the dimensions of student engagement, classroom management, and instructional strategy. The Likert-type scale was scored between 1 (insufficient) and 5 (very sufficient). Sample items were “to what extent can you get students to believe they can do well in schoolwork?” (student engagement), “to what extent can you control disruptive behavior in the classroom?” (classroom management), and “to what extent can you diversify your assessment strategies?” (instructional strategy). The repeated CFA results confirmed the three-dimensional structure (χ²/df = 2.73, RMSEA = 0.041, SRMR = 0.022, CFI = 0.942, TLI = 0.924). The CFA, along with the convergent validity (AVE = .553) and composite reliability (ω = .949), demonstrated a good fit.

TC. The TC scale was originally developed by Cook and Wall (1980), before being revised by Mathews and Shepherd (2002) and adapted into Turkish by the researchers in the present study. The scale consists of 9 items divided into three dimensions: identification, involvement, and loyalty. Items were scored on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Sample items included “I feel myself to be a part of my school” (identification), “knowing that my own work made a contribution to the school would please me” (involvement), and “I sometimes feel like I should leave this job for good” (loyalty). Repeated CFA for the validity of the scale (χ²/df = 2.17, RMSEA = 0.035, SRMR = 0.023, CFI = 0.993, TLI = 0.998), confirmed a good fit. In addition, the value of the convergent validity (AVE = .643) calculated for the discrimination among items was at an acceptable level. Moreover, the omega value (ω = .933) performed for the reliability of the scale items was also acceptable.

Statistical analytic procedure

We employed a person-centered approach to multilevel structural equation mixed modeling (SEMM), designed for evaluating complicated statistical models that are auxiliary models to the focal mixture model (e.g. LPA), by using a robust maximum-likelihood in Mplus 8.8 (Muthén and Muthén, 1998–2017). SEMM accurately classifies cases (teachers, schools, etc.) into quantitatively distinct sub-population profiles (Morin et al., 2020). Before starting the LPA, we performed preliminary analyses such as mean, standard deviation, and Pearson correlation analysis. To determine the suitability of the variables for multilevel analysis, we employed intraclass correlation (ICC) analysis.

The LPA for this study followed two main stages. The first stage involved describing and identifying latent profiles of leadership traits using unconditional LPA. Next, utilizing unconditional LPA, we performed unconditional multilevel LPA to determine whether the school-level profiles were more suitable when teacher-level profiles were fixed. An LPA using maximum likelihood was conducted to identify leadership profiles. In the model, we attempted to randomly allocate people into classes and estimate a one-profile solution, a two-profile solution, and so on, until inquiry of fit statistics indicated a best-fitting solution. Optimal model selection was guided by the likelihood ratio chi-square (LRχ²), Akaike information criterion (AIC), Bayesian information criterion (BIC), size adjusted BIC (SABIC), Lo–Mendell–Rubin adjusted likelihood ratio test (LMR-LRT), and bootstrapped likelihood ratio test (BLRT) (Asparouhov and Muthén, 2014). We resolved the optimal profile solution based on the following parameters: (1) the lowest LRχ², AIC, BIC, and SABIC values compared with other model patterns (Nylund et al., 2007); (2) significant LMR and BLRT p-values (p < 0.01) estimating the better fit of a k-profile model than a k − 1 profile model (McLarnon and O’Neill, 2018); (3) higher entropy values (e.g. entropy >0.80) indicating better probabilities of successfully classifying participants in a latent profile (Muthén and Muthén, 1998–2017); (4) adequate sample size in each group (i.e. at least 5% of the total sample of the given class) with a high probability of correct classification and low probability of belonging to other classes; and (5) in line with previous empirical and theoretical studies to make meaningful interpretations (Morin et al., 2020). Based on these parameters, we identified our principal and school profiles and labeled each profile.

In the final stage, we examined the effect of auxiliary variables (e.g. leadership profiles) latent profile membership on TC via the TSE-mediating mechanism. For this, a counterfactual-based causal effect decomposition approach was used on the complex auxiliary model (Asparouhov and Muthén, 2014). The literature proposes two contemporary approaches (VAM and BCH) for testing auxiliary variable relations (Morin et al., 2020). In this study, we used the BCH approach for auxiliary variables, since covariates have strong entropy, a large sample size, and a low amount of missing data. This approach assigns individuals to classes based on their modal posterior probabilities and then sets the classification error on these assignments when predicting class-specific distributions (Asparouhov and Muthén, 2014). Several steps were taken for the BCH analysis. First, BCH weights reflecting the measurement error in the auxiliary variable were calculated. Second, mean-centered TSE and TC outcome variables were included in the model, respectively. Groups with high standardized average scores were selected as reference groups to simplify the interpretation of complex model analyses and to make more accurate pairwise comparisons between profiles. Third, natural direct effect and natural indirect effect values were employed by controlling the non-reference groups to calculate direct, indirect, and total causal effects on the outcome variables of reference groups (McLarnon and O’Neill, 2018). Lastly, to estimate the significance of the counterfactual causal effects, we applied a 2000-sample bootstrap analysis for the confidence interval (CI) of total, indirect and direct effects as suggested by Preacher and Hayes (2008). All variables were aggregated for the school-level mediation analysis.

Results

Step 1: Preliminary analysis

To begin analysis, descriptive statistics (mean, standard deviation, correlation coefficients, and ICC coefficients) for each variable were estimated (see Table 1). The mean scores for the SIL and TL dimensions were all relatively high, ranging from M = 3.94 to M = 4.19. The mean values of outcome variables for TSE (M = 3.83) were relatively high, while they were quite high for TC (M = 4.25). At the teacher and school level, there were moderately positive relationships between the leadership profiles and teacher belief variables (self-efficacy and commitment) (0.300 < r < 0.700; p < 0.001). Finally, a multilevel CFA was performed to examine the compatibility of variables at both the teacher and school levels. The coefficients at both the teacher level (ICC1 >.05) and school level (ICC2 >.50) indicated a good fit for the variables (Fleiss, 1986). Accordingly, all variables were structured with teachers nested within schools, thereby justifying the use of multilevel LPA.

Step 2: The best-fitting teacher-level model and its identifying characteristics

Table 1.

Descriptive statistics and correlations among study variables at the teacher level (n = 890 teachers) and the school level (n = 107 schools).

	M	SD	ICC	1	2	3	4	5	6	7	8	9
Teacher level
1. DSVG	4.01	.04	.619**	—
2. SI	3.94	.03	.601**	.956**	—
3. MSP	4.03	.04	.584**	.966**	.899**	—
4. PLT	4.04	.04	.597**	.940**	.963**	.894**	—
5. VB	4.09	.04	.650**	.924**	.903**	.828**	.906**	—
6. IS	4.19	.05	.647**	.930**	.905**	.832**	.871**	.947**	—
7. IC	4.10	.04	.625**	.951**	.926**	.854**	.924**	.979**	.950**	—
8. TSE	3.83	.01	.515**	.695**	.713**	.609**	.746**	.673**	.644**	.709**	—
9. TC	4.25	.01	.536**	.717**	.721**	.667**	.634**	.662**	.698**	.687**	.541**	—
School level
1. DSVG	—	.49	.171**	—
2. SI	—	.45	.155**	.837**	—
3. MSP	—	.47	.142**	.787**	.835**	—
4. PLT	—	.50	.154**	.794**	.814**	.850**	—
5. VB	—	.57	.179**	.791**	.762**	.770**	.786**	—
6. IS	—	.62	.176**	.717**	.674**	.699**	.699**	.778**	—
7. IC	—	.58	.166**	.772**	.748**	.759**	.776**	.814**	.806**	—
8. TSE	—	.18	.119**	.410**	.397**	.404**	.405**	.409**	.353**	.420**	—
9. TC	—	.26	.132**	.543**	.473**	.464**	.462**	.482**	.451**	.459**	409**	—

**p < 0.001, means (M), standard deviations (SD), intraclass correlations (ICC), developing school vision and goals (DSVG), supervising instruction (SI), monitoring student progress (MSP), professional learning for teachers (PLT), vision building (VB), intellectual stimulation (IS), individualized consideration (IC), teacher self-efficacy (TSE), and teacher commitment (TC).

Next, we performed several analytical processes to identify leadership profiles. First, we estimated an unconditional LPA model to explore school leadership profiles by checking the information criteria at the teacher level (Level 1). The information criteria (AIC, BIC, SABIC, BLRT, LMR-LRT, and entropy) for distinct profile solutions of the teacher-level LPA are based only on the differences in standardized mean (z) scores among the profiles (Table 2).

Table 2.

Model fit indices of latent profile (LP) analysis at the teacher level.

LP	LRχ²	P	AIC	BIC	SABIC	Entropy	LMR-LRT	BLRT
1	−8801.668	14	17631.337	17698.414	17653.95	—	—	—
2	−7133.334	22	14310.668	14416.075	14346.207	0.932	0.000	0.000
3	−5744.577	30	11549.153	11692.890	11597.616	0.968	0.000	0.000
4	−5326.787	38	10729.574	10911.641	10790.960	0.975	0.111	0.000
5	−5021.963	46	10135.926	10356.322	10210.235	0.959	0.098	0.000

Note: The bold values show the information criteria for the optimal profile.

AIC: Akaike information criterion; BIC: Bayesian information criterion; BLRT: bootstrapped likelihood ratio test; LMR-LRT: Lo–Mendell–Rubin adjusted likelihood ratio test; SABIC: size adjusted BIC; LRχ²: likelihood ratio chi-square.

As shown in Table 2, models with one to five latent profiles were estimated and compared to select a leadership style model. The final profile solution that best approximated the data was selected based on theory and model fit indices. The AIC, BIC, and SABIC continued to decline as additional classes were added. The LMR-LRT suggested the three-profile model, while BLRT suggested the five-profile model (p < 0.001). The entropy figures for the three- and five-profile models were 0.96 and 0.95, respectively. We chose the three-profile model as the optimal solution because (1) the three-profile model indicated greater profile separation than the five-profile model, (2) the profiles in the three-profile model were more interpretable, and (3) the three-profile model replicated profiles previously described in the literature (see Urick and Bowers, 2014; Urick, 2016).

Taxonomies based on standardized mean scores (z) were then created for the three profiles of SIL and TL leadership types at the teacher level (see Figure 2). Specifically, Profile 1, which comprised 15% (n = 130) of the sample, was characterized by the lowest levels of SIL and TL (z = −1.676, −1.410) and was the smallest group. The largest group, comprising 51% (n = 455) of the sample, was Profile 2 (z = −.204, −.150), which was characterized by both moderate SIL and TL levels. Finally, Profile 3 comprised 34% (n = 305) of the sample (z = .941, 1.030) and included individuals with high SIL and TL scores, respectively.

We labeled the school leadership profiles as Profile 1 (“limiting”), Profile 2 (“transitioning”), and Profile 3 (“integrating”), based on the studies of Urick (2016) and Urick and Bowers (2014). In addition, we plotted a graph using standardized mean scores to visualize the differences among the three profiles and determine the representative position of each principal TL and SIL dimension (see Figure 3).

The x-axis and y-axis in Figure 3 represent the TL and SIL dimensions, respectively. The centroid for the “limiting” principals’ distribution is in the lowest side left quadrant. These principals demonstrated the lowest levels of TL and SIL practice, indicating that they perform both TL and SIL duties at a lower level in their management roles. The centroid of the “transitioning” group of principals is close to the center of the quadrant, indicating moderate-range responses. Consequently, transitioning principals perform their managerial duties at a middling level. This profile is termed “transitioning” because of its position between the limiting and integrating profiles. The centroid for “integrating” principals’ distribution is located in the highest side right quadrant. These principals had the highest levels of TL and SIL.

Step 3: The best-fitting school-level model and its identifying characteristics

During the initial stage of this step, we estimated the school-level profiles by performing multilevel LPA analysis based on the relative frequency of the three teacher-level profiles (see Table 3).

Table 3.

Model fit indices of latent profile (LP) analysis at the school level.

LP	LRχ²	P	AIC	BIC	SABIC	Entropy
1	−5744.577	30	11549.153	11692.890	11597.616	0.968
2	−5716.696	33	11499.392	11657.503	11552.701	0.854
3	−5714.832	36	11496.665	11668.149	11556.820	0.850
4	−5709.272	39	11496.543	11683.401	11559.544	0.849
5	−5708.950	42	11501.900	11703.132	11569.748	0.843

Note: The bold values show the information criteria for the optimal profile.

AIC: Akaike information criterion; BIC: Bayesian information criterion; SABIC: size adjusted BIC; LRχ²: likelihood ratio chi-square.

As illustrated in Table 3, when examining the information criteria (e.g. AIC, BIC, and SABIC) for the one- to five-profile solutions, all the information criteria decreased strongly in the second profile only. Moreover, compared to solutions with three or more profiles, the two-profile classification's entropy value of 0.854 yielded the optimal solution. Consequently, the two-profile solution was deemed satisfactory at the school level.

In the final stage, we identified the teachers’ classifications by analyzing three school-level profiles (see Figure 4). At the school level, 56% (n = 487) of the teachers were in the first profile and 44% (n = 403) were in the second profile. The first latent profile included schools with relatively higher numbers of principals in the “integrating” profile (n = 269); thus, this group was labeled as “integrated schools.” The second latent profile contained schools where the majority of principals were characterized by the “transitioning” profile (n = 293); thus, we labeled these schools as “transitioned schools.”

Step 4: Investigating the direct and indirect effect of leadership profiles on TSE and TC

Finally, at the teacher level, we analyzed the direct, indirect, and total effects of principal leadership profiles on outcome variables by using the BCH-weighted approach (see Table 4). Furthermore, we identified the limiting and transitioning profiles with the integrating profile as a reference variable and tested its effects on the TSE and commitment outcomes. Overall, limiting versus transitioning (β = .756, SE = .170, p < 0.001), limiting versus integrating (β = 1.111, SE = .166, p < 0.001), and transitioning versus integrating (β = .355) SE = .047, p < 0.001) profiles significantly predicted the probability of teachers having higher self-efficacy. This means that the teachers who perceived their principals as integrating, rather than limiting or transitioning, were more likely to have stronger self-efficacy perceptions. Likewise, the perception of school commitment was higher among teachers who rated their principals as transitioning rather than limiting (β = .677, SE = .097, p < 0.001), integrating rather than limiting (β = 1.338, SE = .107, p < 0.001), and integrating rather than transitioning (β = .662, SE = .075, p < 0.001).

Table 4.

Teacher level and school level latent profile analysis mediation effect model.

	Teacher level						School level
	Limiting versus transitioning		Limiting versus integrating		Transitioning versus integrating		Transitioned versus integrated
	Estimate [95% CI]	SE	Estimate [95% CI]	SE	Estimate [95% CI]	SE	Estimate [95% CI]	SE
Direct Effect
TSE	.756*** [.500, 1.061]	.170	1.111*** [.862, 1.404]	.166	. 355*** [.281, .435]	0.47	.322*** [.282, .360]	.024
TC	.677*** [.506, .827]	.097	1.338*** [1.151, 1.504]	.107	.662*** [.539, .789]	.075	.504*** [.460, .643]	.026
TC on TSE	.120** [.023, .175]	.053	.339** [.196, .560]	.109	.187** [.068, .298]	.070	.535*** [.392, .681]	.087
Natural direct effect	.676*** [.508, .826]	.097	1.414*** [1.232, 1.581]	.107	.714*** [.613, .823]	.064	.594*** [.545, .640]	.029
Natural indirect effect	.141* [.053, .269]	.065	.376** [.207, .662]	.133	.120**[.069, .225]	.044	.148** [.113, .193]	.024
Total effect	.817*** [.622, .989]	.110	1.790*** [1.534, 2.054]	.162	.834*** [.713, .960]	.076	.742*** [.681, .805]	.039

Note. 95% CI: 95% confidence Interval. *p < 0.05, **p < 0.01, ***p < 0.001.

As for the indirect effects, when controlling for the limiting profile, both the transitioning profile (β = .141, SE = .065, p < 0.05) and the integrating profile (β = .376, SE = .133, p < 0.01) were positively and significantly related to TC through TSE. This means that teachers who perceived their principals as integrating were more likely to have a stronger self-efficacy perception than those who indicated that their principals were limiting or transitioning, and such a difference in TSE was associated with increased TC.

At the school level, the teachers in schools belonging to the integrated leadership profile generally had a higher sense of TSE (β = .322, SE = .024, p < 0.001) and TC (β = .504, SE = .026, p < 0.001) than teachers in schools belonging to the transitioned leadership profile. This finding indicates that teachers with high self-efficacy and commitment are likely to be working with an integrated school leader. In addition, the indirect effect of the integrated school profile on TC was higher than that of the transitioned profile (β = .148, SE = .024, p < 0.01). In other words, teachers in schools with integrated leadership have much higher TSE compared to their peers working in schools with transitioned leaders, and such difference is associated with even larger TC in schools with integrated principals.

Discussion

Despite the shift from the variable-centered approach and leadership styles perspective toward person-centered statistics and a leadership typology model in educational leadership research (Urick and Bowers, 2014; Veletić and Olsen, 2021), few studies have investigated how leadership types might be related to school processes, people, and outcomes (e.g. Urick, 2016). In this study, we identified teacher- and school-level leadership profiles and examined the relationship between each profile and TSE and TC by using multilevel latent profile mediation analysis.

Our first research question aimed to identify school principals’ leadership profiles. Educational leadership scholars have favored two school leadership styles, SIL and TL, for their considerable influence on the process and outcomes of schools (Hallinger and Heck, 2010; Marks and Printy, 2003). Both empirical and theoretical studies of contingency theory provide compelling support for combining the practices of these leadership styles to increase the impact of school leadership (Hallinger, 2003; Krüger et al., 2007). The present leadership combined these leadership practices and used LPA to identify different groups of principals who demonstrated similar patterns. While previous studies have already implemented this strategy to identify various leadership profiles in the United States (e.g. Urick, 2016; Urick and Bowers, 2014), we conducted similar analyses with attention to the contextual differences between the United States and Turkey. Indeed, our analysis produced slightly different profiles of school principals in the Turkish context when compared to those of the previous studies.

Contingency theory supports the idea of combining multiple leadership behaviors. Depending on the context, different combinations of leadership behaviors might prove more effective (Hanson, 1979). While in the present research we did not focus on the influence of the context in predicting the type of leadership profiles, we attempted to create profiles that could help better identify the state of school leadership in Turkey. We found that teachers are grouped into three profiles (limiting, transitioning, and integrating), while two profiles exist at the school level (transitioned and integrated). These groups were named according to the previous work of Urick (2016) and Urick and Bowers (2014). The limiting profile indicates a lack of leadership sharing around instructional issues and lower TL demonstrated by school principals, while the transitioning group fell in the middle of the scale and showed a moderate level of SIL in school and a moderate level of principal TL. The integrating profile was dubbed as such to indicate teachers whose responses showed evidence of high TL and SIL. The first school-level leadership profile is called “transitioned” because this group involves more transitioning teachers than limiting or integrating teachers; similarly, the integrated schools include a larger number of teachers from the integrating group. However, discovering the contextual factors that explain which schools or principals belong to each type (e.g. Urick and Bowers, 2014) is beyond the scope of the present study. Future research based on contingency theory might investigate which school and principal context factors determine the profiles identified in the present study.

Our second and third research questions examine how the leadership profiles might be related to TSE and TC at the teacher and school levels. Although the literature clearly demonstrates that principal leadership and TSE factors are substantial antecedents of TC (Geijsel et al., 2003; Hallinger et al., 2018), such research has yet to provide an understanding of the relationship between leadership, self-efficacy, and commitment when considering a person-centered approach to leadership. Consistent with several previous studies that reported augmented student achievement (Kwan, 2020; Marks and Printy, 2003), more effective classroom practices (Bellibaş et al., 2021a), and higher teacher retention (Urick, 2016) when multiple leadership types were integrated, the present study found that teachers in the integrating and transitioning groups are more likely to have a higher sense of TSE and TC than teachers in the limiting group. In addition, membership in the integrating group positively influenced TSE, which in turn contributed to teachers’ commitment to the profession. We also add nuance to the literature by providing evidence of an indirect relationship between leadership profiles and TC through TSE.

These results were also consistent with the findings of traditional variable-centered research, indicating that improved leadership practices were associated with higher TSE and TC (Al-Mahdy et al., 2018; Cansoy et al., 2022; Liu et al., 2021; Liu and Hallinger, 2018). This could be attributed to the usage of two favorable leadership styles (SIL and TL), which might have led to linearity in the position of the profiles on a coordinate grid (see Figure 2). As a result, the limiting profile (weak SIL and TL) corresponded to weak TSE and TC, and similarly, moderate leadership (the transitioning profile) was associated with moderate TSE and TC. Using contradicting leadership styles (e.g. transactional versus transformational) might result in a different set of profiles and consequently a more complex relationship between profiles and teacher behaviors.

Limitations

The following limitations of this study should be highlighted when interpreting and applying its results. First, the cross-sectional survey design of our study, where data represent only a fixed snapshot in time, does not allow for causal inferences regarding relationships between the leadership, TSE, and TC constructs. Longitudinal or experimental models are needed to determine whether a change in leadership profile results in any improvement or decline in teacher behaviors. Second, although the BLRT value is among the important information criteria for determining how large a sample should be for power analysis, there is no minimum agreed-upon sample size for profile studies (Dziak et al., 2014). While the sample size for this study was substantial (n = 890), larger sample sizes may still be preferable for detecting small differences between profiles. Finally, research based on teachers’ self-reports has the potential to raise a social desirability issue. While this is less likely to be a concern in the present study because of the multivariate nature of the person-centered analyses employed (Meyer and Morin, 2016), the inclusion of multi-source data in future studies could help minimize these types of biases.

Conclusion

Educational leadership scholars have long researched how school leaders might contribute to teacher behaviors, thus improving TSE and commitment to their school. This study takes those endeavors one step further by treating leadership profiles as predictors and linking them to two teacher behaviors: self-efficacy and commitment (e.g. Urick, 2016). The results of our study support previous research on the person-centered leadership approach by indicating that leadership practices are related; thus, it seems less likely that principals have too much variation in their enactment of different leadership styles (e.g. Urick and Bowers, 2014). When school leaders perform strong TL, they are likely to also demonstrate strong SIL practices with other stakeholders (integrated leaders) and vice versa. The results also affirm the integrated leadership literature by concluding that the impact of leadership on school processes and outcomes is stronger when multiple leadership styles are combined (Bellibaş et al., 2021b; Kwan, 2020; Marks and Printy, 2003; Urick, 2016).

The leadership typologies outlined in this present study and their link to teacher behaviors hold considerable implications for policymakers, practitioners, and researchers. Policymakers might focus on school leaders who exhibit a weak leadership profile and develop new policies (e.g. principal assignment, professional development, and mentorship) to improve their leadership skills. Researchers might investigate antecedents of leadership profiles: for instance, by examining the school and principal characteristics that make a principal integrating versus limiting. Similarly, future research could also provide quantitative evidence regarding the moderation effect of context factors in the effect of leadership on school processes and outcomes. The results of the present study might also help school principals to reflect on the SIL and TL practices in their schools to identify their own strengths and weaknesses as school leaders. A shared vision and practices of instructional leadership along with a strong TL orientation could help principals promote stronger belief in teachers’ capacity to address educational issues and improve student learning, which in turn could bolster teachers’ commitment to their school and the teaching profession.

Footnotes

Declaration of conflicting interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Mehmet Şükrü Bellibaş

Mahmut Polatcan

Muhammet İbrahim Akyürek

Author biographies

Mehmet Şükrü Bellibaş is working as an associate professor in the Department of Educational Sciences at Adıyaman University, Turkey. His research interests include comparative and international educational leadership, instructional leadership, teaching practices, and teacher professional learning. His scholarly works are based on quantitative research methods that aim to investigate the relationship between principal leadership and teacher organizational behaviors and student learning outcomes.

Mahmut Polatcan is currently working as an associate professor of educational administration at the Faculty of Letters, Karabük University, Turkey. He received his PhD in Educational Administration from Ankara University in Turkey. His research interests include principal leadership, teacher professional learning, and instructional practices.

Muhammet İbrahim Akyürek is an assistant professor in the Department of Educational Sciences at Selçuk University in Turkey. He conducts research on school leadership, particularly in authentic and democratic leadership. Her research interests also include distance education, whistleblowing, self-efficacy, and psychological and social capital.

References

Al-Mahdy

YEH

Emam

Hallinger

(2018) Assessing the contribution of principal instructional leadership and collective teacher efficacy to teacher commitment in Oman. Teaching and Teacher Education 69: 191–201.

Allen

Grigsby

Peters

(2015) Does leadership matter? Examining the relationship among transformational leadership, school climate, and student achievement. International Journal of Educational Leadership Preparation 10(2): 1–22.

Allen

Meyer

(1990) The measurement and antecedents of affective, continuance and normative commitment to the organization. Journal of Occupational Psychology 63: 1–18.

Asparouhov

Muthén

(2014) Auxiliary variables in mixture modeling: Three-step approaches using Mplus. Structural Equation Modeling: A Multidisciplinary Journal 21(3): 329–341.

Balyer

Özcan

(2012) Cultural adaptation of headmasters’ transformational leadership scale and a study on teachers’ perceptions. Eurasian Journal of Educational Research 49(1): 103–128.

Bandura

(1983) Self-efficacy determinants of anticipated fears and calamities. Journal of Personality and Social Psychology 45(2): 464–469.

Bandura

(1997) Self-Efficacy: The Exercise of Control. New York: Freeman.

Bass

(1985) Leadership and Performance Beyond Expectations. New York: The Free Press.

Bass

Avolio

(1990) The implications of transactional and transformational leadership for individual, team, and organizational development. In: Woodman

Pasmore

(eds) Research in Organizational Change and Development. Greenwich, CT: JAI Press, pp.231–272.

10.

Bauer

Shanahan

(2007) Modeling complex interactions: Person-centered and variable-centered approaches. In: Little

Bovaird

Card

(eds) Modeling Contextual Effects in Longitudinal Studies. Mahwah, NJ: Lawrence Erlbaum, pp.255–283.

11.

Bellibaş

MŞ

Gümüş

Liu

(2021b) Does school leadership matter for teachers’ classroom practice? The influence of instructional leadership and distributed leadership on instructional quality. School Effectiveness and School Improvement 32(3): 387–412.

12.

Bellibaş

MŞ

Kılınç

AÇ

Polatcan

(2021a) The moderation role of transformational Leadership in the effect of instructional leadership on teacher professional learning and instructional practice: An integrated leadership perspective. Educational Administration Quarterly 57(5): 776–814.

13.

Bellibaş

MŞ

Polatcan

Kılınç

AÇ

(2022) Linking instructional leadership to teacher practices: The mediating effect of shared practice and agency in learning effectiveness. Educational Management Administration & Leadership 50(5): 812–831.

14.

Bergman

Trost

(2006) The person-oriented versus the variable-oriented approach: Are they complementary, opposites, or exploring different worlds? Merrill-Palmer Quarterly 52(3): 601–632.

15.

Berkovich

(2016) School leaders and transformational leadership theory: Time to part ways? Journal of Educational Administration 55(5): 609–622.

16.

Blase

(2000) Effective instructional leadership: Teachers’ perspectives on how principals promote teaching and learning in schools. Journal of Educational Administration 38(2): 130–141.

17.

Bolman

Deal

(2018) Reframing the Path to School Leadership: A Guide for Teachers and Principals. Thousand Oaks, California: Corwin.

18.

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, OECD Publishing, Paris.

19.

Burns

(1978) Leadership. New York: Harper & Row.

20.

Bush

(2014) Instructional leadership in centralized contexts: Rhetoric or reality? Educational Management Administration & Leadership 42(1): 3–5.

21.

Bush

Glover

(2014) School leadership models: What do we know? School Leadership & Management 34(5): 553–571.

22.

Cansoy

Parlar

Polatcan

(2022) Collective teacher efficacy as a mediator in the relationship between instructional leadership and teacher commitment. International Journal of Leadership in Education 25(6): 900–918.

23.

Çapa

Çakıroğlu

Sarıkaya

(2005) The development and validation of a Turkish version of the teachers’ sense of efficacy scale. Education and Science 30(137): 74–81.

24.

Chesnut

Burley

(2015) Self-efficacy as a predictor of commitment to the teaching profession: A meta-analysis. Educational Research Review 15: 1–16.

25.

Collins

Lanza

(2013) Latent Class and Latent Transition Analysis: With Applications in the Social, Behavioral, and Health Sciences. Hoboken, NJ: John Wiley & Sons.

26.

Cook

Wall

(1980) New work attitude measures of trust, organizational commitment and personal need non-fulfilment. Journal of Occupational Psychology 53(1): 39–52.

27.

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.

28.

DeVellis

(2003) Scale Development Theory and Applications (2nd ed). Thousand Oaks, California: Sage Publication.

29.

Dziak

Lanza

Tan

(2014) Effect size, statistical power, and sample size requirements for the bootstrap likelihood ratio test in latent class analysis. Structural Equation Modeling: A Multidisciplinary Journal 21(4): 534–552.

30.

Fernandez

Cho

Perry

(2010) Exploring the link between integrated leadership and public sector performance. The Leadership Quarterly 21(2): 308–323.

31.

Fleiss

(1986) Reliability of measurement. In: Fleiss

(ed.) The Design and Analysis of Clinical Experiments. New York: Wiley, pp.1–12.

32.

Geijsel

Sleegers

Leithwood

, et al. (2003) Transformational leadership effects on teachers’ commitment and effort toward school reform. Journal of Educational Administration 41(3): 228–256.

33.

Grissom

Egalite

Lindsay

(2021) How Principals Affect Students and Schools. New York: Wallace Foundation. Retrieved from http://www.wallacefoundation.org/principalsynthesis.

34.

Gümüş

Bellibaş

Esen

, et al. (2018) A systematic review of studies on leadership models in educational research from 1980 to 2014. Educational Management Administration & Leadership 46(1): 25–48.

35.

Hallinger

(2003) Leading educational change: Reflections on the practice of instructional and transformational leadership. Cambridge Journal of Education 33(3): 329–352.

36.

Hallinger

(2018) Bringing context out of the shadows of leadership. Educational Management Administration & Leadership 46(1): 5–24.

37.

Hallinger

Heck

(2010) Leadership for learning: Does collaborative leadership make a difference in school improvement? Educational Management Administration & Leadership 38(6): 654–678.

38.

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.

39.

Hallinger

Huber

(2012) School leadership that makes a difference: International perspectives. School Effectiveness and School Improvement 23(4): 359–367.

40.

Hanson

(1979) School management and contingency theory: An emerging perspective. Educational Administration Quarterly 15(2): 98–116.

41.

Howard

Hoffman

(2018) Variable-centered, person-centered, and person-specific approaches: Where theory meets the method. Organizational Research Methods 21(4): 846–876.

42.

Krüger

Witziers

Sleegers

(2007) The impact of school leadership on school level factors: Validation of a causal model. School Effectiveness and School Improvement 18(1): 1–20.

43.

Kwan

(2020) Is transformational leadership theory passé? Revisiting the integrative effect of instructional leadership and transformational leadership on student outcomes. Educational Administration Quarterly 56(2): 321–349.

44.

Lambert

(2002) A framework for shared leadership. Educational Leadership 59(8): 37–40.

45.

Leithwood

(1994) Leadership for school restructuring. Educational Administration Quarterly 30(4): 498–518.

46.

Leithwood

Jantzi

(1990) Transformational leadership: How principals can help reform school cultures. School Effectiveness and School Improvement 1(4): 249–280.

47.

Leithwood

Jantzi

(2000) The effects of transformational leadership on organizational conditions and student engagement with school. Journal of Educational Administration 38(2): 112–129.

48.

Leithwood

Sleegers

(2006) Transformational school leadership: Introduction. School Effectiveness and School Improvement 17(2): 143–144.

49.

Leithwood

Sun

Schumacker

(2020) How school leadership influences student learning: A test of “The Four Paths Model.” Educational Administration Quarterly 56(4): 570–599.

50.

Leithwood

Jantzi

Steinbach

(1999) Changing Leadership for Changing Times. Buckingham: Open University Press.

51.

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.

52.

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.

53.

Marks

Printy

(2003) Principal leadership and school performance: An integration of transformational and instructional leadership. Educational Administration Quarterly 39(3): 370–397.

54.

Mathews

Shepherd

(2002) Dimensionality of Cook and Wall's (1980) British organizational commitment scale revisited. Journal of Occupational and Organizational Psychology 75(3): 369–375.

55.

McLarnon

MJW

O’Neill

(2018) Extensions of auxiliary variable approaches for the investigation of mediation, moderation, and conditional effects in mixture models. Organizational Research Methods 21(4): 955–982.

56.

Meyer

Morin

AJS

(2016) A person-centered approach to commitment research: Theory, research, and methodology. Journal of Organizational Behavior 37: 584–612.

57.

Moolenaar

Daly

Sleegers

PJC

(2010) Occupying the principal position: Examining relationships between transformational leadership, social network position, and schools’ innovative climate. Educational Administration Quarterly 46(5): 623–670.

58.

Morin

AJS

McLarnon

MJW

Litalien

(2020) Mixture modeling for organizational behavior research. In: Grip

Hansen

(eds) Handbook on the Temporal Dynamics of Organizational Behavior. Massachusetts, USA: Edward Elgar Publishing, pp.351–379.

59.

Murphy

Neumerski

Goldring

, et al. (2016) Bottling fog? The quest for instructional management. Cambridge Journal of Education 46(4): 455–471.

60.

Muthén

(1998–2017) Mplus User's Guide (8th ed.). Los Angeles, CA: Muthén & Muthén.

61.

Newmann

Smith

Allensworth

, et al. (2001) Instructional program coherence: What it is and why it should guide school improvement policy. Educational Evaluation and Policy Analysis 23(4): 297–321.

62.

Nylund

Asparouhov

Muthén

(2007) Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling 14: 535–569.

63.

Park

(2005) Teacher commitment and its effects on student achievement in American high schools. Educational Research and Evaluation 11(5): 461–485.

64.

Porter

Steers

Mowday

, et al. (1974) Organizational commitment, job satisfaction, and turnover among psychiatric technicians. Journal of Applied Psychology 59(5): 603–609.

65.

Preacher

Hayes

(2008) Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods 40(3): 879–891.

66.

Printy

Marks

(2006) Shared leadership for teacher and student learning. Theory into Practice 45(2): 125–132.

67.

Printy

Marks

Bowers

(2009) Integrated leadership: How principals and teachers share transformational and instructional influence. Journal of School Leadership 19(5): 504–532.

68.

Ryu

Walls

Seashore Louis

(2022) Caring leadership: The role of principals in producing caring school cultures. Leadership and Policy in Schools 21(3): 585–602.

69.

Shengnan

Hallinger

(2021) Unpacking the effects of culture on school leadership and teacher learning in China. Educational Management Administration & Leadership 49(2): 214–233.

70.

Sheppard

(1996) Exploring the transformational nature of instructional leadership. Alberta Journal of Educational Research 42(4): 325–344.

71.

Spillane

Hallett

Diamond

(2003) Forms of capital and the construction of leadership: Instructional leadership in urban elementary schools. Sociology of Education 76(1): 1–17.

72.

Tan

Gao

Shi

(2022) Second-order meta-analysis synthesizing the evidence on associations between school leadership and different school outcomes. Educational Management Administration & Leadership 50(3): 469–490.

73.

Thomas

Tuytens

Devos

, et al. (2020) Transformational school leadership as a key factor for teachers’ job attitudes during their first year in the profession. Educational Management Administration & Leadership 48(1): 106–132.

74.

Tschannen-Moran

Woolfolk Hoy

(2001) Teacher efficacy capturing an elusive construct. Teaching and Teacher Education 17: 783–805.

75.

Urick

(2016) The influence of typologies of school leaders on teacher retention: A multilevel latent class analysis. Journal of Educational Administration 54(4): 434–468.

76.

Urick

Bowers

(2014) What are the different types of principals across the United States? A latent class analysis of principal perception of leadership. Educational Administration Quarterly 50(1): 96–134.

77.

van Schaik

Volman

Admiraal

, et al. (2020) Fostering collaborative teacher learning: A typology of school leadership. European Journal of Education 55(2): 217–232.

78.

Veletić

Olsen

(2021) Exploring school leadership profiles across the world: A cluster analysis approach to TALIS 2018. International Journal of Leadership in Education: 1–27. DOI: https://doi.org/10.1080/13603124.2021.1953612.

79.

Zee

Koomen

de Jong

(2018) How different levels of conceptualization and measurement affect the relationship between teacher self-efficacy and students’ academic achievement. Contemporary Educational Psychology 55: 189–200.