This study explores how the relationship between transformational leadership and organizational learning varies across schools and examines how school principals’ self-efficacy and their leadership performance, as perceived schoolwide by teachers, moderate this relationship.
Drawing on data from 1,080 teachers and 31 principals across K-12 schools in Hong Kong, the analysis employs a mixed-effects multilevel linear modeling approach to investigate these dynamics.
The findings reveal that four dimensions of transformational leadership – intellectual stimulation, idealized influence behavior, inspirational motivation and individual consideration – are positively associated with teachers’ perception of organizational learning, though their effects vary across schools. Intellectual stimulation emerges as particularly impactful, especially in schools where principals demonstrate higher self-efficacy and where teachers collectively perceive their principal’s effectiveness in this dimension as strong. Therefore, intellectual stimulation, when paired with a principal’s self-efficacy, becomes a powerful catalyst for organizational learning, but its impact is contingent on the schoolwide endorsement of the principal’s leadership effectiveness. This interplay underscores the complex nature of leadership effectiveness and leader self-efficacy.
These findings offer actionable insights for school leaders and policymakers, underscoring the moderating role of principal self-efficacy in tandem with their schoolwide perceived transformational leadership performance and cultivating supportive school environments to drive meaningful organizational improvement.
Introduction
School leadership is a cornerstone of educational transformation, shaping the culture and collective learning environments that define thriving schools (Hollingworth et al., 2017; Qian and Walker, 2021). Among the various leadership paradigms, transformational leadership (TL) stands out for its capacity to inspire and empower not only individuals but also entire school communities, fostering both individual and organizational learning (OL) (Beverborg, 2024; Hoang and Le, 2024; Lei et al., 2024). As a core mechanism linking leadership to organizational outcomes, this approach enhances self-efficacy among school actors, a critical driver of professional growth and organizational success (Damanik and Aldridge, 2017; Le et al., 2018; Leithwood and Jantzi, 2008; Yilmaz and Tore, 2025). While extensive empirical research has illuminated the profound impact of TL on teacher self-efficacy (Hoang and Le, 2024; Polatcan et al., 2021), a significant gap persists in the educational leadership literature: the interplay between a leader’s self-efficacy, leadership performance and organizational outcomes.
Leadership performance and self-efficacy are deeply intertwined, forming a cyclical relationship that defies simple causal prioritization (Lindsey et al., 1995). On one hand, a school leader’s self-efficacy fuels their ability to inspire and enact change; on the other, effective leadership reinforces their confidence and belief in their capabilities (Kelleher, 2016; Ugwuanyi and Pietsch, 2024). Research suggests that while moderate levels of self-efficacy enhance leadership performance, excessively high self-efficacy could diminish it, creating a self-correcting cycle that warrants closer examination (Machida and Schaubroeck, 2011; Salanova et al., 2011). This nuanced relationship, though widely acknowledged in organizational behavior literature, remains understudied in the context of educational leadership.
Despite the wealth of research demonstrating the effects of leadership on OL (Berson et al., 2006; Do and Mai, 2020), to our knowledge, it remains uncharted how leaders’ self-efficacy might moderate this relationship, especially in educational settings (Hesbol, 2019; Röhl et al., 2022). This oversight is particularly striking given the broadly accepted, pivotal role of school leaders’ TL practices in shaping OL culture in K-12 schools (Leithwood et al., 1999; Leithwood and Sun, 2012; Kools and Stoll, 2016). As Leithwood and Sun’s meta-analysis (2012) concludes that the relationship between TL and OL is highly dependent on school context, we address this critical research gap through a multilevel model, seeking to explore the following research questions:
How does the relationship between TL and OL vary across schools?
How do school principals’ self-efficacy and leadership performance, as perceived schoolwide by teachers, moderate the relationship between TL and OL?
In addressing these questions, the study aims to deepen our theoretical understanding of the mechanisms that underpin effective school leadership and contribute to the broader discourse on educational transformation. By shedding light on the interplay between self-efficacy, leadership performance and OL, this work offers practical implications, laying the groundwork for developing targeted interventions that empower leaders to navigate the complexities of educational change with greater confidence and efficacy.
Conceptual framework
This section synthesizes existing research across three interrelated concepts: TL, OL and leader self-efficacy. Through these conceptual lenses, we establish a robust theoretical foundation for the study. We examine how transformational leadership drives OL and systemic change in schools, while identifying gaps in the literature related to the moderating role of leader self-efficacy in this relationship.
Transformational leadership
TL is widely recognized as a cornerstone of effective leadership, characterized by its ability to inspire, motivate and empower individuals and teams (Bass and Avolio, 1994; Bass et al., 2003). In educational settings, transformational school leaders are expected to cultivate a shared vision, foster collaboration and promote an organizational culture of continuous improvement (Leithwood and Jantzi, 2000; Leithwood, 2010). At its core, this leadership paradigm transcends transactional exchanges, focusing on fostering a shared sense of purpose and collective growth (Riggio et al., 2004).
While building on leadership charisma, transformational leaders act as ethical role models, embodying the values and principles they aim to instill in their organizations – a dual-faceted construct referred to as idealized influence, as conceptualized by Bass and Avolio (1994). Idealized influence (attributed) reflects the extent to which followers admire and trust their leader, while idealized influence (behavior) captures the leader’s actions that demonstrate integrity, self-sacrifice and a commitment to shared values (Antonakis et al., 2003). School leaders who are idealized for their charismatic influence are further empowered by their ability to set high expectations for change among teachers and to engender a shared vision throughout the school (Berkovich, 2016; Berson et al., 2015).
Central to this leadership style is the ability to articulate a compelling and aspirational vision that galvanizes the community, a dimension known as inspirational motivation (Bass et al., 2003). By communicating optimism and setting high expectations throughout the school, transformational school leaders inspire teachers to embrace challenges and strive for excellence (Lynch, 2012). This motivational dynamic is further enriched by transformational leaders’ capacity to nurture creativity and innovation, a process referred to as intellectual stimulation, encouraging teachers to question conventional practices, explore new ideas and engage in reflective problem-solving (Lee and Kuo, 2019; Marks and Printy, 2003).
Another essential characteristic of transformational school leaders is recognizing and responding to each individual’s unique needs and aspirations within the school community, a concept termed individualized consideration. When transformational school leaders provide personalized support, mentorship and encouragement, they foster an inclusive and empowering environment where teachers feel valued and understood (Berkovich and Eyal, 2019).
Overall, the execution of above mentioned TL elements may cultivate a collaborative and dynamic school culture (Hesbol, 2019; Leithwood et al., 1999). Through their TL performance, school leaders can drive systemic inquiry and innovative behavior, ultimately contributing to the broader success of the whole school (Vermeulen et al., 2020).
Linking transformational leadership to organizational learning
OL refers to the collective capacity of an organization to acquire, interpret and apply knowledge to improve practices and outcomes (Senge et al., 2000). In schools, OL emerges as a fundamental driver for adapting to changing educational environments, fostering collaboration and innovation, and sustaining long-term improvement (McCharen et al., 2011). This process entails continuous generation, dissemination and integration of knowledge across the school, enabling dynamic responses to internal and external challenges (Schecter et al., 2022).
A foundational element of OL is creating continuous learning opportunities for all members of the school community (Kools et al., 2020). By providing access to professional development, encouraging experimentation and fostering a growth mindset, schools might empower teachers to expand their skills and knowledge (Miller et al., 2015). In response to this need, transformational school leaders may enhance this process through their ability to inspire intellectual stimulation, encouraging teachers to question conventional practices and explore new ideas (Marks and Printy, 2003; Vashdi et al., 2019).
Transformational leaders, with their ability to articulate a compelling vision, are also uniquely positioned to promote OL through research orientation and systems thinking within schools (Shaked and Schecter, 2020). As Godfrey (2014) asserts, school leaders must excel at facilitating OL through systematic inquiries and research across educational ecosystems, enabling teachers to effectively capture and share knowledge. Information sharing and team learning tendencies are also essential aspects of OL culture, where educators collaboratively co-construct knowledge and tackle complex challenges (Beverborg, 2024; Senge et al., 2000). The link between TL and team learning approach of OL forges stronger interpersonal bonds within schools while concurrently strengthening collective learning and working capacities across educational settings (Welsh et al., 2021).
Another critical aspect of OL culture is the promotion of inquiry and dialogue among school stakeholders, where open communication and critical questioning are actively encouraged (Hosseini and Haghigi-Shirazi, 2021). Schools that prioritize this dimension create safe spaces for educators to engage in meaningful discussions, challenge assumptions and explore new ideas (Welsh et al., 2021). To create such a school culture, transformational school leaders use their capacity to inspire and motivate educators, fostering an environment where inquiry and dialogue are not only encouraged but also valued as essential components of professional growth, ultimately contributing to overall school achievement (Hesbol, 2019). Furthermore, such leadership practices inspire a culture of trust and psychological safety, enabling educators to take intellectual risks and innovate in their teaching approaches (Bryk and Schneider, 2002).
In addition, the intellectual curiosity stimulated by TL behaviors could encourage school actors to connect with their external environment, engaging stakeholders, community members and broader educational trends (Shaked and Schecter, 2020). As such, transformational leaders play an essential role in encouraging organizational members to tap into external resources to achieve organizational goals (Wu and Le, 2020). Overall, transformational school leaders ensure that learning is deeply embedded in the school’s culture and systems, enabling sustained improvement and innovation (Hsiao and Chang, 2011).
Missing moderator: leader self-efficacy
Self-efficacy, defined as a belief in the ability to perform tasks and achieve desired outcomes effectively (Bandura and Wessel, 1997), plays a pivotal role in shaping leadership performance (Machida and Schaubroeck, 2011). In educational settings, self-efficacy perception transcends being a mere personal trait; it functions as a dynamic force that influences how school leaders navigate challenges, inspire their teams and drive organizational success (Ugwuanyi and Pietsch, 2024). Leaders with high self-efficacy approach their roles with confidence (Özdemir et al., 2024), enabling them to make innovative decisions even in the face of uncertainty (Afsar and Masood, 2017). This confidence empowers school leaders to persevere through setbacks, maintain focus on long-term school goals and foster resilience within their schools (Röhl et al., 2022). As transformational leaders generally effectively implement these practices, their self-efficacy might be reinforced, creating a positive feedback loop that further enhances their ability to lead effectively (Hesbol, 2019).
However, the relationship between self-efficacy and leadership performance is not unilaterally straightforward. Machida and Schaubroeck (2011) contend that leaders with excessively high self-efficacy may perceive little need to invest time and energy in developing their leadership abilities, leading to a decline in their perceived performance. Conversely, leaders with low self-efficacy may face significant challenges in developing their skills, though incremental increases toward moderate levels and beyond can serve as a leverage point for growth (Lindsey et al., 1995). Moderate levels of self-efficacy thus represent an optimal baseline for work engagement, enhancing TL performance without the pitfalls of overconfidence (Qiao et al., 2025).
Present study
Building on the outlined conceptual framework, the present study includes four hypotheses (H1–4). First, all dimensions of TL are positively associated with OL at the individual (teacher) level (H1). School leaders’ self-efficacy may act as a moderating factor, either amplifying or diminishing the effect of TL on OL within the schools (H2). Similarly, schoolwide perceived TL could moderate the relationship between TL and OL at the individual level (H3). Lastly, school leaders who achieve a balance between their self-reported efficacy and schoolwide perceived TL are better positioned to foster collaborative learning cultures (H4). This balance enables schools to adapt and innovate effectively, creating an environment where OL thrives. Conversely, leaders with disproportionately high self-efficacy relative to their teacher-perceived TL performance may struggle to inspire or sustain these efforts.
By examining these four hypotheses, this study aims to provide a nuanced understanding of the mechanisms through which leader self-efficacy and TL influence OL in K-12 school settings. As shown in Figure 1, these relationships are tested in a school-dependent and context-sensitive way by a multilevel analytical model, reflecting the nuanced interplay between leadership self-efficacy and teacher-perceived TL performance across schools.
The model is divided into two levels. The top level is labeled “Between-school level”, and the bottom level is labeled “Within-school level”, separated by a horizontal thick line. At the top level labeled “Between-school level”, three text boxes are shown. The text box on the left is labeled “Schoolwide Perceived T L Performance”, the one on the right is labeled “Leader Self-Efficacy”, and the third, positioned between them slightly below, is labeled “T L Performance cross Self-Efficacy”. A downward dashed arrow connects “Schoolwide Perceived T L Performance” to “T L Performance cross Self-Efficacy”, and a dashed downward arrow connects “Leader Self-Efficacy” to “T L Performance cross Self-Efficacy”. On the right of these text boxes, a circle labeled “Org. Learning” is positioned. Above “Org. Learning”, four horizontally arranged text boxes are labeled from left to right as “Principal Gender”, “Principal Age”, “Principal Education”, and “Leadership Experience”. Four individual dashed arrows emerge from these text boxes and point to “Org. Learning”, labeled as “0.11 (0.06)”, “0.01 (0.01)”, “negative 0.26 (0.23)”, and “0.01 (0.01)” respectively. A solid inward arrow also connects to “Org. Learning” from the right. In the bottom level labeled “Within-school level”, a large circle labeled “Organizational Learning” is positioned at the center. On the right of “Organizational Learning”, seven vertically arranged text boxes are labeled from top to bottom as “Continuous Learning”, “Inquiry and Dialogue”, “Team Learning”, “Empowerment”, “System Connections”, “Embedded Systems”, and “Strategic Leadership”. On the left of “Organizational Learning”, five vertically arranged text boxes are labeled from top to bottom as “T L - I S”, “T L - I C”, “T L - I M”, “T L - I B”, and “T L - I A”. Four solid rightward arrows emerge from “T L - I S”, “T L - I C”, “T L - I M”, and “T L - I B” and point to “Organizational Learning”. These arrows are labeled “0.22 double asterisk (0.04)”, “0.14 double asterisk (0.04)”, “0.12 double asterisk (0.004)”, and “0.16 single asterisk (0.06)” respectively. A dashed rightward arrow labeled “0.05 (0.05)” emerges from “T L - I A” and points to “Organizational Learning”. Below “Organizational Learning”, four horizontally arranged text boxes are labeled from left to right as “Teacher Gender”, “Teacher Age”, “Teacher Education”, and “Teaching Experience”. Each of these text boxes connects to “Organizational Learning” with upward arrows labeled respectively as “0.01 (0.03)”, “negative 0.01 (0.00)”, “0.00 (0.02)”, and “0.00 (0.00)”. Within the “Within-school level”, three solid downward arrows labeled “0.37 double asterisk (0.12)”, “0.50 asterisk (0.27)”, and “negative 0.24 (0.11)” emerge from the top-level text boxes “Schoolwide Perceived T L Performance”, “T L Performance cross Self-Efficacy”, and “Leader Self-Efficacy”, respectively. These three arrows point downward to the arrow connecting “T L - I S” to “Organizational Learning”.Conceptual model for cross-level moderation. Note: * indicates statistical significance at the 0.05 level (p < 0.05); ** at the 0.01 level (p < 0.01); ✝ denotes marginal significance (p = 0.06). Dashed lines indicate interaction terms, created as the product of schoolwide perceived TL dimensions and principal’s self-efficacy variables. Dotted lines indicate statistically non-significant relations (p > 0.10). Non-significant between-school predictors for the five dimensions of school-wide perceived TL performance on OL culture, as well as non-significant main and interaction effects at the within-school level, and error terms were excluded from the model for clarity. Source: Authors’ own creation
The model is divided into two levels. The top level is labeled “Between-school level”, and the bottom level is labeled “Within-school level”, separated by a horizontal thick line. At the top level labeled “Between-school level”, three text boxes are shown. The text box on the left is labeled “Schoolwide Perceived T L Performance”, the one on the right is labeled “Leader Self-Efficacy”, and the third, positioned between them slightly below, is labeled “T L Performance cross Self-Efficacy”. A downward dashed arrow connects “Schoolwide Perceived T L Performance” to “T L Performance cross Self-Efficacy”, and a dashed downward arrow connects “Leader Self-Efficacy” to “T L Performance cross Self-Efficacy”. On the right of these text boxes, a circle labeled “Org. Learning” is positioned. Above “Org. Learning”, four horizontally arranged text boxes are labeled from left to right as “Principal Gender”, “Principal Age”, “Principal Education”, and “Leadership Experience”. Four individual dashed arrows emerge from these text boxes and point to “Org. Learning”, labeled as “0.11 (0.06)”, “0.01 (0.01)”, “negative 0.26 (0.23)”, and “0.01 (0.01)” respectively. A solid inward arrow also connects to “Org. Learning” from the right. In the bottom level labeled “Within-school level”, a large circle labeled “Organizational Learning” is positioned at the center. On the right of “Organizational Learning”, seven vertically arranged text boxes are labeled from top to bottom as “Continuous Learning”, “Inquiry and Dialogue”, “Team Learning”, “Empowerment”, “System Connections”, “Embedded Systems”, and “Strategic Leadership”. On the left of “Organizational Learning”, five vertically arranged text boxes are labeled from top to bottom as “T L - I S”, “T L - I C”, “T L - I M”, “T L - I B”, and “T L - I A”. Four solid rightward arrows emerge from “T L - I S”, “T L - I C”, “T L - I M”, and “T L - I B” and point to “Organizational Learning”. These arrows are labeled “0.22 double asterisk (0.04)”, “0.14 double asterisk (0.04)”, “0.12 double asterisk (0.004)”, and “0.16 single asterisk (0.06)” respectively. A dashed rightward arrow labeled “0.05 (0.05)” emerges from “T L - I A” and points to “Organizational Learning”. Below “Organizational Learning”, four horizontally arranged text boxes are labeled from left to right as “Teacher Gender”, “Teacher Age”, “Teacher Education”, and “Teaching Experience”. Each of these text boxes connects to “Organizational Learning” with upward arrows labeled respectively as “0.01 (0.03)”, “negative 0.01 (0.00)”, “0.00 (0.02)”, and “0.00 (0.00)”. Within the “Within-school level”, three solid downward arrows labeled “0.37 double asterisk (0.12)”, “0.50 asterisk (0.27)”, and “negative 0.24 (0.11)” emerge from the top-level text boxes “Schoolwide Perceived T L Performance”, “T L Performance cross Self-Efficacy”, and “Leader Self-Efficacy”, respectively. These three arrows point downward to the arrow connecting “T L - I S” to “Organizational Learning”.Conceptual model for cross-level moderation. Note: * indicates statistical significance at the 0.05 level (p < 0.05); ** at the 0.01 level (p < 0.01); ✝ denotes marginal significance (p = 0.06). Dashed lines indicate interaction terms, created as the product of schoolwide perceived TL dimensions and principal’s self-efficacy variables. Dotted lines indicate statistically non-significant relations (p > 0.10). Non-significant between-school predictors for the five dimensions of school-wide perceived TL performance on OL culture, as well as non-significant main and interaction effects at the within-school level, and error terms were excluded from the model for clarity. Source: Authors’ own creation
Methods
This study employs a mixed-effects multilevel model to analyze data at both the individual (teacher) and school levels, focusing on the cross-level effects of school principals’ self-efficacy, schoolwide perceived TL performance and their interaction. Following Heisig and Schaeffer’s (2019) recommendation that cross-level interactions should always include random slopes on the lower-level components of those interactions, we hierarchically test three dependent analytical models. First, the null model establishes baseline regression terms, indicating the magnitude of variance that can be explained across schools. Second, the random slope model controls for between-school variance in regression terms. Finally, the cross-level moderation model examines the moderating effect of principal self-efficacy on the relationship between TL and OL.
Sample and data collection procedure
Data were collected as part of a research and development project aimed at the leadership development of school leaders in the Hong Kong SAR. Before data collection, ethical approval was obtained, and an informative letter outlining the project was sent to local schools. Consequently, 38 schools that showed interest in participation were invited to complete the initial teacher and principal surveys, indicating a non-probability sampling method. During the data collection process, six principals did not complete the questionnaires, and in one school, none of the teachers responded. As a result, these seven schools were excluded from the study.
Of the remaining 31 schools with a total of 1,792 teachers, 1,424 teachers completed the survey, yielding a response rate of 79.46%. After applying listwise deletion to remove missing and incomplete responses, the final dataset comprised 1,080 teachers and 31 principals. This resulted in an average of 34.84 observations per school, thus meeting Kreft and de Leeuw’s (1998) recommendation of having 30 Level 1 observations within 30 Level 2 units, namely the 30/30 rule. Among participating teachers, 359 were male and 721 were female. In terms of education, 620 teachers held undergraduate degrees, while 460 held master’s or higher degrees. The average age of teachers was 38.40 (SD = 10.21), with an average teaching experience of 13.27 years (SD = 9.95). Among the participating 31 principals, 14 were male and 17 were female. The majority held postgraduate degrees, with 26 holding master’s degrees, three holding doctoral degrees and two holding undergraduate degrees. The average age of principals was 50.61, with an average teaching experience of 26.87 years and an average leadership experience of 15.03 years.
Measurement instruments
Two tailored research surveys were administered to principals and teachers separately, each incorporating well-established scales with robust psychometric properties. The teacher survey included a demographic questionnaire capturing gender, age and years of teaching experience, followed by the Multifactor Leadership Questionnaire 5× (MLQ 5×; Bass and Avolio, 2004), a widely validated instrument designed to assess TL behaviors (Antonakis et al., 2003). The five-point response scale of MLQ 5× measures the frequency of TL behaviors, with options ranging from 1 (Not at all) to 5 (Frequently, if not always). In the present study, MLQ 5× demonstrated strong reliability, with Cronbach’s α values ranging from 0.84 to 0.89. The construct validity of the MLQ 5× has been supported in multiple studies, with significant factor loadings (Hemsworth et al., 2013; Sosik et al., 2002) and good fit indices (Batista-Foguet et al., 2021; Lee et al., 2011). In the current study, construct validity was further supported, with all significant factor loadings ranging from 0.65 to 0.87 and satisfactory fit indices (NFI = 0.94; CFI = 0.95; RMSEA = 0.076).
Additionally, the Dimensions of Learning Organization Questionnaire (DLOQ; Watkins and Marsick, 1997) was used to measure OL practices, yielding satisfactory internal consistency values of Cronbach’s α between 0.83 and 0.89. These values align with prior research, affirming the DLOQ’s reliability (Goula et al., 2020). The Likert-type DLOQ scale assesses the respondents’ perceived OL culture, ranging from 1 (Never) to 5 (Always). A cross-cultural meta-analysis demonstrated its construct validity across diverse contexts (Watkins and Dirani, 2013). In the current study, the original seven-factor measurement model of the DLOQ was retested and further validated, with all factor loadings between 0.75 and 0.87, as well as good model fit indices (NFI = 0.96, CFI = 0.97, RMSEA = 0.055).
The principal survey included the General Self-Efficacy Scale (GSE; Schwarzer and Jerusalem, 2010), a psychometrically sound instrument showing a high reliability score in the study (Cronbach’s α = 0.88), along with demographics. The four-point scale options of the GSE, ranging from 1 (Not at all true) to 4 (Exactly true), measure the respondents’ perceived self-efficacy. High construct validity of the GSE composite score has been cross-culturally confirmed (NFI = 0.97, GFI = 0.97, RMSEA = 0.05) (Scholz et al., 2002). In the current study, the overall factor structure of the GSE was also confirmed, with all factor loadings ranging from 0.31 to 0.91, and acceptable model fit indices (NFI = 0.96, CFI = 0.96, cf. RMSEA = 0.14). Together, these instruments provide a comprehensive and reliable framework for assessing the study’s key constructs at both individual and organizational levels.
To statistically control for potential confounding factors, the present study included several teacher and principal demographic variables (age, gender, education level, teaching experience and leadership experience – only for principals) as control variables in the analysis. Previous relevant studies have commonly controlled for teacher gender, age and education level due to their potential confounding effects, though findings regarding their significance have been mixed (Banoğlu et al., 2022; Mulford et al., 2004; Pogodzinski, 2014). Additionally, years of experience in teaching and leadership positions were controlled for due to their likely relevance to perceived leadership performance and the level of OL (Patrick, 2022; Schechter and Qadach, 2011). Detailed analytical models are provided in Appendix A.
Results
Key findings from three regression models, a null model, a random slope model and a cross-level interaction model, are presented in Table 1.
Mixed-effects multilevel regression models
| Parameters | Null model | Random slope model | Cross-level effect model | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Estimate | SE | Sig | Estimate | SE | Sig | Estimate | SE | Sig | |
| Within-school level | |||||||||
| Transformational leadership | |||||||||
| Intellectual stimulation (IS) | 0.21** | 0.03 | 0.00 | 0.22** | 0.04 | 0.00 | |||
| Cross-level: Schoolwide teacher perceived IS | 0.37** | 0.12 | 0.00 | ||||||
| Cross-level: Princiapal self-efficacy (SE) | −0.24* | 0.11 | 0.03 | ||||||
| Cross-level: IS × SE interaction | 0.50✝ | 0.27 | 0.06 | ||||||
| Idealized Influence Behavior (IB) | 0.13* | 0.05 | 0.02 | 0.16* | 0.06 | 0.01 | |||
| Cross-level: Schoolwide teacher perceived IB | 0.06 | 0.19 | 0.77 | ||||||
| Cross-level: Princiapal SE | 0.08 | 0.12 | 0.51 | ||||||
| Cross-level: IB × SE interaction | −0.96 | 0.63 | 0.13 | ||||||
| Inspirational Motivation (IM) | 0.13** | 0.04 | 0.00 | 0.12** | 0.04 | 0.00 | |||
| Cross-level: Schoolwide teacher perceived IM | −0.09 | 0.13 | 0.48 | ||||||
| Cross-level: Princiapal SE | 0.07 | 0.11 | 0.53 | ||||||
| Cross-level: IM × SE interaction | −0.16 | 0.40 | 0.69 | ||||||
| Idealized influence attribute (IA) | 0.06 | 0.04 | 0.16 | 0.05 | 0.05 | 0.29 | |||
| Cross-level: Schoolwide teacher perceived IA | −0.29 | 0.20 | 0.15 | ||||||
| Cross-level: Princiapal SE | −0.08 | 0.14 | 0.58 | ||||||
| Cross-level: IA × SE interaction | 0.70 | 0.61 | 0.26 | ||||||
| Individual consideration (IC) | 0.19** | 0.03 | 0.00 | 0.14** | 0.04 | 0.00 | |||
| Cross-level: Schoolwide teacher perceived IC | 0.04 | 0.11 | 0.70 | ||||||
| Cross-level: Princiapal SE | 0.11 | 0.14 | 0.43 | ||||||
| Cross-level: IC × SE interaction | −0.02 | 0.44 | 0.97 | ||||||
| Control variables | |||||||||
| Teacher gender | 0.01 | 0.03 | 0.84 | ||||||
| Teacher age | −0.01* | 0.00 | 0.02 | ||||||
| Teacher education level | 0.00 | 0.02 | 0.91 | ||||||
| Teacher experience in years | 0.00 | 0.00 | 0.86 | ||||||
| Between-school level | |||||||||
| Transformational leadership | |||||||||
| Intellectual stimulation | 0.30 | 0.33 | 0.92 | ||||||
| Idealized influence behavior | 0.12 | 0.44 | 0.79 | ||||||
| Inspirational motivation | −0.08 | 0.19 | 0.69 | ||||||
| Idealized Influence Attribute | 0.39 | 0.32 | 0.21 | ||||||
| Individual Consideration | −0.01 | 0.29 | 0.96 | ||||||
| Control variables | |||||||||
| Principal gender | 0.11 | 0.06 | 0.08 | ||||||
| Principal age | 0.00 | 0.01 | 0.99 | ||||||
| Principal education level | −0.26 | 0.23 | 0.26 | ||||||
| Principal teaching experience in years | 0.00 | 0.01 | 0.88 | ||||||
| Principal leadership experience in years | 0.00 | 0.00 | 0.83 | ||||||
| ICC | 0.13 | ||||||||
| Within variance/Residual | 0.36** | 0.02 | 0.00 | 0.13** | 0.01 | 0.00 | 0.13** | 0.01 | 0.00 |
| Intercept (Between variance) | 0.05** | 0.02 | 0.00 | 0.07** | 0.02 | 0.00 | 0.02✝ | 0.01 | 0.08 |
Note(s): **<0.001; *<0.05; ✝<0.10
The ICC confirms that 13% of the variance in OL is attributable to between-school differences, with moderate reliability of school-level means. The within-school variance decreased from 0.36 (p < 0.001) in the null model to 0.13 (p < 0.001) in the cross-level interaction model, indicating that the inclusion of teacher- and school-level predictors significantly explains the within-school variance in OL. Nevertheless, the between-school variance decreased from 0.05 (p < 0.001) in the null model to 0.02 (p = 0.08) in the cross-level interaction model, suggesting that school-level factors account for a smaller proportion of the variance compared to teacher-level factors. Below, the findings are presented in alignment with the study’s two research questions.
Relationship between TL and the OL
Bivariate correlation values for the TL and OL variables are provided in Appendix B. While these correlations offer an initial overview of the associations among variables, the primary focus of RQ1 lies in examining how these relationships vary across schools. Accordingly, the multilevel analysis results reveal significant variation in the relationship between TL dimensions and OL across schools. Intellectual stimulation (IS) consistently demonstrated a strong positive association with OL in both the random slope model (b = 0.21, p < 0.001) and the cross-level interaction model (b = 0.22, p < 0.001). Similarly, idealized influence behavior (IB) showed significant positive effects in the random slope model (b = 0.13, p = 0.02) and the cross-level interaction model (b = 0.16, p = 0.01). Inspirational motivation (IM) and individual consideration (IC) also exhibited significant positive relationships in both models (IM: b = 0.13, p < 0.001; IC: b = 0.19, p < 0.001). However, while controlling for other TL behaviors, the idealized influence attribute (IA) did not show significant effects in any model. Findings suggest that IS, IB, IM and IC play a critical role in fostering OL, with their effects varying across schools, partially confirming the H1.
Moderating roles of self-efficacy and leadership performance.
The cross-level interaction model further examines how principals’ self-efficacy and their leadership performance, as perceived schoolwide by teachers, moderate the relationship between TL and OL. Figure 1 illustrates the proposed cross-level moderation model with the obtained statistics.
Specifically, the schoolwide mean of perceived IS per se had a significant positive effect (b = 0.37, p < 0.001), while principals’ SE per se demonstrated a significant negative effect (b = −0.24, p = 0.03) on the strength of the relationship between IS and the OL culture as perceived by individual teachers. In other words, these findings suggest that stronger schoolwide perceptions of IS amplify the relevance of IS to OL at the teacher level, whereas higher levels of principal SE appear to weaken this relationship. Therefore, both H2 and H3 were confirmed, but only for the IS dimension of TL. Notably, the intensifying effect of schoolwide IS performance is observed when principals’ self-efficacy is assumed to be at average levels relative to their peers. Similarly, the diminishing effect of SE is conditional upon being perceived as average in TL performance by teachers within the respective schools.
This finding highlights a critical issue: an autonomous increase in leader self-efficacy, when not accompanied by a corresponding increase in schoolwide perceived TL performance, could have a detrimental impact. However, increased schoolwide IS performance, holding the leader SE constant, further amplifies the positive relationship between IS and OL. This raises a question of what happens when both self-reported self-efficacy and teacher-perceived TL performance increase together. The results show that the interaction between mean IS and principal self-efficacy approaches significance (b = 0.50, p = 0.06). This suggests that the relationship between IS and OL is stronger in schools where principals exhibit higher self-efficacy and where IS is perceived more positively by teachers, confirming the H4. In practical terms, a one-point increase in the interaction term corresponds to a 0.63-point increase in teachers’ endorsement of OL culture within their school (see Appendix C).
For other dimensions of TL, such as IB, IM, IA and IC, neither schoolwide perceived leadership performance nor principal self-efficacy showed significant moderating effects and, thus, they are not shown in Figure 1. These results underscore the nuanced role of principal self-efficacy and schoolwide perceptions in shaping the impact of IS on OL, while indicating limited moderating effects for other TL dimensions.
Additionally, at the teacher level, teacher age had a statistically significant but substantively negligible adverse effect on OL (b = −0.01, p = 0.02), while teacher gender, education level and experience were inconclusive. At the school level, principal gender approached a marginally significant increase (b = 0.11, p = 0.08), but other principal-level controls, including age, education level, teaching experience and leadership experience, were not significant.
Discussion
The findings of this study weave valuable insights into the multifaceted nature of TL and its effects on OL culture, particularly through the intricate moderation of principal self-efficacy and their schoolwide perceived leadership performance – a relationship that could not be adequately examined in a unilevel, context-free research design. By employing a mixed-effects multilevel modeling approach, the current study underscores the variability of leadership’s impact across different school contexts and illuminates the subtle interplay between leaders’ internal confidence in their general abilities and their external effectiveness as perceived by teachers. These results invite a deeper exploration of how school leadership practices, when harmonized with contextual realities and collective perceptions, can catalyze meaningful organizational growth.
TL serves as both a charismatic and visionary model for teacher groups and a means of individual teacher empowerment (Windlinger et al., 2019). This dual lens underscores the necessity of adaptive leadership behaviors through a context-dependent approach, attuned to the unique dynamics of individual teacher perceptions pertaining to each school’s own organizational ecosystem (Kark and Shamir, 2013; Wang et al., 2010). Our findings reveals that school leaders who skillfully navigate these contextual dynamics – reaching more teachers through their TL practices – are better positioned to foster environments where a culture of OL thrives (Leithwood and Sun, 2012).
Central to the findings from the study’s random-effects model is the diverse influence of school context on how teachers perceive and respond to the principal’s various TL behaviors. The dimensions of intellectual stimulation, idealized influence behavior, inspirational motivation and individual consideration all demonstrate positive associations with OL culture across schools (Alanoglu and Karabatak, 2025; Wilson Heenan et al., 2023). However, their individual effects are far from uniform (Windlinger et al., 2019). This between-school variability prompted us to employ a cross-level effect model to examine the moderating role of the complex interplay between leader self-efficacy and schoolwide perceived leadership performance.
Consequently, the study disclosed the pivotal role of intellectual stimulation in shaping teacher perceptions and, thereby, bridging TL practices to OL outcomes. This might be because intellectual stimulation inherently requires ongoing dialogue, trust and openness to new ideas (Liao et al., 2017). However, unlike other dimensions of TL, the effectiveness of intellectual stimulation is more deeply connected to the collective perception of the principal’s leadership credibility and self-perceived efficacy (Khan et al., 2022; Lee and Kuo, 2019). Therefore, we conclude that a strong collective endorsement of principal leadership and alignment with self-efficacy is particularly crucial for intellectual stimulation to translate into a vibrant OL culture.
In schools where principals are collectively perceived as providing strong intellectual stimulation, the transformative potential of school leadership on OL culture is more likely to flourish (Mulford and Silins, 2009). Conversely, in school environments marked by instability in teacher perceptions of the principal’s TL performance, these same practices may falter (Hauserman and Stick, 2013). When leaders are seen as effective in challenging their teams to think creatively and critically, the ripple effects on OL are profound (Liao et al., 2017). Yet, this relationship is not merely a function of what leaders do; it is also shaped by how their actions are interpreted and internalized by teachers as part of the school culture (Gill et al., 2018).
Here, the alignment between principals’ self-efficacy and teachers’ perceptions becomes critical for better understanding how TL may influence OL culture in schools (Hesbol, 2019). The detrimental impact of low leader efficacy on perceived leadership performance has been theoretically well-established by Bandura and Wessels (1997) and extensively discussed in the leadership literature (Machida and Schaubroeck, 2011). In the present study, we extend these findings to the field of educational leadership by highlighting that school leaders who exude confidence but fail to translate it into their intellectual stimulation practices – as perceived by teachers – risk undermining their own leadership effectiveness. This misalignment underscores the delicate balance leaders must strike between their internal belief in their capabilities and their external capacity to inspire and engage their teams.
Given the significant interaction terms in the cross-level model, the study also unveils a paradox in the role of principal self-efficacy. Moderate levels of self-efficacy sustain the positive effect of TL on OL culture, especially when teachers, on average, perceive their principal’s TL practices more positively compared to those in other schools. This finding aligns with the efficacy-spiral argument, where moderate self-efficacy serves as a baseline for understanding whether leadership development will follow an upward or downward trajectory (Lindsley et al., 1995). Our results also contend that high confidence can paradoxically dilute the impact of intellectual stimulation on OL culture when it was not endorsed by schoolwide perceived leadership performance. This finding invites a reexamination of the interplay between confidence and humility in leadership, particularly in terms of fostering intellectual stimulation.
The results obtained from this research hold dual significance for the existing body of leadership development theory. Firstly, in relation to the first hypothesis, the findings affirm theoretical propositions that the impact of school leadership on teachers’ OL culture is highly context-dependent, shaped by teachers’ individual sense of collectivity and varying according to teacher perceptions across schools (Schechter and Qadach, 2011; Schechter et al., 2022). More provocatively, findings associated with the remaining hypotheses indicate that the intellectual stimulation of teachers by school leaders emerges as a key driver in promoting OL culture, particularly when the teachers widely recognize it.
To illustrate, in a school led by a self-confident principal whose intellectually stimulating leadership is recognized not only by the individual respondents but also by most other teachers, the individual teachers perceive a stronger OL culture in their school. In contrast, a school led by an overconfident principal who similarly attempts to stimulate intellectual engagement, but most teachers do not perceive the principal’s efforts as effective, even those who personally rated the principal’s intellectual stimulation higher may not experience an enhanced sense of OL culture. This underscores the importance of both principal self-efficacy and the broader staff consensus in translating intellectual stimulation into a thriving OL culture. This set of evidence constitutes a potentially original contribution to the literature, highlighting the critical role of humility and self-awareness in effective leadership practices and in fostering sustainable school improvement.
Leaders who are overly assured in their abilities may inadvertently become less receptive to feedback, less adaptable to change and less attuned to the collective needs of their teams, which can lead to low performance (Guillen, 2020). However, school leaders who balance self-assurance with a willingness to learn and develop their TL practices are better equipped to foster school environments where OL flourishes (Hesbol, 2019). This duality highlights the cyclical nature of self-efficacy and performance: while self-efficacy fuels performance to some extent, it is balanced by the perceived organizational context (Bachrach et al., 2022). Interventions aimed at enhancing self-efficacy, therefore, must focus on creating opportunities for school leaders to build confidence through meaningful, incremental achievements that reinforce their sense of competence without fostering overconfidence (Röhl et al., 2022; Ugwuanyi and Pietsch, 2024).
Together, these insights portray the hardwired interaction of multiple psychosocial factors through which TL drives OL culture in schools. They underscore the importance of school context, the centrality of schoolwide perceived leadership performance and the delicate balance between confidence and leadership performance. As the field of school leadership continues to evolve, these findings offer a compelling foundation for future research and practice, inviting school leaders to reflect on their own practices and the ways in which they can cultivate school environments where a culture of collective learning can thrive.
To support the relationship between school leadership and OL culture, particularly considering the moderating effect of principal self-efficacy and schoolwide leadership perceptions, several actionable strategies can be considered. Introducing 360-degree feedback processes enables principals to gather diverse perspectives on their leadership practices and adjust their approach accordingly. Feedback derived from internal evaluation procedures may help principals strike a balance between their self-efficacy and actual leadership performance. Beyond providing metacognitive support for principals’ self-evaluation, involving teachers in shared decision-making stimulates intellectual engagement and empowers teachers to engage in collective learning mechanisms, thereby strengthening the schoolwide perceived OL culture. Such internal evaluation mechanisms help ensure that school leaders remain adaptive and open to continuous learning, thus maximizing the positive impact of their leadership.
These findings also invite more research on the role of self-efficacy in leadership performance and its broader outcomes. While the current study employed a cross-sectional design, future longitudinal research would be valuable to uncover the dynamic and potentially causal mechanisms underlying these relationships. Although our dataset was robust and our analyses were rigorous, the reliance on non-probability sampling may have limited the representativeness of the findings. Thus, future studies should consider larger, more diverse, and representative samples across various educational contexts to enhance generalizability. Additionally, the present study’s findings should be interpreted with caution, given the cultural specificity of the Hong Kong educational context.
The supplementary material for this article can be found online.
