國小校長教學領導與教師教學實踐關聯之多層次分析：檢驗教師自我效能的中介效果與不同國家的調節效

A Multilevel Analysis of the Relationship between Primary School Principals' Instructional Leadership and Teachers' Instructional Practices: Examining the Mediating Role of Teacher Self-Efficacy and the Moderating Effect of Country Differences

校長的領導風格對學校辦學成效具有關鍵影響，而教師的教學實踐則直接關係到學生的學習成果。作為提升學校教學品質與學生學習成效的重要因素，校長教學領導如何影響教師的自我效能與教學實踐，成為教育研究的重要議題。本研究利用2018年教學與學習國際調查（TALIS）臺灣、日本及澳洲三國共364所國小校長與9，832位教師的資料，透過多層次中介效果模型，探討校長教學領導、學校教師自我效能與教師自我效能對教師教學實踐的影響。同時，考量不同國家的學校文化與學生特性可能影響此關係，研究亦納入國家變數進行調節效果檢驗。研究結果顯示，教師層次的教師自我效能對教師教學實踐具有顯著正向影響，而學校層次的學校教師自我效能亦展現顯著的脈絡效果。此外，校長教學領導透過教師自我效能的部分中介對教師教學實踐產生間接影響。然而，國家變數未對校長教學領導與學校教師自我效能的關係產生顯著調節效果，顯示三國在此機制上的一致性。研究結果進一步驗證校長教學領導在促進教師自我效能與教學實踐方面的關鍵作用。最後，本文針對研究發現進行理論與實務層面的討論，並對未來研究方向提出建議。

Principal instructional leadership has long been regarded as one of the most influential forms of school leadership for improving teaching quality and enhancing student learning. In contemporary research on school effectiveness, principals are no longer viewed merely as administrative managers responsible for organizational maintenance, but increasingly as key instructional leaders who shape the professional climate of schools, guide curriculum and instructional improvement, and promote teachers' continuous professional growth. Through setting instructional goals, monitoring teaching quality, allocating resources for professional development, and fostering a culture of collaboration, principals are expected to exert a profound influence on the conditions under which teaching and learning take place. At the same time, teachers' teaching practices constitute the core processes of classroom instruction and represent the most proximal determinants of students' learning experiences and academic outcomes. Extensive research in educational psychology has consistently shown that variations in instructional clarity, cognitive activation, and classroom management are closely associated with students' cognitive engagement, motivation, and achievement. Understanding how principals' instructional leadership is translated into effective classroom practices therefore remains a central issue at the intersection of educational leadership and educational psychology. Yet, the mechanisms through which leadership at the school level is transmitted to instructional behavior at the classroom level remain incompletely understood. Within this line of inquiry, teacher self-efficacy, grounded in Bandura's social cognitive theory, may be a key psychological mechanism linking leadership practices to teachers' instructional behaviors. Teacher self-efficacy refers to teachers' beliefs in their capability to plan, organize, and implement instructional activities that lead to desired student outcomes. These beliefs influence how teachers interpret instructional challenges, how much effort they invest in teaching, how persistently they cope with difficulties, and how resilient they remain in the face of setbacks. A substantial body of empirical research has demonstrated that teachers with higher self-efficacy are more likely to adopt effective instructional strategies, manage classrooms successfully, persist when encountering instructional problems, and promote students' cognitive and motivational development. Although previous studies have generally suggested that principals' instructional leadership may strengthen teachers' beliefs in their instructional capabilities by articulating a clear school vision, promoting curriculum development, and cultivating a supportive professional climate, teacher self-efficacy itself constitutes a key psychological resource that facilitates teachers' active use of diverse instructional strategies, effective classroom management, and the promotion of student engagement. From this perspective, teacher self-efficacy is theoretically well positioned to serve as an important mediating mechanism between principals' instructional leadership and teachers' instructional practices. That is, leadership practices may influence teaching not only directly, but also indirectly by shaping teachers' beliefs about their own professional competence. However, existing research has largely relied on single-level analyses focusing on teachers' individual perceptions, with limited attention to the nested structure of educational data and the cross-level processes through which school-level instructional leadership is transmitted to classroom-level teaching practices. By treating leadership, self-efficacy, and teaching practices as operating at a single level, prior studies have often conflated within-school and between-school effects and overlooked the dual role of teacher self-efficacy as both an individual belief and a shared school-level property. Moreover, few studies have formally tested cross-level mediation models that incorporate both individual- and contextual-level components of self-efficacy, and empirical evidence regarding the moderating role of national context remains scarce. As a result, it remains unclear whether and how instructional leadership operates similarly across educational systems characterized by different cultural, institutional, and accountability conditions, highlighting a critical gap that warrants multilevel and cross-national investigation. To address these gaps, the present study examined the cross-level relationships among principal instructional leadership, teacher self-efficacy, and teachers' teaching practices using large-scale international data. Drawing on the 2018 Teaching and Learning International Survey (TALIS) conducted by the Organisation for Economic Co-operation and Development (OECD), this study analyzed data from 9,832 primary school teachers nested within 364 schools in Taiwan, Japan, and Australia. These three countries were selected to represent contrasting cultural and institutional contexts: Taiwan and Japan as East Asian systems traditionally characterized by strong academic emphasis and relatively centralized governance, and Australia as a Western system emphasizing school autonomy and teacher professional discretion. The study pursued four main objectives. First, it examined the direct effect of school-level principal instructional leadership on teacher self-efficacy. Second, it tested the direct effect of teacher self-efficacy on teachers' instructional practices. Third, it investigated whether teacher self-efficacy mediated the relationship between principal instructional leadership and teachers' teaching practices within a multilevel framework. Fourth, it explored whether national context moderated the relationship between principal instructional leadership and teacher self-efficacy. Given the nested structure of the data, a multilevel 2-1-1 mediation model with an unconflated specification was employed. This modeling strategy allowed for a clear separation of within-school and between-school components of teacher selfefficacy and enabled a rigorous test of cross-level mediation from school-level instructional leadership to teacher-level teaching practices. Country was included as a school-level moderator in the leadership–efficacy path to examine potential cross-national differences. The results yielded several noteworthy findings. First, principal instructional leadership had a significant positive effect on teacher self-efficacy at both the individual and school levels. Schools in which principals reported higher levels of instructional leadership tended to exhibit higher average levels of teacher self-efficacy, and teachers within those schools also reported stronger personal efficacy beliefs. Second, teacher self-efficacy exerted a strong positive effect on overall teaching practices. Teachers with higher self-efficacy demonstrated more frequent and consistent use of effective instructional behaviors in their classrooms. Third, multilevel mediation analyses indicated that teacher self-efficacy partially mediated the effect of principal instructional leadership on teaching practices. Importantly, this mediation operated through both individual-level and schoollevel components of self-efficacy, with the between-school (contextual) component playing a particularly salient role. This finding suggests that teacher self-efficacy is not merely an individual psychological attribute, but also a collective property of schools that reflects shared professional confidence and norms, which in turn shape instructional practices. Contrary to theoretical expectations, the moderating effects of country were not statistically significant. The strength of the relationship between instructional leadership and teacher self-efficacy did not differ significantly across Taiwan, Japan, and Australia. This result suggests that, at least at the primary school level, the basic mechanism linking instructional leadership to teacher self-efficacy may be relatively robust across diverse cultural and institutional contexts. The present study makes several important contributions to the literature. Theoretically, it advances understanding of the leadership–instruction nexus by providing multilevel evidence for a cross-level mediation mechanism linking principal instructional leadership to classroom practices through teacher self-efficacy. Methodologically, it demonstrates the value of unconflated multilevel mediation models for disentangling within-school and between-school processes in educational research. Empirically, it offers rare cross-national evidence on the generalizability of instructional leadership effects, highlighting both the universality and the contextual boundaries of leadership influence. From a practical perspective, the findings underscore the importance of leadership practices that systematically strengthen teachers' professional confidence at both individual and collective levels. Leadership preparation and professional development programs should therefore emphasize strategies for building school-wide efficacy climates, such as fostering collaborative professional learning communities, providing formative instructional feedback, and supporting continuous professional growth. Future research is encouraged to extend this work by incorporating longitudinal designs, richer cultural indicators, and additional contextual moderators in order to further elucidate how leadership effects unfold across time and educational settings.