Introduction

Researchers have emphasized the importance of understanding the influence of school leadership in context (Tan, 2014, 2018; Tan et al., 2020; Berkovich, 2018; Day et al., 2009; Hallinger, 2018; Urick, 2016; Urick & Bowers, 2014). Indeed, schools have operated in multiple contexts that enable or circumscribe teaching and learning trajectories (Hallinger, 2018). School leadership and context can be related in two ways. First, context may shape leadership, such that leaders respond by adapting to their environment (Wasserman et al., 2010). For example, leaders can capitalize on the strengths of their schools, address areas for improvement, identify and exploit opportunities in parental engagement, and mitigate against external threats. Second, leaders may shape their context, such that they can envision a desired future and develop school conditions to realize their vision (Hendriks & Scheerens, 2013).

Research evidence tends to support the notion that context shapes school leadership instead of school leaders shaping their context (Leithwood et al., 2020). However, the influence of specific contexts on leadership practices remains unclear (Hallinger, 2018). The present study addresses this knowledge gap. Two research questions inform the study, namely, “What is the association between family and school contextual resources and the leadership practices of principals?” and “What is the association between the leadership practices of principals with the science learning of students?” Specifically, the science learning of students is the variable of interest given the educational and work opportunities and social mobility prospects in science, technology, engineering, and mathematics. Additionally, science learning is more dependent on the quality of school processes, including leadership, compared with other subjects such as reading (Reynolds et al., 2014).

Figure 1 provides the conceptual framework, which underscores the manner in which contexts shape the enactment of principal leadership and how principals then influence the learning attitudes and outcomes of students. In addition to principal leadership, family (e.g., SES) and sociocultural (e.g., Confucian values emphasizing learning) factors influence student learning outcomes, so these factors are included in the analysis.

Fig. 1
figure 1

Conceptual framework

Full size image

Literature review

The study first conducts a literature review on school contexts (i.e., the availability of family and school resources), school leadership functions, relationships between school context and leadership, influence of family and school resource availability on school leadership, and influence of school leadership on student learning.

Family and school resources

The study examines three variables pertaining to family and school resources. The first pertains to the proportion of qualified teachers (e.g., those with professional educational qualifications and subject specialism), which is important, because teachers are responsible for teaching and learning in the classroom. For example, Ingvarson and Rowley (2017) find that mathematics teachers from education systems with more robust quality assurance policies for teachers exhibited high levels of mastery of mathematics knowledge. Therefore, qualified teachers are more equipped to use evidence-based pedagogies to help students learn coherent, focused, and rigorous content (Urick et al., 2018). The second relates to the importance of the quality of the educational resources of schools, because these resources can be used to support student-centered pedagogies (Cohen et al., 2003). This notion is evident in the emphasis on effective, equitable, and efficient resource-funding models on the basis of multiple criteria such as improvement in the academic performance of students (David-Hadar, 2018). Lastly, the socioeconomic status (SES) of individuals refers to their relative access to valued economic, cultural, and social resources in society that contribute to valued life outcomes (Sirin, 2005). It is commonly measured using the level of education of parents, family income, occupational status of parents, and, increasingly, home resources (Sirin, 2005). School SES refers to the mean level of the SES of individual students in a school (Sirin, 2005). School SES is important because students with high SES obtain more learning opportunities, which benefit academic achievement (Schmidt et al., 2015). Conversely, research shows that low-SES students are less likely to develop interest in science and to pursue science-related career opportunities (Archer et al., 2015). School SES influences the science learning of students in many ways (Sun et al., 2012). Teachers with high-SES students receive more support from parents in reinforcing school learning; moreover, high-SES parents imbue their children with positive learning attitudes; and high-SES schools are perceived as more conducive for student learning (Bellibas & Liu, 2018).

The present study examines three variables of context, namely, teacher quality, quality of educational resources of schools, and school SES, compared with other variables. Teacher quality and the quality of the educational resources of schools are used as indicators of contextual conditions related to the school, because they represent distal sources of environmental opportunities or constraints that may shape school leadership (Tan et al., 2020). In contrast, processes, such as academic press, classroom climate, and instructional practice, are more proximal to classroom teaching and learning (Holzberger et al., 2020) and, hence, are likely to be shaped by school leadership. Additionally, examining teacher quality is important given that teachers exert the greatest influence on student learning within the school followed by school leaders (Leithwood et al., 2020). The present study examines school SES as a family-related variable of context, because decades of research since the Coleman Report (Coleman et al., 1966; Hanushek, 2016) generally find a strong positive association between school SES and the learning outcomes of students (Van Ewijk & Sleegers, 2010; Willms, 2010).

School leadership functions

Four leadership functions characterize the school practices of principals (Leithwood et al., 2006). First, principals galvanize the resources of the school to realize the desired vision, mission, and goals (Murphy & Torre, 2015). Second, principals manage instructional programs by leading teachers in teaching and learning, which promotes effective instructional practices and emphasize the holistic development of students (Hitt & Tucker, 2016). Third, principals facilitate the professional development of teachers by building on strengths and needs, which inculcate teacher responsibility for professional development, and by cultivating professional learning communities for teachers (Murphy, 2015; Opfer & Pedder, 2011). Lastly, principals empower teachers by promoting collaborative decision-making processes such as reviewing management practices and contributing to school improvement. A recent review by Leithwood et al. (2020) indicates that these four leadership practices characterize contemporary leadership practices.

Context and leadership

Numerous studies suggest that context shapes principal leadership; as such, principals adapt by responding to contextual opportunities and challenges (Hallinger, 2018; Leithwood et al., 2020; Wasserman et al., 2010). For example, Tan (2014) conducts an international study and finds that variables of family and school contexts (shortage of school information technology resources, school size, student academic failure, and parental academic pressure) were associated with the leadership focus of principals (academic focus and operational involvement). Urick and Bowers (2014) conduct an analysis of data from the Schools and Staffing Survey for 1999–2000 and demonstrate that principal and school contexts (school size, urbanicity, and performance accountability) predicted a typology of shared leadership between the principal and teachers. Furthermore, schools with small enrollments and those with more minority students were more likely to have balkanizing principal leadership (low-centralized leadership, and high teacher autonomy) and shared integrated leadership (instructional and other core leadership tasks), respectively (Urick, 2016). Goldring et al. (2008) report that principals spent more time on responsibilities involving instruction and student affairs instead of focusing on one if they led schools with less disadvantaged students but high levels of academic press and student engagement. Day et al. (2009) propose that heads of challenging schools (measured through low school improvement and low-SES students) emphasized policies for pupil learning, teaching standards, physical environments, improvements in teaching and learning, and establishing positive school cultures.

The influence of family and school resources on leadership

The literature review then clarifies the influence of family and school resources, as specific characteristics of context, on school leadership practices. First, principals that lead schools with qualified teachers may find that implementing school plans emerging from envisioning exercises is easier, because these teachers have the capacity to realize the vision and goals of the school (Notman & Henry, 2011). Similarly, these principals can concentrate on instructional management, because qualified teachers are more equipped to use student-centered pedagogy. Furthermore, principals can empower teachers to use their professional knowledge for school improvement. In the area of science, principals can focus on instructional management and teacher empowerment when they have qualified science teachers to support them. For example, Lochmiller and Acker-Hocevar (2016) find that principals in public high schools in the United States who lack content knowledge in mathematics and science resorted to hiring teachers who could teach effectively and work collaboratively, supporting teacher collaboration, and providing professional development. Principals that lead schools with qualified teachers may also enhance teacher capacity by providing continued professional development to enable teachers to use student-centered pedagogy.

Next, principals leading well-resourced schools are motivated to establish and communicate the school vision and goals due to having adequate resources for realizing these ambitions. Indeed, Murphy and Torre (2015) maintain that school vision should be infused into key aspects of the school organization, including budgets, operating procedures, structures, and policies. Principals leading well-resourced schools can also easily focus on instructional management, because the resources for implementing innovative pedagogy are available (Cohen et al., 2003). Chang et al.’s (2008) study on Taiwanese elementary schools exemplifies the need for school resources for instructional plans. The authors report that shortage in budget and technological facility hindered the implementation of school plans for technology-enabled instruction. Student-centered science education is more resource-intensive compared with other subjects; thus, the realization of the academic vision and goals and the implementation of instructional management initiatives are dependent on the availability of state-of-the-art school resources for science teaching. The availability of school resources also enables principals to leverage on teacher empowerment and professional development, such that teachers can effectively use these resources for student learning, particularly in an esoteric subject such as science, where principals may lack up-to-date content knowledge (Lochmiller, 2016; Lochmiller & Acker-Hocevar, 2016).

Principals leading high-SES schools are likely to envision academic goals, because high-SES parents have high academic expectations. For example, the academic expectations of parents from schools may predict the academic focus and operational involvement of principals (Tan, 2014). Similarly, principals leading high-SES schools find that implementing instructional management is easy, because high-SES parents are more equipped to support the learning of their children (Notman & Henry, 2011). However, the relationship between school SES and principal leadership, which involves the empowerment and professional development of teachers, remains unclear. On the one hand, teachers in high-SES schools may be more competent and motivated compared with their peers in low-SES schools (Clotfelter et al., 2006), such that principals in these schools may not need to further empower or professionally develop teachers. On the other hand, principals of high-SES schools may desire to use empowerment and professional development to maximize the academic capacity of schools.

Influence of leadership on student learning

This review next examines the influence of principal leadership on the attitudes of students toward science learning (enjoyment and interest) and achievement. Enjoyment of and interest in science are two complementary variables that can be used to measure their attitudes toward science (Osborne et al., 2003). According to Fredrickson (2001), students who enjoy learning have “the urge to play, push the limits, and be creative,” while those who are interested in learning have the “urge to explore, take in new information and experiences, and expand the self in the process” (p. 220). Therefore, students who enjoy learning may be interested to learn different topics in a subject. Hidi (2006) argues that the interest of students is a unique motivational variable that occurs during interactions with specific objects of interest. In science education, students who enjoy learning and are interested in learning about different topics in contemporary science may internalize scientific principles more easily and, therefore, have high levels of science achievement (Acosta & Hsu, 2014).

When principals implement envisioning and instructional management, students can learn from qualified, motivated teachers who employ instructional practices with requisite educational resources (Hendriks & Scheerens, 2013; Leithwood, 2005, 2012; Sebring et al., 2006). Furthermore, these leadership practices enable a close alignment among curriculum, instruction, and assessment through continuous monitoring and evaluation (Leithwood, 2012; Murphy et al., 2006; Robinson et al., 2008). The active involvement of teachers in building the school vision also promotes their sense of shared ownership (Leithwood, 2012; Sebring et al., 2006; Supovitz et al., 2009). Therefore, students whose principals exercise leadership in envisioning and instructional management acquire more positive attitudes toward science learning, which benefit achievement.

Principals can empower teachers when the latter demonstrate high levels of competence. Empowering teachers enables better decision-making based on their professional inputs (Leithwood & Mascall, 2008; Supovitz et al., 2009). In addition, teacher empowerment creates positive school conditions (e.g., trust, care, risk-taking, and continuous learning), which promote student learning (Hunzicker, 2012). Individually, students learn better when they are taught by qualified teachers. To illustrate, Woolnough (1994) examines A-level students and finds that one of the strongest variables that predicted whether or not students selected science as a subject was the quality of science teaching. This variable can be derived from well-qualified graduate science teachers who possess science expertise and subject affiliation, enthusiastically taught authentic science lessons, conducted structured and stimulating science lessons, and talked to students about science. Therefore, qualified teachers, when empowered by principals, are more capable of contributing to decision-making that addresses the needs of students. Moreover, empowered teachers can support the professional growth of peers, including providing the latter with quality and relevant professional learning and assisting with issues related to pedagogical content knowledge (Wenner & Campbell, 2017), which, thereby, benefit student learning.

Professional development nurtures teachers who are cohesive, professional, competent, and efficacious (Hendriks & Scheerens, 2013). Multiple factors are required to enable the professional development of teachers to positively impact student learning (Opfer & Pedder, 2011). These factors include simultaneous, mutually reinforced changes in the professional beliefs and practices of teachers after professional development and the acquisition of subject-specific pedagogical skills. Furthermore, professional development benefits teaching and learning if principals can provide the necessary organizational support. Therefore, student learning is beneficial when professional learning for teachers is extensive.

The present study

The present study (a) elucidates the relationship between two sets of contextual conditions proximal to student learning, namely, the availability of family (SES) and school resources (qualified science teachers and school science resources) and four leadership functions of principals (i.e., envisioning, instructional management, promotion of professional development, and empowerment of teachers). Furthermore, the study (b) investigates the contribution of principal leadership to the science learning of students (science enjoyment, interest, and achievement).

The decision to use science contextual variables to ascertain the influence of context on leadership is informed by two considerations. First, it enables the scrutiny of specific relationships pertaining to science education. Second, well-funded schools with qualified science teachers and school science resources are also likely to possess resources for support teaching and learning in other subject areas. Thus, focusing on science contextual variables does not compromise the characterization of other general school contexts.

The study examines principal leadership instead of teacher leadership as the independent variable, because previous research demonstrates that principal leadership constitutes the most important source of leadership in schools (Tan et al., 2021, 2022). For example, Leithwood and Jantzi (2000) find that principal but not teacher leadership contributed to student engagement in Canadian schools. Day et al. (2009) conduct a study on the improvement of schools in England and find that teachers, governors, and parents perceived headteachers as the key source of leadership that influenced teaching processes.

Furthermore, the study employs multilevel structural equation modeling (SEM) analysis, which included student SES and the Confucian heritage culture (CHC) of countries as control variables. These control variables are included because previous studies illustrate that students from high-SES families and from CHCs exhibit high levels of academic achievement (e.g., Tan et al., 2020).

Method

Participants

The sample comprised 248,620 students from 9370 schools in 35 member countries of the Organization for Economic Cooperation and Development (OECD) who participated in the Programme for International Student Assessment (PISA) 2015 (OECD, 2017). These students were mostly in Grade 10 (55.9%), while the rest was in Grades 7–13. Public schools (69.7%) outnumbered private schools (14.8%); 15.5% of the schools were unclassified. The participating students represented the complete population of 15-year-old students from Grade 7 or higher. PISA 2015 measured the proficiency of 15-year-old students in applying their knowledge and skills learned in science (the focal domain) in addition to reading and mathematics.

Measures

Data on the following variables were derived from the PISA 2015 dataset (https://www.oecd.org/pisa/data/2015database/) for analysis.

Availability of science resources

The study measured the availability of science resources in schools (SciRes) by summing up the responses of the principals (Yes, No) to eight items on the resource availability of the science department (allocation of extra funding to science teaching, levels of education of science teachers, materials for laboratory and hands-on learning, laboratory support staff, and up-to-date science equipment).

Qualifications of science teachers

The qualifications of science teachers (TrQua) were measured using the proportion of science teachers with Bachelor/Master and science major qualifications in schools (principal-reported).

Student and school SES

The study measured student SES (StuSES) using the index of economic, social, and cultural status computed by PISA 2015 (OECD, 2017). The index represented the first principal component derived from student data on the highest level of education of parents, highest occupational status of parents, and home possessions of the students. Data on the highest level of education of parents were derived from student responses (two questions for each parent). The response categories corresponded to “no education,” “primary education,” “lower secondary,” “vocational/pre-vocational upper secondary,” “general upper secondary and/or non-tertiary post-secondary,” “vocational tertiary,” and “theoretically oriented tertiary and postgraduate.” Data on the highest occupational status of parents were derived from student responses on the nature of the main jobs of their parents (two questions per parent). PISA 2015 coded these data and mapped the codes into the international socioeconomic index of occupational status (Ganzeboom & Treiman, 2003). Data on the home possessions of students were derived from student responses to three questions on the availability of various home resources (e.g., study desk, own room, quiet place to study, computer for study, educational software, Internet connectivity, classic literature, poetry books, art works, books to support study, reference books, dictionary, books on art/music/design, televisions, cars, rooms with bath/shower, cell phones with Internet access, tablet computers, e-book readers, and musical instruments). The present study averaged the SES levels of students within a school to obtain the measure of school SES (SchSES).

Principal leadership

The study measured principal leadership with four scales using the responses of principals to 13 items on the frequency of specific leadership practices. Items were rated using a six-point scale (1 = Did not occur, 2 = 1–2 times during the year, 3 = 3–4 times during the year, 4 = Monthly, 5 = Weekly, 6 = More than once a week). The scales corresponded to the four core principal leadership functions that characterize contemporary leadership practices (Leithwood et al, 2006, 2020). The first scale (Envision) measured the framing and communication of principals in relation to school goals and curricular development with four items pertaining to principals using student results to develop the academic goals of the school, aligning the professional development and work of teachers with school goals, and discussing school goals with teachers (α = 0.78). The second scale (Instruct) measured instructional management using three items related to principals promoting research-based teaching practices, praising teachers whose students were learning actively, and emphasizing to teachers the development of critical and social capacities in students (α = 0.78). The third scale (PD) measured the promotion of professional development of teachers by principals using three items: principals taking the initiative to discuss problems encountered by teachers in classrooms, paying attention to disruptive behaviors among students, and collaboratively solving classroom problems with teachers (α = 0.82). The fourth scale (Empower) measured the facilitation of the participation of teachers in leadership by principals using three items related to principals engaging the staff to participate in school decision-making, building a school culture of continuous improvement, and reviewing management practices (α = 0.77). Confirmatory factor analysis (CFA) demonstrated that the four scales satisfactorily explained the variation in the 13 items [χ2(59) = 3,091.61; CFI = 0.93; RMSEA = 0.08].

CHC

The variable CHC assumed a value of 1 if the country was a CHC (e.g., Korea and Japan); otherwise, it assumed a value of 0 (the remaining 33 countries).

Enjoyment of students in learning science

The enjoyment of students in learning science (Enjoy; α = 0.94) was computed using their responses to five items measuring the extent to which they enjoyed science learning using a four-point scale (1 = Strongly disagree to 4 = Strongly agree). These items pertained to students learning, reading on, and working on science topics. CFA displayed satisfactory model fit with the five items loading on a single latent construct (χ2(5) = 3,182.39; CFI = 0.99; RMSEA = 0.05).

Science interest of students

The study measured the science interest of students (Interest; α = 0.82) using their responses to five items on the extent of their interest in different science topics pertaining to the biosphere, motion and forces, energy, the universe, and the application of science to disease prevention. The items were rated using a four-point scale (1 = Not interested to 4 = Highly interested, Don’t know [coded as missing]). CFA pointed to a satisfactory model fit with the five items loading on a single latent construct, with two items allowed to covary with each other (χ2(3) = 3,152.75; CFI = 0.99; RMSEA = 0.07).

Science achievement of students

The science achievement of students was the focal-dependent variable in PISA 2015. Students were not administered the complete set of test items by design, and, therefore, each item featured missing responses. To overcome this limitation, PISA 2015 aggregated the results of individual students to produce scores for groups of students. It also used a set of 10 plausible values (PV1–PV10) for each student to represent the estimated distribution of their science scores similar to the student in terms of responses to the assessment and background items. The analysis specified the data on the science achievement of students as “TYPE = IMPUTATION” in SEM. Supplementary Tables 1 and 2 summarize the descriptive statistics for the student- and school-level variables examined in the study, respectively. Table 1 summarizes the relationships between the variables.

Table 1 Bivariate correlations
Full size table

Procedure

PISA 2015 used a two-stage stratified sampling design, in which schools were first selected from a national sampling frame of schools with probabilities proportional to size followed by students selected from each of the selected schools (OECD, 2017). The OECD internationally sponsored PISA 2015. All participating countries followed the standardized procedures outlined in the technical standards and manuals provided.

Multiple imputation

Missing values may compromise estimation efficiency and produce biased results (Cheema, 2014). Therefore, the study performed a multiple imputation (fully conditional specification) of variables with missing values using all other variables as predictors. The imputed datasets comprised imputed values at the maximum iteration. The missing values were imputed for school- and student-level variables separately. A total of 10 datasets were imputed for student and school-level data.

Three-level SEM

The study fitted a three-level SEM model (student, school, country at Levels 1, 2, and 3 respectively) to the data with random intercepts, maximum likelihood ratio estimator, and student and school weights to examine the relationships among context, principal leadership, and learning outcome of students while controlling for student SES and CHC using Mplus8.1. Student SES was specified as a Level-One variable; school context and principal leadership variables were designated as Level-Two variables; and CHC was specified as a Level-Three variable. Enjoy and Interest were not specified as belonging to any specific levels, because they were assigned as the dependent variables of the Level-Two variables and the independent variables at Level 1. PV was not specified at any specific levels, because it was modeled as the dependent variable at Levels 1 and 3. The four leadership variables were correlated with one another (Envision with Instruct: r = 0.72; Instruct with Empowerment: r = 0.64; Instruct with PD: r = 0.59; Envision with Empowerment: r = 0.59; PD with Empowerment: r = 0.54), so they were allowed to covary with one another in the SEM. Various model fit indicators, such as χ2, CFI, and RMSEA, were used to evaluate the models.

Results

Figure 2 depicts the SEM model fitted. The model fit statistics were satisfactory (CFI = 0.92, RMSEA = 0.02), and χ2(df) was non-significant at the 0.05 level. The SEM model explained 11% of the variance at Level 1 in the science achievement of students. The results demonstrated that the levels of certain resources predicted the variables of leadership (Table 2). Specifically, principals in schools with more science resources were more involved in all four leadership practices, namely, envisioning (β = 0.53), instructional management (β = 0.75), promotion of the professional development of teachers (β = 0.36), and empowerment of teachers (β = 0.38; p < 0.001). Principals in high-SES schools were less likely to be engaged in envisioning (β =  − 0.11, p = 0.043), promote the professional development of teachers (β =  − 0.17, p = 0.002), or empower teachers (β =  − 0.14, p = 0.014). In contrast, the study found no relationship between school SES and instructional management (β =  − 0.03, p = 0.600). The proportion of qualified science teachers was not significantly related to the envisioning of principals (β = 0.15, p = 0.057), instructional management (β = 0.12, p = 0.197), promotion of the professional development of teachers (β =  − 0.05, p = 0.433), or empowerment of teachers (β = 0.10, p = 0.136).

Fig. 2
figure 2

SEM models: contexts shaping principal leadership

Full size image
Table 2 Standardized effects
Full size table

Furthermore, instructional management was positively related to the science enjoyment (β = 0.02, p = 0.003) and Interest (β = 0.01, p = 0.006) of students. In contrast, the promotion of the professional development of teachers was negatively related to the science enjoyment (β =  − 0.02) and interest (β =  − 0.02, p < 0.001) of students. The study observed no relationship between envisioning and the science enjoyment (β = 0.00, p = 0.966) or interest (β = 0.00, p = 0.825) of students. Similarly, no association existed between empowerment of teachers and science enjoyment (β = 0.00, p = 0.939) or interest (β = 0.00, p = 0.613) of students. The science enjoyment of students predicted their science interest (β = 0.56, p < 0.001). Students with high levels of science enjoyment (β = 17.64) and interest (β = 13.97) exhibited high levels of science achievement (p < 0.001).

Discussion

Elucidating the relationships between context and leadership

The results of the current study support the argument of Hallinger (2018) on the existence of “a generic set of leadership practices (e.g., goal setting, developing people) which must be adapted to meet the needs and constraints that describe different school contexts” (p. 5). These results provide evidence of the nature of relationships between context and leadership, that is, contexts shape principal leadership. They caution against the adoption of a simplistic perspective of leadership as an “agency without structure,” which focuses on “simplistic solution-seeking rather than appreciation of complexity and paradox in the leadership experience” (Close & Raynor, 2010, p. 209).

The literature has documented the myriad tasks that confront principals on a daily basis (Goldring et al., 2008); thus, the present study provides insights on contextual considerations to inform principals on the prioritization of their time. For example, the results demonstrate that principals leveraged the strengths of their schools (e.g., availability of school resources) when they focused on envisioning, instructional management, and the professional development and empowerment of teachers. In low-SES schools, they address challenges by focusing on envisioning and the professional development and empowerment of teachers. These results extend those reported in other studies on the implementation of principals of contingent leadership practices in response to challenging school contexts (Day et al., 2009; Notman & Henry, 2011).

Strategic importance of school resources

The results of the influence of school science resources on the four leadership practices underscore the strategic importance of this contextual variable. This finding necessitates a revisitation of the role of resources in education. Researchers focus on assessing the impact of school resources on student learning but report dismal results (Hanushek, 1996). Indeed, Cohen et al. (2003) question whether availability is less important than the instructional use of school resources. Notably, the relationship between school resources and leadership practices is strongest for instructional management, which the literature regards as the most contributory factor of student achievement (Hitt & Tucker, 2016). The implications are stark for poorly resourced schools that need to circumvent resource inadequacy (compromising student learning) and that lack strong instructional management (as a result of resource inadequacy). If the availability of school resources influences leadership practices, then fine-tuning school funding models is imperative for policymakers. Toward this end, David-Hadar (2018) proposes a more comprehensive funding formula on the premise of the demonstrated improvements of schools in terms of student learning in addition to school input or outcome, which merits further consideration.

Principal leadership not related to the proportion of qualified teachers

The results demonstrate that the envisioning, instructional management, and promotion of the professional development and empowerment of teachers by principals was unrelated to the proportion of qualified teachers in schools. These results are counter-intuitive given the complexity of the leadership and management of schools. For example, principals may be more inclined to undertake the complex task of galvanizing the school community to support their vision (Murphy & Torre, 2015) if they are working with a competent staff of teachers. They are also expected to empower a staff of qualified science teachers to a greater extent, especially when they lack the requisite content knowledge in science (Lochmiller, 2016). Therefore, the reason showing an absence of a relationship between principal leadership and the proportion of qualified teachers in schools remains unclear. One possible reason for explaining this finding is that the majority of principals appreciate the need to undertake the four leadership practices examined in the study to address their school needs. This notion is indicated in the mean scores for the leadership scales, which range from 3.41 for envisioning to 4.48 for the professional development of teachers out of a maximum score of 6 (Supplementary Table 2). Therefore, little variation exists in the leadership practices of principals among schools with varying proportions of qualified teachers. Future research can ascertain the reasons for the nonsignificant associations between the proportion of qualified teachers in schools and principal leadership.

Building teacher capacity in low-SES schools

Principals leading low-SES schools face the challenge of personalizing the school environment to enable students to identify with the schools, incorporating the backgrounds of students in instructional design, and helping teachers to affirm the family capital of students (Hitt & Tucker, 2016). Therefore, principals need to provide extensive professional development for teachers in these schools (Clotfelter et al., 2006). Another possible scenario is that certain low-SES schools may be staffed by ill-qualified, unmotivated teachers. For example, Clotfelter et al. (2006) find that teachers perceived the working environment in high-poverty schools in North Carolina as less favorable. The authors also report that teachers in high-poverty schools were more likely to produce low test scores, graduate from less competitive undergraduate institutions, and hold non-regular licenses, and least likely to be board-certified or experienced. Therefore, principals leading low-SES schools may emphasize professional development to build teacher capacity. Low-SES schools may also be resource-poor, such that principals need to empower teachers to optimize resources for teaching and learning, especially in the subject areas of the teachers (Tschannen-Moran, 2009). These reasons may explain why principals leading low-SES schools are more likely to provide professional development and empower to teachers. Therefore, one method in which principals address resource challenges in low-SES schools is seemingly to build teacher capacity.

Instructional management versus the professional development of teachers for the science learning of students

The results demonstrate that among the four leadership functions, only instructional management was positively related to the science enjoyment and interest of students. Previous studies demonstrate mixed results for the relationships between school leadership and the non-academic learning variables of students, that is, non-significant (Bellibas & Liu, 2018) or positive (Tan et al., 2020; Adams & Olsen, 2017, 2019; Zheng et al., 2017) relationships. However, the majority of the variables are not subject-specific; an example of this exception is language self-efficacy (Zheng et al., 2017). Therefore, the present study advances knowledge by examining the influence of leadership in a specific content area, namely, science.

The finding is also noteworthy in affirming the mainstream consensus that school leadership should focus on teaching and learning (Boyce & Bowers, 2018; Daniëls et al., 2019; Hallinger & Kovacevic, 2019). Indeed, the scholarship on educational administration is dominated by the “cognitive anchor” of “leadership for student learning and development” (Hallinger & Kovacevic, 2019) in the past 60 years. Moreover, research finds that instructional management is the most effective for enhancing the learning of students from disadvantaged homes and schools (Tan, 2018).

The negative relationships between the promotion of the professional development of teachers and the science enjoyment/interest of students remain ambiguous. Notably, these relationships do not measure the effectiveness of professional development but emphasize the building of teacher capacity via professional development by principals. Another aspect worth reiterating is that the results demonstrate that the promotion of the professional development of teachers was more likely to occur in low-SES schools. Therefore, principals may not have successfully equipped teachers in low-SES schools with the knowledge and skills required to address the learning needs of low-SES students.

Limitations

This study has three limitations. First, it analyzes the self-reported data of principals, which can induce bias (Urick, 2016). Future studies can address this limitation by incorporating teacher-reported data on principal leadership. Second, data are cross-sectional; thus, causal inferences must be made with caution. Future research can employ a causal design to ascertain the reported relationships. Third, the study examines the relationships among context, school leadership, and the variables of student outcome. It does not formally ascertain if school leadership mediates the relationship between context and student outcome due to the methodological complexities in the data (e.g., multiple imputation, plausible values, three-level data, and sampling weights at different levels). Indeed, the common analytical approaches for testing for mediation, such as bootstrapping, are unavailable for testing of indirect effects in multilevel mediation (Tofighi & Thoemmes, 2014). The implication of failing to formally test for mediation effects is that one cannot ascertain the indirect effects, in addition to the direct effects, of context on student outcome. Another implication is that the study was unable to determine the intervening variables of school leadership by which contextual factors influence student outcome. Therefore, the study was unable to determine the total magnitude of the influence (comprising direct and indirect effects) of context on student learning or to compare the relative influence of different contextual factors on student outcome. Future studies can explore the use of emerging approaches, such as asymptotic normal theory (using Mplus) or the distribution of the product of coefficients method (using RMediation package) to formally test for mediation effects while incorporating the data complexities (Preacher et al., 2011; Tofighi & MacKinnon, 2011).

Conclusion

The present study makes three contributions to the literature on school leadership. First, it demonstrates that context shapes principal leadership, such that principals are more likely to be responsive to, instead of developing, their contexts. It addresses the clarion call issued by Hallinger (2018) to bring “‘contexts for leadership’ out of the shadows” (p. 6) and the emerging realization that school leadership (especially instructional)) is in urgent need of conceptual renovation (Dimmock & Tan, 2016). Second, the study elucidates how principals may respond to their contexts. Specifically, principals may capitalize on contextual strengths and opportunities (e.g., adequate school resources) for envisioning, instructional management, and the professional development and empowerment of teachers. They may also address challenges (e.g., in low-SES schools) by envisioning and promoting the professional development and empowerment of teachers. Third, the study demonstrates how the examination of leadership across contexts promotes the appreciation of the complexities and paradoxes in the research on the effects of leadership (Close & Raynor, 2010). To illustrate, the results demonstrate that principal leadership in promoting the professional development of teachers was negatively related to the attitudes of students toward science learning. However, this finding can be explained when we juxtapose it with another that illustrates that principals may be promoting professional development to enhance teacher capacity in low-SES schools.