Learning From Leadership: Investigating the Links to Improved Student Learning

Click here to download the full report:
 Learning From Leadership: Investigating the Links to Improved Student Learning

Key Findings

  • While there are many sources of leadership in schools, principals remain the central source.
  • How leadership is distributed in schools depends on what is to be accomplished, on the availability of professional expertise, and on principals‘ preferences regarding the use of professional expertise.
  • No single pattern of leadership distribution is consistently linked to student learning.
  • Principals are involved in many leadership activities; others who act as leaders in the school ordinarily do so in respect to one or a few initiatives.
  • Leadership is more distributed for practices aimed at "developing people" and "managing instruction" than it is for "setting directions" and "structuring the workplace."
  • More complex and coordinated patterns of distributed leadership appear when school improvement initiatives focus directly on student learning goals, as distinct from the implementation of specific programs.


Leadership can be conceptualized and studied as an individual or an organizational phenomenon. The former conception orients us toward an analysis of the beliefs, actions, personal traits, and influence of individuals recognized by others as leaders. An organizational perspective suggests that leadership is unlikely to be constituted solely of the actions and influence of an individual. According to this view, we need to examine the range of leadership sources, beliefs, actions, interactions, and influences recognized by participants in those settings.

Section 1.1 of our report describes influence arising from various sources of leadership as that influence comes to bear on school decisions, teachers‘ work, and student learning. Section 1.2 describes leadership shared among principals and teachers as that leadership relates to instruction, trust, professional community, and student achievement. These two sections are based on evidence from teacher surveys and student achievement data. In contrast, Section 1.3 is based on evidence from principal and teacher interviews. We analyze this evidence in an effort to answer four questions:

  • Who participates in leadership distribution?
  • What patterns does leadership distribution take?
  • How is responsibility for "core" leadership functions (described in other sections) distributed?
  • How is leadership distribution related to school improvement goals?

Prior Evidence

Scholars recently have focused considerable attention on the properties and complexities of leadership distribution in schools and districts—sources, focal points, functions, interactions, contexts, and outcomes.106 We know that leadership may be distributed in various patterns, though consensus on a typology and terms remains elusive. Furthermore, we know little to nothing about how different forms of leadership distribution enhance or do not enhance the accomplishment of organizational goals.

Gronn (2002) refers to holistic and additive models of leadership distribution. The additive model refers to a dispersed pattern of leadership in which multiple members of an organization provide leadership for varying goals and/or tasks. Different members may provide leadership for different purposes, without coordination or a shared focus. The holistic model suggests greater interdependency and coordination among varied sources, focused on shared goals and tasks.

At a more micro-level, Spillane (2006) identifies three arrangements for distributing leadership responsibilities: division of labor (different leaders for different tasks), co-performance (multiple leaders together for same task), and parallel performance (multiple leaders perform the same tasks but in different contexts). Similarly, Goldstein (2003) and Gronn (2002) distinguish between situations in which leadership for specific tasks is enacted by multiple leaders, together or separately. Spillane expands upon this formulation, defining three types of co-performance: collaborated distribution (multiple leaders jointly enact the same leadership practice in the same context); collective distribution (multiple leaders perform separate but interdependent tasks in different contexts and in support of the same goal); and coordinated distribution (interdependent actions of multiple leaders are performed in a particular sequence).

Recently, Leithwood and his colleagues have conceptualized a typology that offers a more general theoretical framework for exploring the distribution of leadership in organizations.107 The framework, grounded in a research-based definition of leadership, identifies four categories of "core" leadership functions: setting directions, developing people, redesigning the organization, and managing the instructional program.108 This typology emphasizes variability in the alignment of leadership functions and in beliefs associated with different forms of alignment: planful alignment, spontaneous alignment, spontaneous misalignment, and anarchic misalignment.

The analysis in Section 1.3 builds on past theory and research to explore the nature and patterns of leadership distribution in schools, focusing on sources of leadership influence and the relation of leadership influence to student performance. We pay particular attention to the role principals play in the distribution of leadership.

New Evidence


Data for our analyses arise from interviews with school personnel in a sub-sample of schools participating in the site-visit component of the larger study. The teacher survey administered to all participating schools during the first round of data collection included a set of items designed to measure the relative influence of those in multiple roles on school decision making (see Section 1.1). From these items, we derived a measure of collective leadership that enabled us to make comparisons across schools by reference to the range of sources of leadership influence and the strength of that influence on teachers.

We selected a purposive sample of site-visit schools for this analysis. First, we classified all site-visit schools as high, medium, or low on the collective leadership and student performance measures. From the resulting matrix, we selected five schools for qualitative analysis of leadership distribution. These schools varied widely on collective leadership scores and student performance. The sample (Table 1.3.1) included elementary and middle schools, schools in high- and low-SES settings, and schools in inner-city, suburban and rural settings across four states (Texas, Missouri, Oregon, and New Jersey).

We collected data for each school, using all school administrator and teacher interviews conducted during the first site visit (8-10 interviews per school). We transcribed all interviews and entered the transcripts into an NVivo project data base that included leadership as one of the core codes.

We employed a three-stage process of analysis. In stage one we created descriptions of leadership activities in each school derived from the NVivo data queries. We developed a findings template that drew upon Spillane‘s conceptualization of leadership practice.109 The template enabled us to construct descriptions of (1) sources of leadership linked to (2) specific actions and (3) goals in (4) specific contexts, along with (5) the co-participants in those situations, (6) the reported effects of those actions, and (7) the reported factors influencing those leadership variables. This analysis generated 15-25 leadership scenario templates per school.

In stage two we recoded each scenario according to the core leadership practices exemplified (here we used operational definitions derived from Leithwood & Jantzi, 2006). Then we wrote an analysis of the leadership distribution patterns we discerned in the scenarios, applying concepts from research on leadership distribution as appropriate.

In stage three we wrote a case report for each school, integrating findings from the scenario analyses and structured according to the research questions. The findings presented and discussed here highlight key themes and findings that emerged from the cross-case analysis.


TABLE 1.3.1
Sample School Characteristics



Collective Leadership

Student Achievement110


London Elementary



Size: 537 Pupils
Diversity: High
Poverty: High

Overton Elementary



Size: 221 Pupils
Diversity: Med
Poverty: Med

Gregory Elementary



Size: 581 Pupils
Diversity: Med
Poverty: High

Playa Junior High



Size: 345 Pupils
Diversity: High
Poverty: High

Forest Elementary



Size: 443 Pupils
Diversity: High
Poverty: Med

Who Participates in Leadership Distribution?

Consistent with the findings of others,112 we found that school personnel did not attribute leadership actions and influence only to one source, and not always to the principal. The individuals or groups identified as providing leadership included a mix of principals, assistant principals, teachers in formal leadership roles (e.g., grade or subject team leaders) and teachers with specialist positions (e.g., literacy specialists, technology specialists, counselors). Teachers also identified other teachers informally recognized by peers as influential; school leadership or management committees; school program teams or committees (e.g., Special Education, Gifted and Talented, Limited English Proficiency); parent involvement personnel; district administrators and professional staff; and external consultants linked to particular areas of curriculum, program, and teacher development priorities at the school level.

What Patterns Does Leadership Distribution Take?

Mere identification of the various individuals and groups contributing to school leadership provides scant insight into the actual distribution of leadership. Overall, principals were more likely than any other source to be implicated in multiple leadership responsibilities. Three overall patterns of distribution appeared across the five schools:

  • Pattern One (London, Overton, and Gregory Elementary Schools). The leadership influence of the principal was evident across various focal points of schoolimprovement activity. Principals were seen to exercise influence in planful collaboration with influential school-based teacher leaders (individuals and groups) and with outside sources (district specialists, external consultants) associated with particular goal-oriented initiatives. In these schools there was a strong emphasis on professional collaboration among teachers, including teachers in instructional leadership roles that crossed curriculum and grade boundaries. These schools had high collective leadership ratings on the survey measure.
  • Pattern Two (Playa Jr. High School). The leadership influence of the principal extended across various focal points of school-improvement activity, but the evidence was less robust for influential sources of teacher leadership and for principal collaboration with teachers and/or external change agents. Teacher leadership was limited to traditional grade-level or program-specific structures, and there was less emphasis, school-wide, on teacher collaboration.
  • Pattern Three (Forest Elementary School). The principal interacted administratively with various focal points of school-improvement activity, but she had little influence on implementation. Key teachers or external agents were identified with support for different improvement initiatives, yet teachers attributed little influence to their enactment of those roles. Teachers did not report an emphasis on, or culture of, teacher collaboration within or across school organizational structures.

These findings from the five schools are consistent with the higher collective leadership scores in London, Overton and Gregory Schools, and with the lower scores in Playa and Forest Schools. Sometimes leadership is conceptualized as a school-level phenomenon; sometimes it is conceptualized for a specific, goal-oriented activity. Gronn‘s (2002) distinction between additive and holistic leadership is useful for describing leadership distribution here. Among our cases, Forest Elementary provides the clearest example of a school in which the overall pattern of leadership distribution corresponded to an additive pattern, at least in a formal, bureaucratic sense (teachers attributed little actual influence to those in formal positions of leadership responsibility).

The distribution of leadership sources in London, Overton, and Gregory Schools conformed more closely to the holistic pattern of leadership distribution. This is clearly reflects the extension of the principal‘s leadership influence across various focal points of school improvement. Playa School did not clearly fit either an additive or a holistic pattern of distribution in leadership sources, in part because there was no strong teacherleader presence.

Teachers’ Collective Influence as a Pattern of Distributed Leadership

Teachers in several schools talked about the collective influence of teachers, not merely the influence of colleagues identified as teacher leaders. Collective influence, these teachers reported, was instrumental in school decisions and in broader decisions about school improvement. They framed it as a function of whether the principal and district authorities invited, valued, and acted upon input from teachers. This qualitative finding reinforces the teacher survey-based findings on collective and shared leadership presented in Sections 1.1 and 1.2.

In London School, for example, teachers reported that a previous principal rarely solicited teacher input; when she did, teachers said, she rarely acted in ways that acknowledged the value of that input. They felt unsupported, and increasingly they kept their opinions and ideas themselves, thereby decreasing the potential for broader teacher influence on decisions in the school. That changed when a new principal came in—one who was perceived as genuinely seeking and respecting teacher input and influence on school decisions. Teachers and principals in Overton and Gregory Schools also affirmed the presence and influence of a strong collective voice from teachers, facilitated by the principal‘s orientation to teacher input and to organizational structures enabling that input. These findings stand out in contrast to discussions, widespread in the profession, that focus narrowly on the leadership contributions of individually influential teachers.

Formal Role Designations and Patterns of Distributed Leadership

It is tempting to associate the bureaucratic distribution of roles, responsibilities, and authority with the distribution of leadership sources and influence. Beyond the pervasive role of the principal, however, our findings paint a more complex picture.

First, the bureaucratic allocation of responsibility to perform certain functions and tasks does not necessarily mean that the persons or groups so designated will be perceived as influencing what others think and do. Spillane (2006) argues that leadership sources and acts can be recognized as such even if they do not yield their intended effects. But that argument is difficult to sustain against evidence (from Forest School, for example) about people in formal leadership positions whose actions are not seen by school personnel to make much difference.

Second, bureaucratic structures do not determine how patterns of leadership distribution will be enacted through any given bureaucratic structure. A given bureaucratic structure may be compatible with more than one pattern of leadership enactment. The schools examined here all had multi-stakeholder school leadership committees and special program committees (e.g., special education, bilingual education); they all had a similar array of formal teacher-leader positions, including subject and grade team leaders. Some had teachers assigned to instructional leadership roles associated with priorities for improvement in program and instruction (e.g., in literacy and mathematics). However, actual patterns of leadership influence varied from school to school. Even in single schools, we found examples of variation over time in how leadership was enacted and distributed through the same bureaucratic structures. Principal succession was a factor in each of these situations.

In London School the current principal and her predecessor both worked with a School-Based Management Team, grade-level teams, cross-grade subject teams, special program committees (gifted education, bilingual education, etc.), and specialist roles (counselor, literacy teacher, parent involvement coordinator, etc.). Under the previous principal, the leadership distribution pattern had been highly additive, and the principal was uninvolved with school-improvement initiatives. These initiatives were mandated by the district; they proceeded in an uncoordinated manner, guided and managed by grade team leaders, specialists, and external consultants. The new principal took on a proactive leadership role, exercising influence within existing governance structures in a way that spanned multiple focal points of school-improvement activity. That change yielded a more holistic pattern of leadership distribution.

In Gregory Elementary School a previous principal led an effort to implement the Accelerated Schools comprehensive school reform model. This effort entailed formation of five curriculum cadres, a school-site council, and a school-improvement planning process. While the cadres and council were chaired by teachers, and teacher influence on school directions, improvement plans, and professional development was reportedly strong, school personnel said that the previous principal played a more overt coperformance leadership role within those structures than the current principal. The current principal and assistant principal talked about deliberately stepping back from a coperformance leadership role to a more indirect advisory role in the cadres and site council. Teachers also reported that adherence to the needs assessment and planning processes became less stringent under the new principal. These cases show that formal organizational structures create an institutional framework for the distribution and enactment of leadership, but they do not determine how leadership plays out over time.

In sum, it is important to distinguish the formal allocation of leadership roles and responsibilities from what Leithwood et al. (2007) define as the planful alignment of leadership sources, practices, and influence. Formal bureaucratic structures do not necessarily require or facilitate the kind of consensus building, communication, interaction, and collaboration that we would associate with the planful alignment of leadership.

How Responsibility for Core Leadership Functions Is Distributed

Analyses of our case study data indicate that patterns of leadership distribution and influence can vary by core leadership practices—not only between schools and districts, but also for different focal points of activity within a given school. Overall, leadership is more commonly distributed for developing people and managing instruction than it is for setting directions and structuring the workplace. This emphasis probably reflects the influence of external policy, which may limit the freedom of principals and teachers to set goals or to redesign the workplace. Principals‘ beliefs about their own expertise and expertise from other sources also affect direction setting, and they are a key factor shaping the distribution of leadership for developing people and program management.

For all the schools and districts sampled in our study, state and federal curriculum policies, standards, and accountability systems influenced direction setting pervasively. Flexibility for principals and teachers depended greatly on the extent to which state and district authorities tended to mandate programs or to enable schools to select their own priorities and programs. Ultimately, however, leadership distribution for direction setting is shaped by how the principals view and enact their roles within the context of state and district policies, priorities, and leadership traditions, as illustrated in the following contrasting examples.

The principal and teachers at Forest Elementary School portrayed themselves as complying with state- and district-mandated programs (e.g., in reading and mathematics) and procedures (e.g., curriculum mapping, student data reports). The principal described herself and the School Leadership Committee as managing the implementation of externally mandated directions, not as setting directions per se. In contrast, the state and district did not mandate commercial or local programs at Overton Elementary School. While district authorities established system priorities for improvement based on results from state testing (e.g., in mathematics), the principal focused her leadership influence less on setting or enforcing program or achievement targets for improvement than on structuring the workplace (e.g., through a Leadership Committee, curriculum teams, and coaches), facilitating teacher learning (through lesson study and book study teams), and managing the instructional program (by monitoring teaching and teachers‘ professional learning plans) in ways that guided teachers to establish their own directions for improvement, collegially, in the context of state standards, test results, and district priorities.

Our cases highlight two circumstances in which principals may be more prone to act directly and less collaboratively to influence school directions for improvement:

First, a principal known to possess specific expertise in curriculum or instruction may be inclined to press forward on the strength of that expertise. At London Elementary School, for example, the principal was well known for her expertise in reading. She decided that children in her school would do better in reading if teachers were to adopt and implement a wider variety of teaching strategies. She communicated that goal to teachers, provided training herself and via an external expert, and she monitored teachers‘ implementation of new strategies in the classroom and in grade team meetings. At the same time, she facilitated ongoing improvement efforts mandated at the district level prior to her appointment (curriculum writing, implementation of a commercial mathematics program)—collaborating with grade team and subject leaders, specialist teachers, and trainers provided by the externally developed mathematics program.

Second, a principal who believes that his or her teachers have become complacent may be inclined to press forward independently, launching efforts to set higher standards for teacher performance and student learning. At Playa Junior High School, for example, the principal sought school improvement through an effort to get teachers to be less didactic in their teaching, to broaden their repertoires of instructional strategies, and to focus on higher-order learning expectations. She explained her initiative as a strategy to motivate teachers and to help them improve student performance beyond the predominantly "acceptable" ratings the school had received under the state‘s accountability system. She reported that she coached teachers, and made use of external consultants for in-service training, with this in mind. Teachers at Playa were also involved in curriculum writing projects in response to a district mandate. The principal delegated responsibility for leading and managing the curriculum development work to traditional subject heads and teams.

The general point of these accounts is that patterns of leadership distribution and influence can and do vary for different dimensions of leadership practice (i.e., setting directions, developing people, workplace [re]design, and managing the instructional program)—not only between schools and districts, but also for different focal points of improvement within a given school. Here, as in many other areas of interest, professional practice is more varied and complicated than the simplified patterns that often stand out in scholarly discussions.

Complexity of Leadership Distribution as a Function of Goal Type and Breadth

Leadership distribution patterns are affected by the goals that school personnel associate with leadership activity. Some goals (e.g., improving student performance in mathematics, strengthening professional community) are more encompassing than others (e.g., implementing a specific mathematics program, standardizing student discipline policy and practices). The more encompassing the goal, the greater the likelihood that multiple sources of leadership will be involved, and the greater the range of goal-related activities to which leadership might be attributed.

Contrasting illustrations from Forest Elementary School and London Elementary School will help to clarify this point. Both schools were involved in implementing new, district-mandated, externally developed mathematics programs. Student performance in mathematics at Forest Elementary was below average levels for the state, and the school was not currently satisfying Adequate Yearly Progress expectations; nonetheless, school personnel did not explicitly identify improved achievement in mathematics as a goal. Instead, the goal (one of many program-specific goals in the school) was simply to implement the district-mandated Grade 6-8 mathematics program. A district mathematics consultant visited the school weekly to assist math teachers with implementation. At the same time, two potentially related initiatives were underway. First, the school counselor was preparing student assessment data reports at the beginning of the year, to assist teachers with lesson planning and tracking student progress. These reports were to reach teachers a few weeks prior to state testing dates so that teachers could identify students who might need additional coaching. The principal was reportedly keenly interested in student performance data, though no one could identify any actions that she had taken to influence the use of those data. Second, the school technology coordinator had been trained by district staff to facilitate the implementation of a computerized curriculum mapping and lesson-planning tool. The interview data for Forest Elementary School did not indicate that these strands of activity and the leadership sources and actions associated with them were deliberately coordinated. The result, from a teacher‘s perspective, was a leadership distribution pattern of anarchic misalignment (see Leithwood, Mascall, et al., 2007).

In London Elementary School, the principal‘s vision and goals included improving student success (not limited to mathematics), greater coherence in curriculum and teaching, and improved teamwork focused on student learning among teachers and with other stakeholders (e.g., parents). Although the percentage of London Elementary students performing at or above state standards in mathematics was acceptable (and high, relative to similar schools in neighboring districts), the principal‘s goals emphasized the success of all students and the need to boost learning outcomes beyond those touched on by the tests. Consultants working for the commercially developed mathematics program visited the school every six weeks to provide implementation training and assistance for the teachers. Not unlike the faculty at Forest Elementary, London Elementary faculty members were engaged in a curriculum project (mandated by the district but organized internally) that involved writing curriculum guides and common assessments keyed to the state curriculum in core subject areas. The principal arranged for the writers to get input from external program consultants. She relocated the writers‘ classrooms to ensure that all teachers had convenient, informal access to them for advice. Not only was the principal committed to the use of assessment data for identifying and addressing student learning needs, she delivered data-use training for teachers, and she sat in on grade-level team meetings to facilitate teachers‘ use of assessment data in their planning of six-week tutoring cycles. She also arranged for the parent coordinator to get trained in the mathematics program so that she could prepare ways to show parents how to help their children with mathematics homework. With the exception of the parent involvement piece, the activities related to implementation of the mathematics program in London Elementary were similar to activities at Forest Elementary (external program with in-service training, curriculum mapping aligned to state standards, assistance with data use). At London Elementary, however, these activities and varied sources of leadership were linked in a complex, collective pattern through the principal‘s actions. The overall effort encompassed multiple, core leadership practices (setting directions, developing capacity, workplace arrangements, managing instructional program) and multiple leadership sources associated with the focus on a shared learning goal. The pattern at London Elementary seems likely to produce a greater impact on student learning in mathematics than the pattern at Forest Elementary, where the focus was limited basically to program implementation. The leadership distribution scenario at London Elementary corresponds well to the concept of planful alignment across core leadership practices (Leithwood, Mascall, et al., 2007).

Student Learning and Leadership Distribution

No general claims about the relationship between student learning and school leadership distribution can be made on the basis of evidence derived from qualitative research at five schools. We did not find any obvious relationship between alternative patterns of distributed leadership and state test performance of students in each school from 2002/03 to 2005/06. We, however, consider two explanations for the apparent lack of any relationship related to distributed leadership: changes in leadership personnel, and within-school variation in leadership distribution.

First, any attempt to associate different patterns of distributed leadership with student learning must take into account the potential consequences of changes in key leadership positions. Among the five schools, only one of the principals had been in her position (at Forest Elementary) for more than two or three years. Teachers in London, Overton, and Gregory Elementary alluded to differences in leadership styles, distribution, and practices between the previous and current principals. The impact that these changes in leadership might have on student learning would not necessarily show up in the first year or two of the principals‘ tenure.

Second, our case study findings highlight the need to be sensitive to the focus and scale of leadership distribution and action as they relate to student learning. At the microlevel of specific goals and leadership tasks, different patterns of distribution across leadership sources and actions often co-exist in a school (e.g., improvement in mathematics and reading performance at London Elementary). It would be a logical error to infer that leadership as it is distributed and practiced for one leadership scenario, such as leading a new reading initiative, would necessarily be similar to leadership distribution across other scenarios, such as changes made in the science curriculum. The influence of more general concepts and approaches to leadership distribution on student learning outcomes, such as collective leadership (Section 1.1), shared leadership and professional community (Section 2.2) are more easily and empirically measurable than specific forms and arrangements of distributed leadership.

Implications for Policy and Practice

Four implications for policy and practice emerged from this section of our study.

  1. Efforts to promote greater sharing or distribution of leadership need to operationally identify specific or desired leadership patterns. Simply invoking the term distributed leadership is meaningless, given the many different patterns distributed leadership can take. To understand the distribution of leadership one needs to explore evidence of actual behaviors and influences associated with core leadership practices and specific focal points of school-improvement activity. Principals working in similar organizational structures may enact their leadership roles and engage in distributed leadership in quite different ways.
  2. It would be a serious mistake at this point to "harness" any major school reform effort cart to the distributed leadership "horse." While we now have a better understanding of some patterns of leadership distribution as they operate in practice, evidence about the effects of leadership distribution on schoolimprovement initiatives or student learning is extremely modest. That said, other evidence (see Sections 1.1, 1.2) does suggest that principals‘ sharing of leadership with others in planful, yet diverse, patterns of leadership distribution is probably a worthwhile way to approach improvement in student learning.
  3. The task of encouraging more leadership distribution in schools should be viewed, first and foremost, as the task of nurturing principals‘ dispositions toward such leadership. As school principals enact leadership roles, the beliefs and orientations they bring to the task matter a great deal. The extent to which leadership will be distributed in schools, and the forms it may take, are determined in large measure by what principals believe and feel about the key factors that come into play: external and internal influences on school direction setting, sources and uses of professional expertise (their own expertise, teachers‘ expertise, expertise from external sources), and participatory or shared leadership.
  4. Distributing leadership more widely in schools should not be viewed as a means of reducing principals‘ workload. Leadership from teacher leaders and external sources is more likely to be goal- or initiative-specific. Principals, on the other hand, are responsible for a boundary-spanning role not typically performed by others, nor picked up by others in the absence of active principal leadership. Principals are typically involved in a great many leadership initiatives in their schools, including initiatives for which others have assumed lead roles. Their role to coordinate or link others‘ leadership efforts is essential.

< < Previous | Next > >


106. MacBeath (2005) and Spillane (2006)

107. Leithwood et al. (2007)

108. Justification for these categories is provided in Leithwood & Riehl (2005); Leithwood, Louis, et al. (2004); and Leithwood & Jantzi (2006).

109. Spillane (2006).

110. Student achievement rankings calculated by comparing the percentage of students scoring at or above minimum state proficiency standards on state-mandated assessments in reading and mathematics (2002- 2005) relative to other schools in the states where these schools are located.

111. Diversity (Low=66%+ White; Medium=18%-65% White; High=0-17% White); Poverty (Low=0-17% F/R lunch; Medium=18%-65% F/R lunch; High=66%+ F/R lunch).

112. Evidence of this is provided by Camburn, Rowan, & Taylor (2003); Hall (1992); Heller & Firestone (1995); Leithwood et al. (2004a); and Spillane (2006).