Comparing principal autonomy in time and space: modelling school leaders’ decision making and control

ABSTRACT This article presents the development and application of a model that can be utilized to compare the autonomy of principals in various historical and national contexts. Drawing on former conceptual work on autonomy in education, the model conceptualizes principal autonomy as two-dimensional. The first dimension is the decision making expected from principals, manifested on a continuum of being restricted or extended. The second dimension is the control imposed on principals, also varying between a restricted or an extended nature. In relation to each other, restricted/extended decision making and control can result in four different governance configurations that impact principals’ work complexities. We apply the model for the description of the development of Swedish principals’ work conditions over time and for a comparison of the autonomy of Nordic principals. With this endeavour, we demonstrate how historical and nation-specific particularities in educational governance interact with limitations and opportunities of educational leadership. Moreover, we argue that a focus on the autonomy of various stakeholders in schools might be a fertile strategy for further understanding the confusing concept of school autonomy.


Introduction
The aim of this article is to develop and apply a comprehensive model of principal autonomy for the purpose of understanding and comparing the impacts on various context-specific particularities on practices of educational leadership.With this endeavour, we aim to contribute to the further investigation of the complex and confusing nature of school autonomy in curriculum work and practices of education leaders. 1 Our elaboration of the model draws on theoretical and empirical work on teacher autonomy and considers recent research on autonomy in education.When applied to educational practice, autonomy is a complex concept that can mean a multitude of things.First, when applied to the school level, it becomes evident that schools are complicated social systems in which multiple actors operate in different roles.Concomitantly, the scope of action of some individuals may affect the decision-making capacity of others.The question of who in a school community has that autonomy (e.g. the teachers, the principals, or the learners) has fundamental implications for the ways in which the school operates in different matters and especially in curriculum work.Moreover, the matters over which the members of the school community enjoy CONTACT Wieland Wermke wieland.wermke@specped.su.se; tine.proitz@usn.no;wieland.wermke@specped.su.se; daniel.nordholm@edu.uu.se; maria.jarl@gu.seSpecial Education, Stockholm University, Stockholm, Sweden autonomy have important implications for what school autonomy means in practice (Salokangas & Ainscow, 2017).
In the last few decades, the concept of autonomy has elicited increased attention in research on curriculum work, school improvement and school leadership (cf.Wermke & Höstfält 2014;Heffernan, 2018;Heinrich, 2015;Mausethagen & Mølstad, 2015;Uljens, 2015;Wermke & Forsberg, 2017).Comparable to efforts found in research on teacher autonomy and how this concept could contribute to educational change (e.g.Hattie, 2011), researchers have approached the concept of school leaders' autonomy from different starting points.For instance, some studies have revealed a positive correlation between leadership autonomy, strategic decision-making and student performance (e.g.Fuchs & Wössman, 2004), and between increased autonomy and curricular changes that improve students' results (Caldwell & Spinks, 2013).
The paper argues for an understanding of autonomy as decision-making capacity and control combined to complementary dimensions, which can be extended or limited via various government configurations.Seeing both dimensions on a continuum opens each one and displays various governmental configurations.These configurations can serve as independent variables in future studies on principals' work experiences in different contexts.From our point of departure we take one step back and begin by discussing principal autonomy and the dimension of the concept.In the following sections, we present previous studies on various forms of autonomy in education, pinpointing potential conceptual issues and possible solutions in this field.Then, we develop a model of principal autonomy as a context-specific, multidimensional construction.Third, we elaborate on the model's usefulness by describing Swedish principals' autonomy from a historical perspective and comparing principals' autonomy in a selection of Nordic countries.Finally, we discuss possible ways to go further with our conceptualization.

Previous studies on autonomy
Autonomy is a worthy concept for investigating the complex relations and the possible and actual agency of various stakeholders in a school system (Paulsrud & Wermke, 2019;Vanderstraeten, 2007).Autonomy is a widely used concept in education policy and practice.The etymology of the concept derives from the Greek word autonomos, which means 'having its own laws' (Oxford Dictionaries, 2021).The ability and capacity to self-rule among different actors in education, as well as the governance models that might constrain or enhance individual and group capacity to self-rule, stand at the centre of the autonomy debate (Wermke & Salokangas, 2015).
In the literature on teacher autonomy, autonomy is defined as self-governance that is closely related to teacher agency (Priestley et al., 2015), which refers to the capacity of an agent to determine his or her own actions through independent decisions within a system of principles and laws.Autonomy is something that is actively exercised rather than passively received.Autonomy in education is affected by different context-specific particularities, such as different regimes of governance (Parker, 2015;Wermke & Salokangas, 2015;Wilches, 2007).Education leaders' scope of action must be understood more deeply to comprehend the relation between different entities in education because their relation is the most significant connection between the loosely coupled subsystems (Weick, 1976).Moreover, educational leaders today must often handle hybrid governance systems characterized by a hierarchical political model with top-down decision making from the state and a bureaucratic model for municipality administration (Moos et al., 2016).
Notably, in the literature on school autonomy and leadership autonomy, these concepts are not neatly distinguished but often used together.Ko et al. (2016) have argued for a dimensional relation of school autonomy and school leadership.In their view, school autonomy and accountability are important dimensions of school leadership.It is argued that autonomy must be combined with profound leadership, comprehensive continuous professional development and a positive, collaborative school climate.In reviewing the current body of research, Cheng et al. (2016) have argued for the need for theoretical innovation in research on school autonomy.Too little attention has been paid to cultural autonomy and internal structural autonomy at individual and group levels, as well as to potential conceptual links between school autonomy and learning outcomes.Thus, Cheng et al. (2016) have posited that internal autonomy could be as important as external autonomy for school success.Meanwhile, Hanushek et al. (2013) and Neeleman (2019) have provided additional examples of school autonomy and principal autonomy being frequently intertwined and used as overlapping concepts.According to Hanushek et al. (2013, p. 216), a principal's decision-making capacity is but one of several aspects of school autonomy: a school has full autonomy in a certain domain when either the principal, the school board, the department head or the teacher 'is the only one to carry responsibility '. Neeleman (2019, p. 34) has argued that increased school autonomy and intensified accountability standards 'have led to an increase in school leaders' decision-making responsibility'.Cheng et al. (2016) and Hanushek et al. (2013) have both made significant conceptual contributions on school autonomy.Their models elaborate on three dimensions each: functional, structural and cultural autonomy (Cheng et al., 2016), as well as autonomy regarding academic content in personal decisions and budget allocations (Hanushek et al., 2013).They do not, however, elaborate which matters or tasks could be decided by different actors, for instance, the leader, the school boards, middle-tier administration or teachers' cooperative decision making.
Different local governance systems have different impacts on a school leader's autonomy, and studies show that the perceived autonomy can differ within the same municipality (Saarivirta & Kumpalainen, 2016).School district leadership autonomy is linked to clarity regarding responsibility and authority in the steering chain, expertise, as well as trust and accountability among local politicians, superintendents, mid-level managers and school leaders (Nihlfors & Johansson, 2017).Previous research have shown that the phenomenon of autonomy is multifaceted and that the concept can have a shifting nature in a school context (e.g.Ingersoll, 1996;Wermke & Forsberg, 2017).In this regard, in elaborating the relationship between autonomy and control, Cribb and Gewirtz (2007) have proposed that one approach to understanding the concept is to pay certain attention to the 'loci' of autonomy and control.Regarding teacher autonomy, Cribb and Gewirtz have argued that the locus is the teacher (individual or group level), but it can also have a number of sources at the local, regional and national levels (e.g.school management, peers, parents, local school governance, steering documents).
As the literature attests, to understand the complex organizations of schools, we need ideas on who decides on what in more detail.To illustrate this further, increased school autonomy or local autonomy, as witnessed, for example, in relation to the introduction of independent schools, such as the Friskola movement in Sweden or the Academies movement in England, does not automatically grant teachers or principals an increased scope of action (Dieudé & Prøitz, 2022;Kauko & Salokangas, 2015).
Following this line of thought, this article takes an analogous point of departure, claiming that the duality between autonomy and control becomes equally important for school leaders.

Developing a model for comparing principal autonomy
The theoretical perspective of this article draws on the work of Wermke and Salokangas (2021) on teacher autonomy from a comparative perspective.Their model was built on the reading of Vanderstraeten (2001Vanderstraeten ( , 2007) ) on organizational framed interactions in education systems and of Ingersoll (2003) on power and control in school organizations.This article presents the development of a model of principal autonomy that relates decision making and control of the decisions to each other.From our point of view, these two dimensions are complementary.Consequently, both dimensions can vary, for example, be high and low at the same time.As a result, various autonomy and control configurations correlate with different levels and qualities of complexities and risks in principals' work.This will be elucidated below.

Autonomy in education: complexities and risks
Building on the seminal works of Vandertstraten and Hopmann on important premises of the nature of public education, Wermke and Salokangas (2021) have presented the following considerations of autonomy in education.Developing between at least two persons, a learner and a teacher, public education is a peculiar type of act of education.It can be perceived as an institution in which groups of people come together to learn appropriate behaviours, skills and/or contents to participate in society.Consequently, large masses of people must learn predetermined taught content (Hopmann, 1999;Vanderstraeten, 2001).
Moreover, education builds on interactions.It is commonsensical in sociological research that an interaction is not 100 percent plannable.For example, Goffman (1966) has highlighted that we can never be sure where and how an interaction will develop, and following Luhmann's (1993), we could say that individuals are only able to process communication self-referentially (cf., Vanderstraeten, 2001).This means that we do not know what others think; we only know what they say and we only know what consequences this has for us, being self-referentially alone.In education, this results in ambiguity between what is taught and what is actually learnt (Hopmann, 2007).When we acknowledge this, we must recognize how we understand autonomy in education or, in other words, how autonomous professionals in public education must and can be to make education possible in such conditions.Vanderstraeten (2001), heavily inspired by the German sociologist and functional systems theorist Luhmann, has discussed how an educational system-here seen as a subsystem of the overall system of society-is responsible for the education of society members, how it emerges, or, in his words, 'makes public education possible'.Furthermore, he has argued that the overall system of a society can be seen and divided into smaller differentiated subsystems or the so-called functionally differentiated systems.Conversely, a functionally differentiated system can be seen as a system that fulfils a purpose or a function in relation to a greater and more abstract system that is highly complex (i.e. the society).From a functional perspective, this is interpreted as education needing a certain amount of autonomy to be able to communicate and to develop circumstances for its core function, which is education (Vanderstraeten, 2001).In other words, education, in relation to other functional systems, requires a scope of action to be able to react quickly and appropriately to educational needs that become visible in interactions.Moreover, resources such as facilities are needed to make frequent interactions possible, and a certain degree of hegemony is required for educational solutions.Solutions cannot be approached only through financial or legal means but must have an educational focus.Furthermore, incentives such as internal motivation or certificates that empower students in tackling the future are needed.All these elements add to the complexity of educational organizations.
Therefore, the education system and the overall school structure are built by what Vanderstraeten has called organizationally framed interactions (2001), namely, interactions that are plannable to a certain extent.The education system is characterized by face-to-face social meetings, which open up particular forms of interactions within the system (ibid.).In every school, for example, teachers interact with their students, other teachers, parents and/or the school principal regarding the different matters occurring in the school's everyday work.As described earlier, an important factor is that an interaction never takes the exact same turns, nor can it be seen from the exact same perspective.Instead, it depends on the situation in which it occurs.Thus, an interaction takes on a life of its own depending on the people who interact with one another and the context in which the interaction takes place (ibid.).
Consequently, on the one hand, the professionals in a school need a particular scope of action to be able to respond to certain reactions from students, parents or colleagues.On the other hand, they also need a frame, which decreases the complexity of possible reactions in interactions.Simply put, the 'school' itself cannot handle the social interactions of education but needs specially trained people to do this, namely, teachers and principals who can communicate educational objectives.However, public education professionals need the school to reduce the complexity of possible interactions.For example, due to the school organization, teachers do not have to search every morning for children who are interested in learning.They can regularly meet the same students, who are of the same age, and educate them on certain subjects regulated by the curriculum.Finally, they are paid for their services by the school (Vanderstraeten, 2007).Here, the school principal plays a significant role as the nexus between the educational profession and the organizational structures of the school.
Considering this, this study builds on the following assumptions on autonomy in education: (1) Public education is a social phenomenon that must cope with the problem of planning of interactions for learning.
(2) Professionals in public education need a certain scope of action to formulate their decisions in interactions on the reaction of students in their educational day-to-day life.
(3) To reduce the complexity of possible interactional and educational outcomes, professionals in education have to rely on an organization that helps them reduce the possible complexities in education.An organization (school) thus emerges, for example, teachers meeting the same students on a regular basis at the same place and knowing what they are expected to teach.Principals have a particular position as the nexus between the teaching profession and the organizational framework.(4) Professionals in public education must handle an immense number of students.Unlike established professions, such as medical doctors or lawyers who deal with one customer or patient at a time, educational professions mainly deal with large groups of students.The school absorbs many of the risks associated with such high numbers.

Decision making and control in education
Next, we put forward the work of Ingersoll, a scholar with an intensive interest in schools as organizations.His work has also been presented in more detail by Wermke and Salokangas (2021).Ingersoll (1996Ingersoll ( , 2003) ) has considered that power within an organization belongs to those who control the most important decisions that are to be made.Hence, autonomy has been viewed by Ingersoll (2003) as the stakeholders' control and influence over the key decisions that affect the character, content, processes and evaluation of their daily work.To analyse autonomy, one must define which are the most important decisions that affect the work and purpose of the school organization, as well as determine which criteria should be used for evaluation and how one should examine who controls these decisions.
From a perspective of organization theory, Ingersoll (2003) has claimed that every workplace faces similar challenges in defining, supervising and rewarding the tasks undertaken by its employees to enhance performance and productivity.A certain amount of coordination and control is necessary to achieve organizational goals.Similarly important are the consent and cooperation of the individual members of the organization.According to Ingersoll (2003), the problem of finding the proper degree of control and consent is a cornerstone in organization theory and essential for avoiding demotivation, disorder and low performance.This balance can also be described as the level of bureaucracy within the organization.Ingersoll (2003) has put forward Weber's work as central in this field.Weber has viewed bureaucracy as built on rationality, efficiency and hierarchy, characterized by standardization and control, but also as dependent on loyalty and cooperation.From this viewpoint, some organizations are better suited to bureaucracy than others.Schools are, according to Ingersoll (2003), the archetype of loosely coupled systems-complex organizations that, on the one hand, require bureaucracy in the form of coordination and control, but on the other hand, involve human relations and tasks that are not easily bureaucratized.In such a system, the need for bureaucracy conflicts with the employees' need for flexibility and autonomy, leading to a more intense tension between control and consent than in many other organizations.Conceptualizing autonomy empirically in terms of important decisions and their control is very fruitful.What is important here is the question of which aspects and domains of professionals are allowed to decide for themselves, which is why we must discuss further what we call 'autonomy of autonomy' and potential 'autonomy paradoxes' (Wermke & Salokangas, 2021).Herein, Ingersoll (2003) has proposed a distinction between important and unimportant decisions.
In Ingersoll (1996Ingersoll ( , 2003) ) has focused on control mechanisms.According to him, a control mechanism is needed to define the tasks expected to be carried out by employees and determine whether the results are satisfactory.Furthermore, sanctions should be imposed on employees who have not performed the tasks accordingly.Professional work in education is regulated by a number of rules.Besides standardized local curricula and instructional policies, teachers and principals are expected to enforce behavioural rules for students, often formulated with minimal teacher input.However, Ingersoll (2003) has also claimed that school administrators simultaneously lack the power to enforce the rules and impose sanctions on teachers, implying that there are a few means to reward good performance.Thus, other ways of steering the professionals in schools must be found.Ingersoll (2003) has pointed out that organizational control is not always visible.For example, the hierarchical structure of the organization in itself restricts the power and influence of those members who simply do not have access to the areas of work where important decisions are made.Moreover, rules, regulations and direct control might even be less effective than norms, expectations and other forms of subtle control in organizations.Hence, the absence of visible control mechanisms and formal standardization of working tasks may seem to be an indicator of decentralization and autonomy, although the power within the organization is highly centralized in reality.Ingersoll (2003) has argued, for example, that a principal's most effective means of achieving control is by appearing to have the power and capacity to sanction teachers.Moreover, from Ingersoll's (2003) point of view, teachers are dependent on their principals in a number of issues, especially in the key area of student socialization.Since teachers have limited authority in the disciplining of students, they need their principals to take their side in conflicts with students in order not to appear weak and lose credibility.
Taking this reasoning one step further, Ingersoll (2003) has considered that loose regulations might make professionals in public education vulnerable and that top-down regulations and bureaucratic control might not necessarily be discouraged.Instead, regulations may constitute a buffer zone, protecting the professionals from accountability, conflicts with students and parental pressure because they clarify where teachers' responsibilities start and where they end.This reasoning harmonizes with Wermke and Forsberg's (2017) idea that centralized steering and regulations reduce the complexity of teachers' work; they also have implications for our reasoning on principal autonomy.
As Ingersoll has suggested, for the complex system of education to work, a delicate balance between control and consent is required to ensure equitable and fair education whilst not regulating and controlling the work of professionals in public education too much.The professionals need autonomy to be able to handle the complex and interactive work of teaching, which is impossible to organize from the outside in detail (see , Hopmann, 2007).However, there can also be too much autonomy.This is why Ingersoll (2003) has emphasized the significance of determining which decisions are most important, for example, for teachers, as they are not necessarily the same ones the school organization considers important.Moreover, Wilches (2007) has argued that undesired responsibilities should not be considered as autonomy in their proper meaning.Instead, he considers that autonomy concerns the right to make professional decisions within certain frames.From such a viewpoint, autonomous professionals in public education should have a say in how responsibilities are distributed.
Consequently, we conceptualize formal control as something that frames and determines professional actions through different material determinants.Therefore, the control elements are as follows: national and local governance documents (curricula, school laws or circulars), centrally administered tests and exams, and subject and pedagogical knowledge provided by research.Wermke and Salokangas (2021) have reported that in all their investigated national contexts, teachers express that such sources determine their decision making, mostly in the educational domain.In other words, according to Wermke and Salokangas (2021), professionals in public education do what they are expected to do by relating their decision making to key governance documents and scientific knowledge.This finding can be understood with the help of a complexity and risk perspective.Drawing on legal documents and plans, as well as scientific knowledge, helps educational professionals to reduce the complexity in their work.Decisions become less complex when one accepts the given frames.Moreover, following the rules and regulations of an organization in its decision making means that afterwards, failures and wrong decisions can be ascribed to those frames (Vanderstraeten, 2007).
In relation to the aforementioned arguments, Wermke and Salokangas (2021) have put forward that decision-making capacity can be limited by providing guidance through documents, regulations and bureaucracy.Such structures reduce the complexities and risks that have to be taken.Conversely, the more decisions teachers and principals are expected to make, the more they become accountable and the more they may restrain themselves to cope with the risks of incorrect decisions, thereby creating an 'autonomy paradox'.Consequently, restricted autonomy as limited decision making can be experienced by teachers and principals as something positive.

A model for comparing principal autonomy
In line with our reasoning, we propose a model comparing principal autonomy.The model captures principals' possibilities for decision making and control, particularly the formal control imposed on principals and their work.The model also acknowledges that decision making and control can be limited or extended.Increased decision-making capacity enhances the complexity of principals' work, while increased control increases the risk of the job.Figure 1 elaborates on these aspects.

Decision making, control, complexity and risk and a comparison of national principal professions
The first section (I) applies to principals who have very complex tasks and who are assigned with considerable decision-making capacity.By contrast, the control of principals' work by different governance technologies is rather limited.The second section (II) refers to principals who have very complex tasks, including a plethora of decision-making responsibilities.In this type, the profession's work is also controlled in an extensive manner.The third section (III) describes lower complexity and intense control, meaning lower complexity in relation to types I and II.In section IV, principals are not subject to much control, but their decision-making capacity is also somewhat limited compared to type I and II principals.
It is important to emphasize that the model presented in Figure 1 has a comparative character.By no means does it state that principals' work are not complex or that they are not associated with any risks.It states that some principals must, due to varying governance regimes, bear lower or higher risks/complexities.We elaborate further on this by exemplifying how the model can be used and by simultaneously testing the theories behind the model empirically.We aim to highlight how various education governance regimes that differ in time and space might lead to varying working conditions and discretion of principals.

Contextualizing autonomy in time: the autonomy of Swedish principals (1960-2020)
Next, we utilize the model to analyse the autonomy of Swedish principals for an extended period.The section builds on an extensive re-analysis of literature on the history of Swedish principals.Our journey starts in the 1960s, which saw the introduction of the comprehensive school system, and ends in the present time.This case, which has partly been developed by Jarl et al. (2012), builds on secondary sources, such as research reports and government documents (reform proposals, bills and regulations).We will employ our model shown in Figure 2.

Swedish principals' decision making and control in different time periods
From the 1960s to the 1980s, principals' decision-making capacities were limited due to detailed state regulations.Centralized governance was seen as a prerequisite for achieving the political goal of 'a school for all'.However, principals started a professional project centred on the idea that educational leadership requires certain skills and competencies that separate principals from teachers, which paved the way for increased decision-making responsibilities in the following period.Decentralization reforms of the 1990s entrusted principals with the overall responsibility to organize and lead the day-to-day activities of the school to achieve the national objectives identified by the central government.Until 2006, government agencies supported the improvement activities of municipalities and schools, while inspections were largely missing.In the second half of the first decade of 2000, there were more frequent demands for accountability and inspections.Accountability reforms in the late 2000s and early 2010s undoubtedly increased the control of principals' work.At the same time, these reforms did not limit principals' decision-making capacities, as they were still very extended.We elaborate in detail on this historical journey below.

Principals in the social engineering era (1960s to 1980s)
In the 1960s, a comprehensive school system replaced the parallel school system.The parallel school system consisted of seven-year elementary schools and secondary schools which, after becoming somewhat simplified, attracted students from different social classes.Having one comprehensive school for the whole population was an instrument for the state to foster an equitable society.The notion of a strong central state with detailed state regulations accompanied the reform.The reform included a centralization of responsibility for education.The national curriculum and the syllabuses left little room for local enactment by teachers and principals.Teachers and principals were state employees.The state allocated resources to schools via a central government agency called the National Board of Education (NBE) and the board's regional offices.From a comparative perspective, the Swedish educational system during this period has been argued to be among the most centralized systems in the western world (Lundahl, 2002).
The elementary and secondary schools of the parallel school system did not only have students from various social classes; these schools were also populated by teachers and principals from various cultural and educational backgrounds, which were merged under the comprehensive school reform.The comprehensive school reform fuelled an increased sense of professional belonging among principals from different school forms, as well as an institutional separation of teachers and principals.While principals and teachers used to be considered representatives of a single occupation, principals were seen as the first among equals (the most qualified among colleagues), which, in a sense, gave them a superior position in relation to the teachers (Johansson & Fredriksson, 1993).Principals almost exclusively had a professional background in teaching, and principals and teachers were co-organized in several different unions (Pousette, 2006;Ullman, 1997).In 1966, the Swedish Association of School Principals and Directors of Education was established.As demonstrated by Jarl et al. (2012), the founding of this association marked the beginning of a journey towards a common professional project for principals and the end of a long period of cultural and class-related tensions among different categories of principals (see also, Ullman, 1997).
Further, this professional project was supported by the introduction of a state-led education programme that was exclusive to principals.The programme started on a limited scale before it was implemented nationally in 1976 (Johansson, 2001).As posited by several scholars, the education programme signalled a newly awakened interest among state actors to influence principals' work (Johansson, 2001;Rothstein, 1996;Ullman, 1997).By introducing an education programme directed only to principals, the state indirectly supported their efforts to establish a professional knowledge base.However, the programme's objective was to educate principals to become state servants for implementing education policies.It did not want to educate representatives of a distinct occupation whose practice would be based on professional judgements (Johansson, 2001).

Principals in the decentralization era (1990-2006)
At the beginning of the 1990s, the educational system underwent far-reaching decentralization reforms that transferred the responsibility for education from the state to the local level (i.e. to the municipalities) and to schools.The responsibility for hiring and paying teachers and principals was transferred from the state level to the municipalities, and the responsibility for organizing public education and for allocating resources to schools was decentralized from the state level to the municipalities.The introduction of a voucher system and freedom-of-choice reforms at the beginning of the 1990s paved the way for the expansion of independent schools (Lundahl, 2002).Parents and pupils became entitled to choose among schools and were not, unlike before, required to simply enrol at the nearest school within the municipality.Technically, every pupil was given a voucher for his or her education.Thus, school resources were dependent on the number of pupils enrolled.
Management by Objectives and Results (MbOR) was introduced as the guiding principle for educational governance: the main role of the state was the identification of the overall objectives of schooling and the expected learning outcomes in different subjects in the national curriculum.Municipalities, independent school providers and principals were mandated to decide on the appropriate organization of education at the municipal and local levels.Decentralization reforms included the replacement of the NBE by a new government agency for education, namely, the Swedish National Agency for Education (NAE).The NBE symbolized an obsolete state bureaucracy, while the NAE was seen as a manifestation of new management ideas (Jacobsson & Sahlin-Andersson, 1995).Initially, the new agency took a rather passive role to shield municipal and professional discretion (Magnusson, 2018).
As postulated by Jarl et al. (2012), these reforms increased principals' decision-making responsibility and fuelled the institutional separation of principals and teachers.Decentralization reforms turned schools into smaller organizational units: the number of managerial positions and the actual number of principals increased (Ullman, 1997, p. 233;Ekholm et al., 2000).While most principals still had a background in teaching, reforms led to principals being recruited from school backgrounds that, traditionally, had not functioned as a training ground for principals, particularly preschools and primary schools (Ullman, 1997).Furthermore, there was a change in regulations on the qualifications for the job because 'educational experience' replaced the requirement for a teacher examination, as both pedagogical qualification and teaching experience became the formal eligibility criteria in 15 out of 19 OECD countries (Pont et al., 2008, pp. 183-186).Sweden and a few other countries seemed to be exceptions to the rule.
The reforms also underscored that principals served two entities: the state and the subordinated local authorities (municipalities).The state level defined the principals' general responsibilities in relation to the objectives in the curriculum; they were responsible for ensuring that local schools accomplished the national objectives.In addition, they had responsibilities from the municipality since both principals and teachers were municipal rather than state employees.As a consequence, the hierarchical relationship between teachers and principals changed: principals became the teachers' employers rather than their colleagues (Norberg & Johansson, 2007).Empirical studies report a shift in the principals' occupational values as they tended to view themselves as representatives of the municipal administration rather than of the teachers' collective (Jarl & Rönnberg, 2010).The ultimate expression of the principals' new jurisdiction was that they became responsible for determining the individual salaries of teachers.
Given that they had become municipal entities, tasks such as personnel administration, recruitment, salary negotiations, budgeting, and distributing resources among school classes, as well as organizing and planning the local work of the school, were concentrated to each school's central management.From an OECD perspective, Swedish principals were reported to have a high degree of autonomy in budgetary issues.Sweden was above the OECD average in terms of students enrolled in schools, where school leaders had 'full autonomy in deciding how their budget was spent' (Pont et al., 2008, p. 53).In addition, studies have shown that principals believed that their administrative and managerial duties increased at the expense of responsibilities related to teaching and education (Ekholm et al., 2000;Norberg & Johansson, 2007).

Principals in the accountability and re-centralization era (from 2007)
The 2006 parliamentary election resulted in a right-wing government of an alliance of four parties.The leader of the Liberal Party, which had accountability reforms ranked high on his agenda, was appointed the Minister of Education.Once in office, the government initiated an investigation of the overall structure of agencies in the education sector.Major reforms emerged as a result.First, a new agency was introduced in 2008, namely, the Swedish School Inspectorate (SSI).The SSI overtook the responsibility for educational inspections from the NAE, and the inspection work was intensified (Carlbaum et al., 2014).The SSI conducted regular inspections of individual schools on a four-year cycle.
In addition to formal inspections of schools' work, the state intensified several other techniques of school performance evaluation.A new curriculum with lesser and more distinct overall objectives for education was introduced.In 2010, the Education Act of 1985 was replaced by a new education system that imposed an obligatory evaluation system on municipalities and schools.It became mandatory for municipalities and schools to evaluate the quality of education in line with a specific model labelled systematic quality work (Lundström, 2015).The national assessments of student performances were extended.Several subjects were added, and additional age cohorts were included in the testing.Since the first half of the 2010s, large-scale international comparisons of student achievements, such as PISA or TIMSS, have become increasingly important to Swedish policy making (Forsberg, 2011).
While these reforms imposed an increased control of principals' work, principals were still entrusted with an extended decision-making capacity.There are signs that the complexity of principals' work increased even more.As mentioned above, marketization reforms at the beginning of the 1990s led to an expansion of independent schools.During the first decade of the 21st century, the effects of market reforms became increasingly evident at the school level.For example, marketization led to increased competition of students between schools because student vouchers were valid for both independent and public schools.Competition was also heightened because for-profit schools were finally permitted to operate.Competition meant that the schools with the best attributes attracted the most students, thereby meeting the challenges of the school marketplace (Stenlås, 2011).In a school system where parents and pupils independently decide where to start or continue education, the loss of pupils might jeopardize the existence of an individual school.Moreover, such forms of governance transformed the identities of all stakeholders, including teachers and principals (Norén, 2003).Clients became customers with increased rights (Fredriksson, 2010).

Contextualizing principal autonomy in space: Nordic principals compared
Drawing on empirical work on Nordic principals, we present a further application alternative for our model: a comparison of the autonomy of principals in Sweden, Norway and Finland.Thus, this section focuses on the model's applicability for investigating the impact of nation-specific governance particularities on principals' autonomy.In doing so, we elaborate further on the potential cultural dimension of school leadership and how this dimension interacts with principals' work.Here, our model might be a fertile device for displaying such relations.
From a governance perspective, there are similarities and differences among these Nordic countries, which motivate the comparison.Sweden, Norway and Finland share specific traditions of the Nordic model, which ideologically emphasizes modernizing society in the name of science, rationality, equality and democratic participation (Prøitz & Aasen, 2017).Education is considered an important part of the modern Nordic society, and the state is viewed as responsible for education as a common good (Telhaug et al., 2006;Prøitz & Aasen, 2018).Consequently, the Nordic countries have developed extended comprehensive school systems with no streaming of students and where all classes of society would meet.The Nordic model draws on a robust centralized school system with a strong state.As part of international developments in the western world since the 1990s, decentralization and marketization have been implemented in the following 20-25 years in the respective school systems.These reforms have resulted in various hybrid governance models combining elements of centralized and decentralized governance regimes, which have had significant consequences for the working situations and function of principals (Moos et al., 2016).
The empirical material of this paper was collected in the autumn of 2017, in a research project on autonomy in education (Wermke & Salokangas, 2021) which formed the foundation for two Master theses (Kotavuopio Olsson, 2018;Andersson, 2020).In the first study, 5 Swedish and 6 Finnish principals have been interviewed (Kotavuopio Olsson, 2018), in the second 6 Swedish and 5 Norwegian (N = 22).The principals work mostly smaller towns on both sides of the Swedish-Norwegian and Swedish-Finnish borders, located in the upper North of the countries, where the schools share similar local conditions, rural, close embedding in the civil society, long geographical distances.As written earlier, we assume such a sample would produce many local similarities, but also illustrate national differences.The sample is small, but empirical valuable.Our co-authors, both principals by themselves, having Swedish-Norwegian and Swedish-Finnish cultural and linguistic background could approach principals, due to geographical distances often neglected areas.Moreover, there are not a plethora of principals working alongside the boarders of the respective countries in the very North.The principals worked in K-6 and K-9 schools.The number of students at the schools varied between 15 and 400 in total.All schools had both principal and deputy principal positions, while at the larger schools, there were even school management teams.Our principals worked in leadership positions for between 1.5 and 25 years.
The interview guidelines were constructed following the theoretical work of Ingersoll (2003), and used the following questions.
(1) What do the principals consider to be the most important decisions to be made regarding school function?
(2) Who do the principals consider makes these decisions (the principal, teachers, private or municipality actors, parents, etc.)?
(3) How are decisions monitored and followed up on, and who ensures that decisions are implemented 'correctly'?
The interviews were conducted by the authors of this paper in the principals' native languages (Swedish, Norwegian, Finnish), then transcribed and translated into Swedish (the common language of the project group).Afterwards, the interview transcripts were analysed following a deductive qualitative content analysis (Mayring, 2007, as also exemplified in Wermke & Salokangas, 2021, with a focus on teacher autonomy).We drew on the analytical device presented earlier in Table 1.A group of codes were firstly related to expressions of decision-making capacity and practices in the educational, social, developmental and administrative domains of principals' work.These codes represented teacher autonomy.Secondly, another group of codes were related to expressions of control of principals' work in the given domains.In addition, we coded these two groups in terms of their experienced intensity.For increasing the reliability of our results, at least two members of our study group performed the coding work in varying combinations.
The following quotes illustrate the typical answers regarding decision making: I make a [educational] plan comprising the themes that must be worked with during the term, one that is valid for all classes.[. ..]These are the educational tasks I take the lead for.All teachers make their own plans for their students.(Finnish principal) Actually, there are quite a lot of aspects in which I am allowed to decide.I rule over the whole inner organisation.How we organise the school, how we schedule our staff Regarding control aspects, the typical answers included the following: The control of the council is carried out through a monthly accounting system [. ..] economy, organisation in terms of the number of employees, sick leaves, working environment, relations to society, such as which communal services are employed, issues with school subjects, results of pupil surveys, national curriculum testing and health surveys [. ..]Such accounting does not constrain our freedom.The questions are easily answered and do not entail much work.(Norwegian principal) Do they control the student performances?No, since everything is about how the students reach their goals.However, this is a good question.Perhaps they should control more.(Finnish principal) Our analyses of the empirical material of this study using a framework of decision making and control have led us to the considerations illustrated in Figure 3.

Principals' decision making and control in Finland, Norway, and Sweden
Regarding the Swedish principals, we argue that, compared with their colleagues, their work is highly complex and risky.This is because among all three groups, they have an extended responsibility, with many important decisions to make.Due to high-stake national testing in Sweden, where student grades and results can be used by parents as the basis for choosing schools and pedagogical specializations are arguments in a schooling market, principals must be involved, together with teachers, in the domain of schooling.They also play a significant role in students' access to special support if needed.Regarding social questions, Swedish principals are responsible for a healthy and positive school climate, one that actively prevents bullying and takes students' psychological wellbeing into consideration.Furthermore, they are responsible for systematic school development and professional development for the staff, which must also be documented (systematic quality work).Administratively, the principals are responsible for recruitment, salary negotiations, renting and maintenance of school buildings.This is in addition to other tasks, such as decisions on promotion, scheduling and resource allocation.At the same time, principals in Sweden are also heavily controlled by various stakeholders: the state by way of the school inspectorate, the municipal administration and the parents with their above-mentioned extended 'consumer rights'.The extended decision-making requirements, which are simultaneously controlled, correlate with the risks of failures for which the principals will be made accountable, potentially by taking the blame or losing their position.That is why we placed the case of Sweden in the second section of our model (high complexity/high risk).This high complexity and risk scenario of Swedish principals can be best understood by comparing them with their Finnish counterparts who have fewer decisions to make in practice, a fact which we locate in the first section of our model (high complexity/low risk), quite close to the fourth section (low complexity/low risk).Although they also have significant responsibilities for many educational and developmental decisions, they can still draw on an autonomous teaching profession, making the majority of decisions themselves.Regarding social questions of schooling, Finnish principals make many decisions to achieve a healthy social-psychological climate in their schools.When it comes to administrative issues, they perceive a significant scope for decisions since they have an extended budget responsibility.However, they, like their Swedish colleagues, are not required to make decisions about facilities or negotiate teachers' salaries.Nevertheless, working with educational plans, scheduling and staff and school developments involve considerable complexities.
Regarding the control dimension, there are considerable differences.Put simply, Finnish principals appear to be less controlled and by fewer stakeholders.Finnish teachers do not experience any cross-pressure from the state and municipality administrations.They perceive themselves as municipal employees who are accountable to their local community.Moreover, there is minimal (and not done in public) national standard testing and no school inspections.The prominent explanation of Finnish teachers as trusted professionals is obviously also valid for Finnish principals.
Finally, we placed the Norwegian principals in the first section of our model (high complexity/low risk) but closer to section II (high complexity/high risk).By comparison, we found that Norwegian principals' position is more complex than that of their Finnish colleagues but less complex than that of their Swedish counterparts.They have an extended responsibility for educational and developmental planning, but there is a greater necessity to undertake this in a systematic way.Mandatory national standard testing must be undertaken, and the results must be reported, evaluated and followed up.School and staff developments must be planned in relation to the municipality's school strategies.Notably, Norwegian principals have an extended autonomy to handle the social aspects of schooling, with an explicit requirement for bullying prevention.In the administrative domain, they are in a similar situation to their Finnish colleagues, with budget autonomy but without the need for salary bargaining or facility management.In Finland and Norway, such aspects are addressed at the municipality level.
Moreover, Norwegian principals perceive a different form of control from their Nordic colleagues.They are apparently more extensively controlled than their Finnish colleagues but less so than Swedish principals.Like the Finnish principals, they see themselves clearly as municipal employees, embedded in their local civil societies.In Norway, there are control instruments, such as national testing and inspections of schooling practice, like in Sweden.The difference is that the municipality and the state are integrated, and principals participate more in the discussions of both testing and inspections (Wermke & Salokangas, 2021).

Discussion
This article has investigated the functionality of a model for comparing principal autonomy.We have analysed Swedish principals' autonomy from a historical perspective and compared it in a selection of Nordic countries.The two analytical applications have demonstrated that the model can be useful in demonstrating the dynamic nature of principal autonomy.The model understands autonomy as decision-making capacity and control combined to complementary dimensions, which can be extended or limited via various government configurations.Seeing both dimensions on a continuum opens each one and displays various governmental configurations.These configurations can serve as independent variables in future studies on principals' work experiences in different contexts.
Decision-making can be limited by governance structures (i.e.thorough documents, regulations, and bureaucracy).Such structures reduce the complexities and risks.Conversely, the more decisions principals are expected to make, the more they become accountable and the more they may restrain themselves to cope with the risks of incorrect decisions, thereby creating an 'autonomy paradox'.Indeed, the model follows a prototypical approach and should, therefore, be taken analytically due to its form that reduces the complexity of principal work into two dimensions of autonomy.Moreover, we have limited ourselves to a governance perspective, which means that we have examined, first and foremost, the relation of principals to government structures.Such limitations are important to bear in mind when discussing the ambitions of our conceptualization.However, the comparisons in complexity can support to make improvements in principals' working conditions.The anthropologist Kluckhohn has illustrated this point with an aphorism: 'The fish would be the last creature to discover water' (Erickson, 1986, p. 121).Comparative research, through its inherent reflection of the contingency of practice, makes the familiar strange and interesting again.The commonplace becomes problematic.What is happening can become visible, and it can be documented systematically (ibid.).
How can the understanding of principal autonomy as presented here be a vantage point for a further understanding of school autonomy?Our model of decision making and control could be applied to the conceptualization of autonomy of both teachers and principals (i.e. the model covers the two large professional groups in schools).It has also elucidated how different governance regimes interact with potential professional action.It is probable that the logic of autonomy as a form of decision making and control could also frame the analysis of school administrators' and superintendents' autonomy or be valid for special educators.This multidimensionality that only considers the complex stakeholder relationships in the eventual autonomous school might already challenge the value of 'school autonomy' as an analytical category beyond being a descriptive term of a particular education reform tradition, namely, decentralization.
In education governance research, it is, of course, useful to identify where certain important decisions are made, for example, at the state, municipal/district, or school level.However, such categories are far too simple to fully understand the complex relationships between policy and practice because schools are complex structures, formally and informally.The aforementioned ambiguities between leadership autonomy and school autonomy are not surprising.Both describe different aspects in varying detail.Being more detailed in terms of what constitutes school autonomy -teacher autonomy, principal autonomy, administrator autonomy and so on-could be a fertile strategy to expand the explanations of educational effectiveness or change.Moreover, the comparative methodological approach which our presented model draws on will deliver a suitable perspective for such theorizing work.
We also argue that principal autonomy must be investigated further, in line with a perspective that analytically sees autonomy and control as two complementary, not opposing, phenomena.Moreover, we emphasize the need for a more realistic perspective on principal autonomy and its potential consequences.Simply increasing principals' autonomy might not lead to the desired outcomes because increased decision-making capacity brings further complexity and risk to their work, possibly leading to anxiety, self-restriction, and the eventual rejection of autonomy, as seen in Wermke et al. (submitted).This is particularly true when extended autonomy is paired with extended control, as is often the case in NPM-inspired educational regimes.Finally, autonomy must be seen in its multidimensional nature.In some domains, principals might need and should be able to handle autonomy and control; in other domains, they may not (see, for example, Nordholm et al., submitted).Therefore, the analytical device in this comparative article might also be an invitation to conduct further research on such topics.This will entail knowledge on how different means and measures of governance employed by different actors interact with one another because more empirical studies and conceptual devices are needed that would look into the reality of school and public education professionals from a multi-governance perspective (Frostensson, 2015).Note 1. See for example, an interesting discussion on the relationship between school leadership research and curriculum studies on curriculum work by Uljens (2015) and his arguments on how curriculum construction implementation and evaluation, on different levels, must be understood as a form of educational leadership.

Figure 1 .
Figure 1.Decision making, control, complexity and risk and a comparison of national principal professions.

Figure 2 .
Figure 2. Swedish principals' decision making and control in different time periods.
[. ..]Actually, I decide over most things.[. ..].However, it is not like I want to rule in most things.I want to have the support of my staff, as well as the politicians and the administration [. ..]. (Swedish principal)

Figure 3 .
Figure 3. Principals' decision making and control in Finland, Norway, and Sweden