Decision-making in Context: Swedish and Finnish Teachers’ Perceptions of Autonomy

ABSTRACT This article presents the results from a survey investigating 708 Swedish and 1583 Finnish teachers’ perceived autonomy with a focus on the teachers’ perceptions of who makes the most important decisions in school. Teacher autonomy is seen as exercised at different levels; by teachers individually in the classroom or by teachers as a collective in school; and in different domains of teachers’ work, since the degree of decision-making by teachers is likely to differ between educational, social, developmental and administrational issues. Finnish and Swedish teachers’ perceived autonomy varied in somewhat different ways between the domains. Finnish teachers generally perceived themselves to be more individually autonomous, while Swedish teachers were more collegially oriented.


Introduction
There is no doubt that teacher autonomy is an important factor when explaining good practice in schooling. Teachers' perceived autonomy has shown to be important for teachers' commitment and work satisfaction, and extended teacher autonomy is linked to fewer disciplinary problems and reduced teacher turnover (Ingersoll, 2003;Skaalvik & Skaalvik, 2014). In two research reviews on the phenomenon, Wilches (2007) and Parker (2015) point out that a considerable part of the research in the field regards autonomy as one of the defining features of teacher professionalism. Some studies consider that teacher autonomy is important for attracting students to teacher education, while others even suggest that it increases school efficiency, since, they argue, autonomous teachers with influence over the educational decisions they are urged to implement are more likely to succeed in solving school problems. These are some of the reasons why teachers' autonomy has been the focus of much research during the last decades.
However, both Parker (2015) and Wilches (2007) emphasize the difficulties of measuring teachers' perceived autonomy, since teachers may have different perceptions of the same working conditions. Because of these difficulties, Wilches (2007) highlights the difference between subjective and objective levels of analysis, where a subjective analysis pays respect to teachers' perceived autonomy, which can be constrained by limitations in teachers' competence, job satisfaction, confidence and attitudes towards teaching and learning, while an objective analysis investigates external factors that might constrain teachers' autonomy, such as workload, lack of time and power structures within the school organization, but also external constraints from a political level, where central control by standardized testing and other accountability processes affect teachers' working conditions and inhibit teacher autonomy.
There are also reasons to believe that the issue of teachers' autonomy is complex in terms of how autonomous teachers ought to be. A certain amount of flexibility and autonomy regarding the selection of teaching methods and educational content is necessary, it has been argued, since teaching is complex and largely built on interaction, which is why we can never know exactly what comes out of teaching in terms of students' learning (Hopmann, 2007). On the other hand, while high levels of control and standardization in schools have negative effects on teacher autonomy, they may have positive effects in terms of creating more equivalence in school quality and for decreasing feelings of insecurity among teachers. A certain amount of bureaucracy and steering is also necessary to be able to organize education in a mass schooling system. Bureaucracy can thus be seen as a way to define the role of teachers and reduce the complexity of teachers' work (Ingersoll, 2003). This balance is and has been dealt with in different ways in different organizational, national and historical contexts, governing certain parts of teachers' work in various ways. In order to investigate this balance, we will view teacher autonomy from a holistic perspective by accounting for its multidimensional nature and its context dependency. This article suggests a conceptual and empirical way to do so.
We argue that Sweden and Finland are interesting cases in order to investigate the nature of teacher autonomy in the desired way: In Sweden, for example, the teacher profession and its autonomy has gained increased attention over the last decades, and has been at the centre of a polarized debate, where some claim that teachers are too autonomous and call for more central control of teachers and schools in order to ensure school quality, whereas others argue that the school reforms have led to decreased autonomy and a de-professionalization of teachers (Carlgren, 2009;Stenlås, 2011). In the neighbouring country of Finland, the story is quite different, perhaps due to their strong performances in the PISA tests (Carlgren, 2009). Finnish teachers' high status and their traditional teaching practices have been suggested as contributing factors to the PISA success, but Finnish teachers' high degree of autonomy, ensured by the absence of extensive control and surveillance systems, has also been highlighted in this respect (Simola, 2005).
In this article, Swedish and Finnish teachers' perceived autonomy is compared by presenting the results of a quantitative survey study, completed by 708 Swedish and 1583 Finnish teachers. Because teacher autonomy is seen here as context dependent, it is very important to take into account local and nation-specific structures when comparing teachers' autonomy (Wilches, 2007). The article will therefore start with a presentation of relevant characteristics of the Swedish and Finnish educational landscapes, together with a review of some of the research that has focused on Swedish and Finnish teachers' autonomy. Thereafter, the theoretical foundation of this article's understanding of the teacher autonomy concept is described and an analytical device is presented. This is followed by descriptions of the survey's construction and structure, samples and arguments for the methodological choices made. Finally, the results are presented, analyzed and discussed.

The National Contexts of Sweden and Finland
The Swedish and Finnish school systems share a history of strict regulation by a strong state, accentuating equality, rationality and citizenship, which underwent extensive changes when an intense decentralization phase took place in the 1990s, resulting in local autonomy and central management by goals and results (Telhaug, Mediås, & Aasen, 2006). Furthermore, both Finland and Sweden were influenced by the emerging globalization of education in the 2000s, with the OECD as a new powerful actor. However, the reforms did affect the educational landscape of the two countries in different ways. Unlike in Finland, the rapid decentralization and deregulation of the Swedish school system included reforms which gave parents the right to choose schools for their children. This came with a new system where families were provided with vouchers that financed the schools, and a heavy growth in the number of independent schools followed, which increased school competition. Moreover, the decentralization led to considerable freedom for individual teachers in Sweden to, for example, select educational content, which was soon exposed to criticism, resulting in a backlash with more extensive central control of school quality and results (Carlgren, 2009).
This was not the case in Finland, where schools and teachers as a collective were supposed to develop local curricula (Simola, Rinne, & Kivirauma, 2002), which perhaps regulated the work of individual teachers, who on the other hand have not been as closely evaluated and controlled as their Swedish colleagues. For example, there is no school inspection and no national testing in Finland (Simola, Rinne, Varjo, & Kauko, 2013). In comparison to the Finnish PISA success, the Swedish results were considered a disappointment, which might have been one reason for the intense evaluation and adaption to a globalized standardization of education (Grek, 2017). Regarding the Swedish and Finnish teaching professions, there are some important differences. Finnish teachers have strong unions with considerable influence on educational reforms (Simola, 1993). In Sweden, the unions were weakened during the 1990s due to the decentralization process and the accompanying new local teacher contracts and the introduction of individual salaries (Wermke, 2013). Also, it has been argued that Finnish teachers enjoy a higher status than Swedish teachers, partly because of a more highly considered teacher education, which meets high academic levels (Simola, 1993;Wermke, 2013).

Research on Teacher Autonomy in Sweden and Finland
Carlgren and Klette's (2008) study of the influence of educational reforms on Nordic teachers' working conditions found that curricular documents in both Finland and Sweden place an emphasis on teachers as professional curriculum makers, provided with a large amount of flexibility and responsibility for creating a good learning environment. Despite these similarities, there seemed to be major differences in practice. Teachers from both countries described the major changes of the 1990s as increasing their workload and demands for collegial work, but Swedish teachers also expressed uncertainty concerning how to handle their new situation with a newfound scope of action accompanied with increased external demands. Moreover, they perceived themselves as being increasingly responsible for their students' learning; a pressure the Finnish teachers did not seem to experience. Mølstad's (2015) study of how the central authorities in Norway and Finland exercise control over education, and to what extent the conditions for local curriculum work are promoted or prevented by curriculum design in the two countries, further strengthens the image of the Finnish teacher as trusted and autonomous. Mølstad describes the Norwegian approach as focused on delivery of the curriculum. Detailed guidelines were provided to ensure a correct understanding of the curriculum, and thereby support its implementation. In the Finnish case, teachers were granted more flexibility in their local curriculum work, which was seen as a pedagogical tool. Since Finnish teachers are given the mandate to develop the curriculum, Mølstad considers that they are offered a larger space for didactic reasoning.
In a study by Helgøy and Homme (2007), the autonomy of Norwegian and Swedish teachers was compared. They concluded that Norwegian teachers are characterized by old professionalism, with strong input regulations, which decreases individual teacher autonomy, but with teacher influence at a national level, where Norwegian teachers have been conservative towards school reforms such as individual salaries and standardized testing. Swedish teachers, Helgøy & Homme argue, have changed more towards a new professionalism that implies more individual teacher autonomy, which on the other hand might be restricted by local factors because of the more direct individual accountability that follows. According to Helgøy & Homme, Swedish teachers have lesser influence as a collective at a national level, but have adapted to a new form of collectivism at a work team level to share experiences and protect themselves against individual accountability. Lundström (2015) investigated Swedish teachers' autonomy, and argued that they are not trusted by the government as much as they had been before. Instead, he claims that their autonomy is now strongly regulated due to New Public Management structures, which have encouraged managerialism, standards and accountability. The respondents viewed themselves as autonomous in theory, but at the same time they claimed that their workload had increased due to external demands to implement new tasks without being provided with sufficient resources from municipal politicians. Many teachers also described a new role as school advertisers and an extended pressure to "satisfy the customer" due to school competition. Furthermore, the emphasis on test results and grades as definitions of quality, along with increased evaluations and the standards-based curriculum were pointed out as narrowing the view of knowledge and constraining creativity.
In Wermke and Forsberg's (2017) analysis of changes in Swedish teachers' autonomy over the last few decades they argue that the Swedish teaching profession lost a considerable part of its institutional autonomy and influence in connection with the decentralization wave of the 1990s. At the same time, the autonomy of individual teachers was extended. After the turn of the millennium, the increased control of school quality and results, together with other factors, has also led to a reduction in the autonomy of individual teachers. In comparison, authors points out that the Finnish teacher profession is more influential and enjoys more autonomy at an institutional level, but that a strong teaching profession also takes on the responsibility of regulating the autonomy of individual teachers. Autonomy is from that perspective always controlled in some way, and regulations can be viewed as an instrument for reducing the complexity of teachers' work.

An International Perspective
In order to further discuss education policy trends and its implications for teacher autonomy in Sweden and Finland, both countries will be related to education policy from an international perspective. In doing so, we want to point towards two fields of tension that might affect what is expected from teachers and thereby shape their autonomy in Continental and Northern Europe and in the Anglo-American realm: (i) how professionals relate to state and market and (ii) teachers' approaches to curriculum work.
i) The distinction between the Anglo-Saxon and Northern and Continental European traditions is important because the context in which a profession is autonomous relates to different understandings of what professions actually are, and to whom they are accountable: Svensson and Evetts (2008) distinguish between an Anglo-Saxon and a Continental approach to professionalization. In the latter, professions evolved in relation to nation state building and consolidation, having been entrusted with the above-mentioned risk handling tasks by the state. Consequently, their autonomy is granted by the state, professionals act responsibly towards their tasks in a given framework. They are accountable to the state that judges whether the profession acts according to defined expectations, and it also secures the professions' rights and status. Professions in an Anglo-Saxon tradition, in opposite, are not bound to a state authority in the same way. They are rather actors at a marketplace, offering their services (related to different life risks) to clients, to which they are accountable to (ibid.). Such professionals have more freedom of choice regarding their professional means. ii) Besides the abovementioned differences in understandings of professions, another very well empirically investigated contextual factor affecting teacher autonomy might be the Didaktik and/or curriculum tension (Gundem & Hopmann, 1998). It distinguishes between so-called European didactics traditions in guiding teachers' work, influenced by the philosophy of Friedrich Hegel. Key points include the critical importance of the teacher (not in the sense of teacher-centred instruction) and the subject in instruction. The teacher is responsible for elaborating the intrinsic value of a subject (Bildungsgehalt) for the education of pupils (Künzli, 1998;Westbury, 2000). Pedagogical work revolves around a theory of pedagogical action: the question of the mediation or mediator between theory and practice. In didactics, the autonomy of professional reasoning is crucial. Didactics is "not centred on the expectation of the school system, but on the expectations associated with the tasks of a teacher working within both the values represented by the concept of Bildung and the framework of a state mandated curriculum" (Westbury, 1998, p. 48). This approach is contrasted by the so-called curriculum approach that has its roots in Anglo-American education systems, and influenced by Ralph W. Tyler's rationale in educational planning and that defines the role of teachers and schooling rather differently. Westbury (1998, pp. 48-49) describes this approach in the following: "Curriculum is associated with the idea of building systems of public schools in which the work of teachers was explicitly directed by an authoritative agency which as part of a larger programme of a curriculum containing a statement of aims, prescribed content, (in the American case) textbooks, and methods of teaching which teachers are expected to implement." Both Sweden and Finland have traditionally been associated with the European tradition. Their similarities can be viewed in the light of the Nordic way of constructing the welfare state in the aftermath of World War II, which can be referred to as the Nordic model. The schools were heavily managed by the state and aimed to rebuild and modernize society in the name of science, rationality, equality and democratic participation (Imsen, Blossing, & Moos, 2017). As already touched upon, both Sweden and Finland carried out decentralization reforms in the 1990s that increased local autonomy and municipal responsibility for financial resources and the organization of schools, which could be seen as a step towards a more Anglo-American way of organizing the school systems (Carlgren, 2009;Simola et al., 2002;Simola, Rinne, Varjo, Kauko, & Pitkänen, 2009). However, it could be argued that Sweden have walked further on this path than Finland, who according to Imsen et al. (2017) has resisted much the global education reform movement and kept parts of the Nordic model of education, successfully combined with an academic culture. Sweden, on the other hand, has implemented a number of educational reforms after the turn of the millennium, with the purpose to strengthen state control and enhance school quality from above (Carlgren, 2009). This interpretation is strengthened while comparing with the Finnish context, where teachers have been depicted as conservative and lesser controlled, with higher societal status and more influence over school matters (Carlgren, 2009;Simola, 1993Simola, , 2005. These differences might be a reason to why Finnish teachers are often believed to be more autonomous than Swedish teachers.

Teacher Autonomy as a Multi-dimensional and Context-dependent Phenomenon
The view on autonomy in this article is built on the definition outlined by Ballou (1998) as self-governance; or "the capacity of an agent to determine their own actions through independent decisions within a system of principles and laws within which they operate" (p. 105). From this perspective, teacher autonomy is understood as something that is actively exercised, rather than passively received. This definition is closely related to Priestley, Robinson, and Biesta's (2015) understanding of the concept teacher agency, which also includes a view of the concept as involving the capacity of formulating possibilities for action and the exercise of choice.
Especially important for further developing the theoretical understanding of the teacher autonomy concept in this article is the work of Ingersoll (1996Ingersoll ( , 2003. Ingersoll offers a holistic perspective on teacher autonomy, but his work is not grand theory in itself. Rather, it is a collection of mid-range theories explaining the field of teacher autonomy. The American educational contextual background to Ingersoll's theory is built on principles of decentralization and accountability for teachers, and has traditionally differed considerably from the Nordic school systems and their approach to the professionalization of teachers. Today, the Nordic countries have moved toward a more similar system of product control, Sweden more so than Finland. However, Ingersoll's perspective is organizational rather than national, which makes his research less dependent on education policy. His research can thus be considered highly relevant to this study. We argue that Ingersolĺs organizational focus contributes with a new perspective on teacher autonomy, since dichotomies entail a risk that the researcher gets stuck in preconceptions while analysing the cases in focus. 3.1. Teacher Autonomy as Decision-making Ingersoll (1996Ingersoll ( , 2003 is interested in power distribution and conflict within the school organization, and his research revolves around the amount of power and autonomy given to teachers and its relation to school functioning. Ingersoll considers that power within an organization belongs to those who make the most important decisions. Hence, teacher autonomy is viewed by Ingersoll as teachers' influence over the key-decisions that affect the character, content, processes and evaluation of their daily work. A fruitful way of investigating teachers' autonomy is thus, according to Ingersoll, to ask them who holds the power to decide about important issues in a school. This is a central key for operationalizing the concept in this study, and entails a slight difference from the teacher agency concept as put forward by Priestley, et. al., since decision-making relates to the question what, while agency can be considered to deal more with issues regarding the how. When investigating teachers' perceived autonomy empirically, one cannot expect valuable results from simply asking teachers how autonomous they are, since the same amount of autonomy might be experienced differently by different teachers. The concept needs to be clearly defined and it must be possible to measure and compare between teachers. By operationalizing teacher autonomy as decision-making, the complexity of the phenomenon is reduced and it becomes easier to create relevant variables that can be used to measure and compare teachers' autonomy. Ingersoll (1996) claims that more decision-making power for teachers has varied effects on their autonomy depending on which decisions are to be made. That is why it is especially important to separate important and unimportant decisions in order to gain an appropriate understanding of the meaning of power distribution for teachers' autonomy. Less important decisions might be delegated, leading employees to believe they have real power, while key-decisions remain firmly controlled at a higher level within the organization. Ingersoll (2003) highlights administrative, instructional and social issues as the three main areas of teachers' work where important decisions are made. Ingersoll (2003) claims that teacher autonomy is mostly exercised through instructional decisions within the classroom, such as selecting educational concepts and instructional methods. Choices of textbooks, the establishment of a local school curriculum and other decisions at a schoolwide level, though, are issues where teachers have little input. Administrative decisions about teachers' schedules, class sizes, student tracking and resource allocation are, according to Ingersoll (2003), almost exclusively made on a managerial level, while teachers' power over decision-making regarding social issues is more varied. Decisions regarding student discipline within the classroom has traditionally been in the hands of the teachers, but teachers' authority to exclude students from the classroom, and their influence over behavioural rules is not as evident. It is agreed by many researchers that the complex and diverse work of teachers implies that teachers may be more autonomous in some areas of their work than in others, and other scholars have also conceptualized teacher autonomy as exercised in different areas or domains. Wilches (2007) suggests four domains: The teaching and assessment domain includes teachers' control over decisions regarding choice of educational goals, content, methods and materials as well as social issues, such as the procedures around student behaviour. In the curriculum development domain, teachers' implementations, interpretations and re-phrasings of the curriculum are investigated. School functioning is the third domain. It comprises teachers' decision-making in administrative tasks, such as budget planning, timetabling and class-composition. The fourth and final domain concerns teachers' professional development, and deals with teachers' opportunities to engage in professional development, but also their opportunities to decide for themselves its educational content and when and where it takes place.

Teacher Autonomy at Different Levels
According to Ingersoll (2003) most theory on school organization considers schools as consisting of two separate zones. The schoolwide zone comprises mainly administrative activities, such as management, planning and resource allocation, while the classroom zone includes teaching and other educational activities. This kind of view of teacher autonomy as exercised at different levels; by individual teachers in the classroom, and by groups of teachers at local or even national level, can also be found in the work of other researchers in the field as shown below. Such an understanding assumes that teachers may enjoy a higher degree of autonomy at one level at the expense of reduced autonomy at another.
For example, Wermke and Höstfält (2014) draw in their reasoning on Hoyle's (1975) distinction of teacher professionalism meaning what defines the teaching profession as a whole and teacher professionality being that what teachers do in their professional day-to-day work. Authors argue that the institutional autonomy of the teacher profession is extended in a school system where the government exercise its control by standardizing schooling and educating a competent teaching profession, responsible for the evaluation of pupils with respect to curricular goals. The status and authority of the profession is, however, based on the constraining of individual teachers' service autonomy by peer-control, grounded in a professional code of ethics and the use of didactics as a standardizing tool for instruction. When the government instead evaluates the outcomes of schooling, the institutional autonomy is constrained, since other agents are in power over decisions regarding the evaluation of pupils and what is adequate knowledge. At the same time, the service autonomy of the individual teacher is extended, since neither the state nor the teacher's colleagues restrict the choice of educational content and teaching methods. Frostenson (2015) provides an alternative understanding of Wermke and Höstfält (2014) conception by dividing the service autonomy component into two separate levels, which describe teacher autonomy as exercised by individual teachers on one hand, and by groups of teacher colleagues, on the other. In Frostenson's conceptualization, general professional autonomy deals with the authority of the teacher profession in issues around organizing the framing of teachers' work, such as teacher education, entry requirements, curricula and the organization of the school system. At the level of collegial professional autonomy, teachers' influence on organizational and pedagogical issues within the local context of teachers' work are analysed. Finally, individual autonomy refers to individual teachers' decision-making in educational matters within the classroom context.

An Analytical Device
An analytical matrix has been constructed in order to be able to measure, analyse and compare teachers' perceived autonomy. The matrix is based on conceptualizations of teacher autonomy highlighting its multidimensionality; mainly the work of Frostenson (2015), Wermke and Höstfält (2014), Wilches (2007) and Ingersoll (2003). All these scholars underline that autonomy can be exercised at different levels and in different domains. The model presented in the analytical matrix is a summary of these scholars' suggestions of such domains and levels and has been discussed and validated in expert workshops as described below.
The matrix consists of a vertical dimension with four domains. Planning, instruction and assessment are some of the duties regarding teaching and learning that are included in the educational domain. The second domain is social and contains the socialization of students. A few examples of teachers' work that are located in this domain are disciplining of students, student grouping and the treatment of students with special needs. In the developmental domain, teachers' formal professional development is emphasised. Central issues in this domain are teachers' influence and involvement in these matters. Finally, the administrative domain refers to teachers' opportunities to make decisions regarding the use of resources, timetabling, teachers' salaries and other administrative activities (Table 1).
The classroom-related dimension relates to the individual teacher's scope of action within a classroom context. In the second, school-related dimension, the autonomy of teachers as a collective is highlighted in a local context. At this level, the school is viewed as a social arena with many actors who affect the autonomy of one another. In particular, Frostenson (2015) and Wermke and Höstfält (2014) suggest even a professional level that describe the relation of the teaching profession to other stakeholders in the school system and society. In this paper, we focus on teachers' perceptions on decision making and control, drawing on Ingersoll's organizational theorizing on autonomy and control in schools. That is why only the classroom level and the school level are put in the centre of attention. We believe that regarding the autonomy at these two levels, our respondents will be able to give valid answers.
This analytical grid presents an attempt at complexity reduction in terms of the empirical reality of the subjects of study: teachers in different countries. In relation to the findings of former conceptualizing of teacher autonomy, we are interested in a holistic perspective on teacher autonomy, where each dimension can be understood in relation to each other, which of course is only one of many possible perspectives for investigating teacher autonomy.

Designing the Survey
The study at hand is part of a project comparing teacher autonomy in Sweden, Finland, Germany and Ireland. The process of designing the questionnaire started with a literature study, where empirical instruments measuring teacher autonomy were explored and analysed in terms of their utility and suitability for the project. The literature review resulted in an analytical grid, which was tested and further developed in a qualitative comparative study before undertaking the survey (during 2016). Roughly 100 teachers in four countries were interviewed individually or in groups, and approximately two weeks were spent in schools in four countries observing the organization of schooling.
In relation to and after the literature review, the research project was accompanied by an expert group with two to five teachers and teacher educators per country. The group met, in differing constellations at four expert workshops, each two days long. This study followed the recommendations of Broadfoot, Osborn, Gilly, and Bucher (1993) on the conduct of cross-national studies, i.e., linguistic and concept equivalence above all. Such cross-national expert group meetings helped to address such requirements. In particular, the meeting of experts from the participating countries enabled ensuring equivalence in the operationalization. In order to sensitize the experts to cross-cultural comparisons, school visits as well as filmed instructional lessons and discussions of dilemmas (in vignettes) were included. All discussions were filmed or recorded and listened to again.
The meetings were organized in the following order: The experts first discussed (at workshop one, May 2016) the conditions of the teaching profession in relation to which are the most important decisions to be made for the teaching profession; who makes them, and how the decisions made are controlled (Ingersoll, 2003). At workshop two (March 2017), the participants discussed in detail the operationalization of the research questions in a questionnaire. In the first development step, all questions were formulated in English. After this workshop, the translated questionnaire versions were sent for testing and commenting to the experts involved. All questionnaires were distributed in the countries' languages (in Finland: in Swedish and Finnish). In workshop three (November 2017) and in workshop four (March 2018), the questionnaire's descriptive results and scales were discussed in order to test their plausibility, as well as issues that occurred during the data collection process. In the online questionnaire, teachers had the opportunity to comment on the questionnaire. These comments were also used to find problematic issues in the questionnaire ex post. All these steps aimed to make the questionnaire valid and reliable, and ensure comparability between countries.

The Questionnaire
The main part of the questionnaire, and the focus of this article, is the teachers' perceptions of who makes important decisions that affect their work, and this follows a holistic approach. Consequently, we are first and foremost interested in how teachers decision-making in the most important aspects of their work, relate to each other in different national contexts. Literature on the subject, the teacher interviews and input from expert groups formed a knowledge base for choosing which decisions should be considered important and appropriate for the questionnaire. The decisions were then divided into the four domains comprising educational and social matters as well as issues regarding teachers' continuous professional development and administrative work. Some of the decisions fell into the classroom level category and some were classified as school level decisions. Also this classification was built on the previously described pre-study and interviews with experts in the field, with the aim to make transnational comparisons possible. The dividing of decisions into different levels could have been made in a different way, depending on perspective. In other words, we do not claim any exclusiveness in regard of the distinctions made, and invite further discussion here. The matrix is in that sense interesting for discussing which decisions are actually made on which level. However, related to our expert workshops, we use the following most important decisions as categorized in Table 2.

Data Organization and Aggregation
This study approaches the most important decisions by asking teachers which actors might be involved when decisions are made in schools in the particular national context. The respondents had the possibility to choose two alternatives, and for each decision the respondents could choose between: 1. Individual teachers 2. Teachers with colleagues 3. Teachers with others (such as parents council or school management) 4. School management 5. Local actors such as municipalities or school owners 6. Central government 7. Do not know/Do not want to answer The high number of decision-making variables in the study at hand is of nominal variable nature. Each decision consists of seven variables, leading to each dimension (level * domain) comprising between 30 and 40 variables. The nominal nature of the variables excludes factor analysis as a complexity reduction strategy. There are today statistical strategies that can investigate latent relations between categorical variables. Such latent class analyses are very complex and include several interpretative steps that actually become more complex with each variable added to the model. The high number of nominal variables in this study, given by its methodological design (in total over 200), would challenge such statistical analyses too much. Since we are interested in a holistic perspective on teacher autonomy, such approaches cannot be seen as practical here.
In order to reduce the complexity of the material, the seven alternatives were instead first summarized into four categories: individual teachers (alternative 1), teachers collectively (alternatives 2 and 3), school management (alternative 4) and actors outside school (alternatives 5 and 6). If one of these alternatives was chosen by a respondent, it was assigned a point. The points were then summarized for each domain and level, and the respondents were sorted into different decision-making types that indicate which actors they perceive to make the most decisions within the different domains and levels. If a respondent, for example, for the most part stated that individual teachers made the decisions within a specific domain at a specific level, the respondent was considered individually oriented with respect to that dimension. In the same way, respondents who mainly perceived the decisions to be made by teachers together with colleagues, by the school administration or by the state or municipality, were described as peer oriented, principal oriented or policy/administration oriented.
The described arithmetical means of summarizing the items does have methodological consequences, even if we argue for the validity of the scales due to their theoretical embedding in the study's analytical device. One consequence is that we assume all of the most important decisions to be made per dimension as equally important in relation to the decision-making and control perspective employed. The types are strictly quantitative and it can only be understood that a particular stakeholder has the most or least, less or more decision-making power. That means, they do not, according to factors, explain relations between the individual decisions. Due to our theoretical vantage point manifested in the teacher autonomy device, we can, however, relate the eight different dimensions to each other, and by doing so take a holistic approach to teacher autonomy. In relation to Ingersoll (2003), this means that we only look at the most important decisions, but follow his idea of less or more importance while comparing the results of the different dimensions.
Moreover, we decided to pursue a conservative method of handling missing values. Besides deciding not to answer questions, teachers could choose to answer almost all questions with "not relevant/ I do not know". This is important for the survey's validity. The teachers answer only when they feel confident to do so, in relation to their contextual situation (by country, school form, school etc.). In other analyses, a higher or lower number of teachers answering I don't know can also be analysed in terms of how much a teaching profession is aware about who decides on the most important aspects of schooling. For clarification, such decisions have evolved from extensive work with literature and teachers in different national contexts. However, such analyses are not the focus of this article. Until then, it has been decided to implement a categorical deletion of missing values. In this method, an entire record is excluded from analysis if any single value is missing. This produces a greater number of missing values in our scales. Due to the large overall size of the sample and in favour of validity, this strategy was chosen, even at the price of psychometric arguments such as the power and appropriateness of our scales.

Sample
The examining of teachers' autonomy is a complex process, where many different factors need to be regarded as possibly affecting the results. Thus, the need for a large sample is high; a precondition with impact on the choices made in the process of data collection. In order to reach many teachers, the data was collected by contacting municipalities and teacher unions, who distributed the questionnaires to their teachers to be filled in between September and November 2017. The Swedish sample was smaller, with 708 teachers participating in the study in comparison to the Finnish sample of 1583. This was due to a higher response rate among the Finnish teachers. The reasons for this is unclear, but one possible interpretation is that the subject of the questionnaire seemed more attractive to the Finnish teachers. Possible reasons for this might be feelings of pride and shame in the sense that Finnish teachers perceive themselves more autonomous than Swedish teachers. The gender distribution was roughly the same within the two samples, but regarding the issue of school grade, Sweden differed from Finland with a larger number of primary school teachers. Another difference between the two countries was that more Swedish teachers categorized their school location as urban, which might also explain why the average school size was larger in the Swedish sample. Furthermore, the Finnish teachers seemed to be a little more experienced than the Swedish teachers and working full time to a larger extent. When comparing the sample to the teacher populations in the two countries, it showed a larger share of women, and in the Swedish case also a smaller share of secondary teachers. However, neither the gender nor the grade variable seem to have any considerable effect on the results that could be identified at this stage of the study. The other differences between the characteristics of the Swedish and Finnish samples, such as school size or location did not seem to noticeably affect the results.
The variables categorizing teachers as different decision-making types and control types are nominal, which is why Chi-square tests were employed for testing the significance of eventual differences between Swedish and Finnish teachers in these cases. We report only the significance value p. Values of significance (i.e., below p = 0.05) in the results would then imply significance if the selection had been an independent random selection, which it is not as described earlier. By performing significance tests and describing the impact on the results by the background variables differing the most between Swedish and Finnish teachers and between the sample and the population, an analytical discussion regarding estimations of generalization will be possible.

Results
Similarities and differences between Swedish and Finnish teachers' perceived autonomy were explored through cross tabulations of the decision-making types, which show the share of teachers who perceived that the decisions within each domain and level were mostly made by individual teachers, teachers with colleagues, the school management or the municipality/state. The total number of cross tabulations was eight (four domains and two levels). These will first be presented, after which the overall results will be highlighted by use of the analytical matrix, which displays the one or two actors most teachers perceived to be the main decision-makers in each dimension.
The similarities between the countries in the educational domain were more prominent than the differences. Most Swedish and Finnish teachers perceived their autonomy to be high at both levels. Generally, few teachers in either context perceived the school management, politicians or officials at local or national level as having much decisional power over these issues; a finding in line with previous research in the field (e.g., Ingersoll, 2003  The most obvious national difference at the classroom level (Figure 1) was that a larger proportion of Finnish teachers perceived the decisions to be mostly made by individual teachers, whereas the Swedish teachers were more collegially oriented. The school level (Figure 2) proved to be a field for collegial decisions to be made by teachers, according to most teachers' perceptions. Ingersoll's (2003) research has found that these kinds of decisions have tended to be mainly controlled by actors other than teachers. However, this discrepancy might depend on the different national contexts in the Nordic and the US school systems, in which Ingersoll's research took place. An interesting relationship for analysis is the one between individual teachers' autonomy and collegial autonomy. It certainly seems that the teacher faculty is powerful regarding educational issues in both Swedish and Finnish schools, although Finnish teachers perceive themselves as more individually autonomous in educational classroom issues.
One interpretation of this difference is that the Finnish teachers' role might be more defined and regulated in advance by an extensive education with practice schools, which strengthens the uniformity of the profession and reduces the need for large-scale cooperation among colleagues in decision-making (Wermke, 2013;Simola, 1998). Another reason may be that it is a part of the Swedish tradition to make important decisions in consensus (Wermke, 2013), which is why the results could be understood as a consequence of a Swedish kind of mentality, where meetings are used to ensure that decisions are made in a cooperative way.
Compared to the educational domain, more teachers perceived the school management to wield power over decisions in the social domain. The most obvious school management power seemed to be exercised at the classroom level (Figure 3). Regarding the comparison between the two countries, the similarities once again overshadowed the differences. Teacher involvement in social decisions was primarily exercised by teachers as a collective, especially at the school level (Figure 4). The question that could be raised here concerns the level at which teachers prefer to have influence over the social decisions that are to be made; as individuals in the classroom or as a collective at school? It could be argued that social work in the classroom is most closely connected to teachers' everyday work and hard to separate from instructional work, since it is all intertwined.

Individual teachers Teachers with colleagues
School management Municipality/owner or state From that perspective, the individual autonomy of teachers at the classroom level would be more important for teachers' well-being, and perhaps also for effective instruction. According to Ingersoll (2003), social issues are the ones where teachers' influence entails the most advantages for school functioning. Moreover, he argues, this domain is also the one where teachers' dependence upon the school management is highest, since teachers typically are responsible for keeping order in  the classroom but do not have the power to sanction students. A high degree of power by the school management in these issues could then be viewed as a kind of subtle control mechanism. On the other hand, decision-making also entails responsibility and accountability, and it could be argued that social decisions are accompanied with risks and potential conflicts with students and parents. Another way of viewing this power distribution would thus be that if the school management takes on this responsibility, they absorb the risks for teachers, who can focus more on educational decisions.
In the developmental domain, differences between the two countries were highly significant at the classroom level ( Figure 5). Among teachers in the Swedish sample, the school management was perceived by most of the teachers to be the actor most often deciding about these issues, while individual teachers had very little influence. In Finland, the distribution was more even, although the largest share of teachers in this case also attributed the most decision-making power to the school management. However, it should be noted that the principal in Finnish schools is a part of the pedagogical profession and is defined as a head teacher. This is not always the case in Sweden, where a teaching background is not required. Hence, decisions made by the school management regarding teachers' CPD are indeed in a sense made by teachers in the Finnish case; an important addition to the analysis.
At the school level (Figure 6), the differences between Swedish and Finnish teachers were not as distinct as at the classroom level. The perceived teacher influence diminished, and instead, actors outside school, but most of all, the school management, were perceived as the most powerful stakeholders by the teachers. According to Ingersoll (2003), however, control and power can be less visible, and they do not have to comprise formal regulations and standardized work tasks. Teachers might therefore not perceive how the hierarchy of the school system is constructed, and that decisions about teachers' CPD might originate from a much higher level than the results show.
It certainly seems, though, that Finnish teachers have considerably more influence over their own continuing professional development, which is very interesting when discussing teacher professionalism in the two countries. It is possible that Finnish teachers' comparatively higher status (Simola, 2005) and the societal appreciation of their knowledge (Wermke, 2013)   they are more trusted to judge which CPD is the most appropriate. The alleged de-professionalization of Swedish teachers put forward by many scholars (Lundström, 2015;Stenlås, 2011), might then be taking place in this domain, rather than in the educational one, since an occupation that is not in charge of its own CPD could hardly be classified as a profession in the traditional sense.  At the classroom level (Figure 7), teachers from both Sweden and Finland perceived their own influence on administrative issues as very limited, whereas the school management was considered by the vast majority of teachers in both Sweden and Finland as the most powerful actor in this area. This was no surprise, since researchers in the field have found that administrative decisions are the ones where teachers traditionally have the least input (e.g., Ingersoll, 2003). At the school level (Figure 8), Swedish teachers differed significantly from their Finnish colleagues due to a much higher level of perceived collegial autonomy. According to many Swedish teachers, the teaching faculty had a considerable amount of influence in these matters.
The greatest part of the Finnish teachers, on the other hand, perceived the school management and actors higher up in the school hierarchy to be in more exclusive command over the decisions, whereas teacher involvement was regarded to be more modest. The school management was still the actor most teachers in both countries considered to be in control over the most decisions within the domain, though, and individual teachers rarely had any influence on these matters. However, this domain turned out to be the only one where the degree of teacher involvement was perceived as higher among the Swedish teachers. Once again, an important question to ask is whether this kind of decision-making is desirable from a teacher perspective? According to Ingersoll (2003), unimportant decisions are often delegated in order to create a false sense of autonomy, while the really important decisions are made without much teacher input. Since we asked after the most important decisions regarding the various dimensions, we might interpret here that different dimensions are more or less important for teachers. Consequently, autonomy in certain domains can produce a false sense of autonomy (Ingersoll, 2003) or overstretch the capacities of teachers (Wills & Sandholtz, 2009).
There are some nation-specific differences between the contexts of Swedish and Finnish teachers' work that might be relevant here. In the wake of the decentralization of the school system in the 1990s, new local regulations and contracts for teachers involved new working conditions for Swedish teachers, such as the introduction of individual salaries and office hours where teachers' physical presence at school, even outside of class-time, became scheduled. This was not the case in Finland. Such differences might affect the volume of administrative working tasks. It is possible that the situation where a teacher suddenly has to bargain for pay rises would entail a new relationship with the school management, where the need to prove value might lead to the teacher taking on more administrative duties. It is also possible that the required time spent at work makes it easier to assign administrative work tasks to teachers that cannot be done at home, which also would imply that Finnish teachers have more time for making individual educational decisions at home.

Discussion
Tables 3 and 4 summarize the overall results of the Swedish and Finnish teachers' perceptions of their autonomy in terms of decision-making in the four domains and at the two levels.
Among the findings, there are some that can be considered more relevant than others, and that, for example, display interesting similarities or differences between Swedish and Finnish teachers' perceived autonomy, or between the teachers' perceived autonomy in different domains of their work: . Teachers from both Sweden and Finland perceive themselves as autonomous when it comes to the most central area of their work: teaching. Overall, the similarities between Swedish and Finnish teachers were more striking than the differences. . Generally, more Finnish teachers perceived themselves to be individually autonomous, whereas Swedish teachers experienced more collegial autonomy. . The results also indicate that there are large differences in teachers' perceived autonomy between the different domains of their work, with the highest degree of autonomy experienced in the educational domain and the lowest in the administrational domain. Moreover, Finnish and Swedish teachers' perceived autonomy varied in somewhat different ways between the domains, with Finnish teachers seeming to have more influence over their own continuous professional development than Swedish teachers, who instead seem to make more administrative school decisions than Finnish teachers.

Important and Unimportant Decisions: a Nuanced View on Teachers' Autonomy
At this high level of abstraction, differences between Swedish and Finnish teachers' perceived autonomy do not seem to be as great as indicated by much of the previous research in the field, claiming that Swedish teachers have become de-professionalized and lost substantial parts of their autonomy due to the educational reforms which have taken place since the 1990s and the introduction of New public management structures (e.g., Lundström, 2015;Stenlås, 2011). Instead, the results of this study suggest that Swedish teachers' autonomy might not have been so reduced after all, at least not from a who makes the most important decisions perspective. For example, both Swedish and Finnish teachers perceived themselves to be autonomous regarding the core of their work; educational, and to some part social issues. This finding also suggests that the more curriculum-oriented education policy trends in Sweden in comparison to Finland does not necessarily have to mean a lower degree of teacher autonomy. Moreover, the results indicate that there is no simple answer to the question of how autonomous teachers are. Rather, it is a question of which aspects and areas of teachers' work they decide about for themselves. Ingersoll (2003) suggested that it is important to separate important and unimportant decisions. In this study, only decisions considered as important were included. By using the analytical device, however, it became apparent that the perceived autonomy varied greatly between the different domains, and perhaps the domains should even be ranked by their importance to teachers' autonomy.
In the case of this study, the Finnish teachers perceived more individual autonomy than Swedish teachers regarding educational and CPD issues. At the same time, they experienced a very limited autonomy regarding administrative issues, whereas more Swedish teachers considered the teacher faculty to be responsible for decision-making in this area. In the analysis section, Swedish educational reforms and organizational re-structuring of teachers' salaries and working hours were suggested as possible reasons for this. From Ingersoll's point of view, the Finnish alternative might be preferable for teachers, since the opportunity to leave those decisions to other actors would reduce the complexity of the teachers' work and free up time and energy that can be put into more important tasks.
As concluded by Ingersoll, the complexity of schooling means a delicate balance of control and consent in order to organize the school system and ensure equivalence in education without regulating and controlling teachers' work too much. Teachers need autonomy to be able to handle the  complex and interactive work of teaching, which is impossible to organize in detail (see also Hopmann, 2007), but there can also be too much. This is why Ingersoll (2003) emphasizes the significance of sorting out which decisions it is most important for teachers to make. Decisions that are important for the school organization do not have to be important for teachers. It could very well be that the decisions teachers prioritize are the ones most closely related to teachers' main area of work, i.e., educational, social and CPD issues at the classroom level. This would be interesting to explore further in future research. It would also be interesting to investigate teachers' influence on the distribution of responsibilities in the school organization. Wilches (2007) argues that undesired responsibilities are not autonomy in its proper meaning. Instead, he considers that autonomy is about teachers' rights to make professional decisions within certain frames. From such a viewpoint, it could be argued that autonomous teachers need to have a say in how responsibilities are distributed.

Collegial Autonomy: Sharing Risks at the Price of Control
Another dimension can be added to the discussion by focusing on the difference between Swedish and Finnish teachers' autonomy in terms of whether it is mostly individual or collegial. We would even argue that a discussion about the meaning of the different levels of autonomy is of crucial importance for understanding the results. Finnish teachers were more individually autonomous, while Swedish teachers perceived collegial autonomy to greater extent; a difference that could be seen in, for example, educational issues. What does collegial autonomy then mean compared to individual autonomy? As already mentioned, collegial autonomy could be a sign of cooperation among teachers, where they share responsibilities with each other, which could function as a form of risk absorption and break the isolation that according to Ingersoll (2003) creates feelings of insecurity among teachers. This kind of local collectivism has already been put forward by Helgøy and Homme (2007) as a common feature among Swedish teachers. On the other hand, collegial decision-making could also be interpreted as a kind of (subtle) control of teachers' work. By using Michel Foucault's concept of governmentality, Ball, Maguire, Braun, Hoskins, and Perryman (2012) argue that control mechanisms used by the central government are not only meant to implement changes in the practice of schooling, but also to implement the underlying thoughts and opinions. According to Ball et al, the importance of effectiveness and high and measurable results in a school system with output control is therefore implemented among the teachers, and results in "policy carriers" within the teacher profession, who in turn implement the curriculum among their colleagues. From this perspective, it could be argued that individual autonomy is more "real" than collegial autonomy. This perspective can serve as an alternative interpretation of collegial autonomy; that the regulation of individual teachers' autonomy by their colleagues could also originate from central governance rather than from a strong teacher profession. An analysis rather similar to the one made by Ball et al. is Lawson's (2004). Also by referring to Foucault, Lawson illuminates the teacher empowerment concept, which has often been confused with teacher autonomy. Teacher empowerment have, according to Lawson, in some sense come to mean the handing over of decision-making and responsibilities from the government to teachers. However, Lawson argues that empowerment discourse leads to a system of self-governance by individuals, which from a Foucauldian perspective can be seen as a system of discipline. An interesting remark here is that such a system would be equally present for both Swedish and Finnish teachers. Maybe even more so in the Finnish context, following our previous descriptions of the uniform and conservative Finnish teacher.

Conclusions
Teacher autonomy is a complex phenomenon. As indicated in this study and in much conceptual research, it is context dependent, multidimensional and cannot be measured easily. Sometimes the teachers' autonomy seemed to be a kind of compromise or a trade-off, where a high degree of autonomy at one level or in one domain could work as a constraining factor for other levels and domains. At least partly, this is because there is only so much time and energy for making decisions. The Swedish teachers in this study considered decisions to be made collegially to a higher extent than the Finnish teachers did, among whom individual teacher decisions were more frequently perceived. The cooperative orientation among Swedish teachers may protect teachers against risks, but could it also be interpreted as a sign of a profession that dares not be held individually accountable to its clients? Perhaps, the increasingly intense governmental control of schools and teachers in Sweden has raised levels of uncertainty among Swedish teachers to the degree that they choose not to make individual decisions even though they are able to do so. Of course, also other factors might lay behind teacherś choices to make decisions together. For example, teacher collaboration has shown to be beneficial for teacherś motivation and for teaching efficiency (Vangrieken, Dochy, Raes, & Kyndt, 2015), and it might just be that Swedish teachers tend to prefer cooperative work for such reasons. However, also the decisions a teacher perceives to be making individually are affected by many factors, such as the teacheŕs education, institutional norms and professional ethics. The decisions are most likely also affected by the viewpoints of other actors with an interest in the matter. From such a perspective, Finnish teacherś autonomy might already be regulated by their strong teaching profession, which makes them more certain in how to make the "correct" decisions individually.
The results of this study showed that Swedish and Finnish teachers' autonomy was rather similar in the educational and social domains. Swedish teachers' autonomy might therefore not be as limited as much research has indicated. Signs of the de-professionalized Swedish teacher could perhaps instead be more easily found in issues of teachers' continuous professional development, or in the difference between individual and collegial autonomy. Carlgren (2009) points out that Finland is the only Nordic country where students are easily attracted to teacher education and highlights Finnish teachers' autonomy and high status as possible explanations. If the orientation toward individual rather than collegial autonomy and the perceived greater influence over CPD issues among the Finnish teachers in this study are examples of factors that increases teachers' status as educational experts and that makes the teacher profession more attractive, one could ask whether the Swedish teacher profession would benefit from a more "Finnish" means of power distribution in the Swedish school system. And if that is the case, who are the actors that need to initiate change? It is hardly possible for an occupation to passively receive professionalism. More likely, the teacher profession would, for example, need to strive toward playing a greater role in the organization of teacher education and teachers' CPD in order to gain credibility.

Limitations
The large sample is a major factor in making our results relevant and interesting for discussing teacher autonomy in the two national contexts. In order to achieve a satisfying sample size, a practical way of collecting data was chosen. Since the sample was not randomized, and some differences were present in the characteristics of the Finnish and Swedish samples of teachers, we needed to be careful when discussing generalization of the results. For example, the Swedish sample consisted of a higher share of primary school teachers, and more teachers working in larger urban schools, compared to the Finnish sample. It could be argued that secondary teachers in smaller schools have few opportunities for collaboration with colleagues regarding their classroom practice and their subject related CPD, which could be an alternative explanation to why the Swedish teachers seemed to be more collegially oriented in their decision-making than the Finnish teachers. One by one, these variables did not have any considerable effect on the results, and "country" was clearly the most distinguishing variable. However, we must consider that all these other variables together may have affected the results to some extent, and perhaps made differences appear somewhat greater than they really are.
We are also well aware that the high level of abstraction in this article is showing reality somewhat fuzzy or reshaped. On the other hand it also entails possibilities to analyse and discuss the results in ways that would otherwise have not been as easily done. Moreover, one limitation with both our operationalization and our analytical device is that our findings and conclusions have to be seen in light of our theoretical perspective, and that we cannot claim that our results take all possible aspects of teacher autonomy into account. Other research designs may have better prerequisites for reaching a deeper understanding of the concept. Complementary qualitative research on Swedish and Finnish teachers' autonomy is therefore welcome.