Making teachers by policy? The case of teacher evaluation in Norwegian education

ABSTRACT The reported study investigates the emergence of teacher evaluation (TE) in Norwegian green papers between 1988 and 2019. We examine how knowledge dissemination and discourses in Norwegian official reports shape suggestions for TE and discuss implications for how teachers are “made by policy”. Guided by Bacchi’s (2009) approach to critical discourse analysis, “What’s the problem represented to be?” (WPR), we explore where problem representations and their underpinning assumptions are produced and disseminated, and how they invoke new arguments and discursive and subjectification effects. The study contributes new knowledge on the origins of TE as a policy tool for governing Norwegian education in general, and teachers more specifically. Across time and irrespective of government coalitions, we identify the OECD as a main contributor to new problem representations and display how proposals for TE policies based on performance-based accountability (PBA) promote new discourses on teacher professionalism. Implications for research, policy and practice are discussed.


Introduction
Over the last three decades, the neoliberal emphasis on markets, managerialism and performance has influenced national efforts to improve education quality across the Western world (Ball, 2003).Along with new technologies, innovative use of data and increased public transparency, performance-based accountability (PBA) has become a feature of education reforms that seek to measure public sector workers' performance and make them accountable for results (Grek et al., 2020;Jaafar & Earl, 2008).Across different accountability regimes, complex models have developed to measure teaching and schooling outcomes, using statistics and graphs to portray the commitment to transparency and accountability as a "science-based knowledge production" (Lindblad et al., 2018;Porter, 2020).Since the 1990s, teacher evaluation (TE) has been promoted as one means of making what happens in the classroom more transparent and open to public scrutiny (Holloway & Brass, 2018).However, scholars have argued that TE takes shape as a ritual of power and a "ceremony of visibility", in which teachers are politically reconstructed as ethical subjects governed by logics of quality control and performance indicators, consequently affecting the core purposes of teaching (cf.Ball, 1990).
Previous studies have illuminated the various conceptions of PBA informing education system blueprints across Europe (Grek et al., 2020;Gunter et al., 2016;Maroy & Pons, 2019;Verger & Curran, 2014).Recent literature highlights the political force and practice-transforming capacity of PBA policies in Norway (Camphuijsen et al., 2020) and the implications for Norwegian school leaders and teachers (Larsen, 2021;Mausethagen, 2013;Mausethagen & Mølstad, 2015).However, there is a need for further research in this area, not least to expose the empirical manifestations and concrete "products" of PBA policy trajectories in contemporary education.
The present article situates TE motives and rationales in a commitment to control and audit in Norway and beyond.In particular, we analyse and discuss the introduction and retention of TE as a distinctive feature of Norwegian education governing in terms of the problems it is intended to solve.To that end, the study investigates formal decision-making processes and the policy documents produced by ad hoc commissions of inquiry (NOU series) 1 for the Ministry of Education and Research between 1988 and 2019.Informed by the work of Carol Bacchi (2009) and more generally by Foucauldian "governmentality studies" (see for example Ball, 2003;Rose, 1996), we contend that, in any domain of reality, governing entails the construction of particular problems in need of guidance and control.We argue that the increasing role of PBA in education governing and the different ways in which "the problem" of teachers' work is constituted are important issues precisely because governing is enacted and realized through such problematizations.From this perspective, political problems are not external to governing practices but are produced through them.
We were especially interested here in how the members of commissions of inquiry (re)construct and (re)shape problem representations ranging from comprehensive educational assessment frameworks to proposed tools for TE.By identifying these representations in a range of documents, we were able to capture the potential subjectification effects of emerging TE policies.By exploring changes in these problem representations over time, we were also able to describe the ongoing process of the prevailing control policy "as becoming" (cf.Ball, 2003).
The rest of the article is structured as follows.First, we present a brief overview of key studies, identifying research perspectives on PBA and studies that specifically investigate TE.We then outline Bacchi's (2009) theoretical and methodological perspective, which asks "What's the problem represented to be" (WPR) before introducing our research questions and our rationales for case and data selection.After reporting our empirical findings regarding the problems targeted by TE, we elaborate the arguments and underlying assumptions associated with particular problem framings, and discuss how these problematizations contribute to subjectification effects.In conclusion, we review the implications of our findings for policy, practice and future research.

Previous research
In attempting to address the legitimacy crisis of public services and the welfare state, national governments have adopted and recontextualized the new public management approach, with varying results, which have attracted considerable attention among education scholars (Ball, 2003;Gunter et al., 2016;Verger & Curran, 2014).Among these, Ball (2003) explored how new policy technologies invite particular discursive interventions in education reform.Subject positions in market, management and performance discourses are linked to values like competition and individual performative worth within the disciplining logics of competition, efficiency/productivity and fabrication (Ball, 2003).One salient implication in the present context is how recent developments have driven policy makers to continuously reinvent output-oriented tools for evaluation and assessment in order to meet new accountability demands (Gunter et al., 2016;Pons, 2017).Under the influence of what in the research literature often is labelled performance-based accountability (PBA), teachers' roles and associated expectations have changed accordingly, along with tools for evaluating teachers' work.
To a great extent, PBA reforms depend on complex models of TE driven by the "crisis" discourses of international comparative surveys such as the Programme for International Student Assessment (PISA).Many recently developed TE tools (elaborated below) seem to appeal to policy levels as they purport to be "objective" and "neutral" instruments for measuring how teaching relates to learning outcomes (Lewis & Holloway, 2019).By 2013, 22 of 28 OECD member countries had implemented frameworks for TE; Norway was among the minority that had not (OECD, 2013).This in turn increases the pressure on teachers and school leaders to comply with expectations of increased student performance (Lewis & Holloway, 2019).Some scholars have argued that as teachers lose power to define their role, as well as ethical and decision-making autonomy, this deprofessionalization may prevent them from questioning policy orientation or negotiating employment conditions and teaching content (Maroy & Voisin, 2017, p. 16).There is also concern that these reforms and new evaluation tools contribute to the production of new kinds of teacher subject (Ball, 2003;Lewis & Hardy, 2015;Perryman et al., 2017).
In the international literature on TE (predominantly from the USA), value-added models and frameworks for classroom observations are among the tools most widely discussed (Papay, 2012).Value-added models purport to quantify teacher and school effectiveness by measuring the outcomes of student learning as pre-and post-school year differences in student achievement.As these models often inform high-stakes decisions and provisions for teacher accountability, they have attracted increasing international attention (Levy et al., 2019) with reportedly detrimental consequences (Paige & Amrein-Beardsley, 2020).In particular, these models have been criticized for failing to acknowledge important methodological and ethical qualities of relevance to TE (cf. ASA, 2014;Baker et al., 2010;Ravitch, 2013).
TE is often characterized as formative, to the extent that it advances professional development and strengthens the legitimacy of the teaching profession (OECD, 2011a).However, new tools for TE emerge within a socio-political context, and a growing body of research suggests that TE tools and practices embedded in PBA policies are essentially summative mappings of teacher efficiency for the purposes of accountability and governing (Hallinger et al., 2014;Lillejord et al., 2014).One common argument for accountability is that because most schools are publicly funded, quality assurance and accountability is justifiable, as taxpayers and politicians want to be sure that public money is well spent (Gorard, 2010).Thus, in the beginning of the 2000s, TE was frequently promoted as a means to increase transparency and to ensure that teachers and principals were held accountable for all students' outcomes, regardless of socioeconomic background (Reynolds, 2007).One OECD report identified teacher quality as the "determinant of pupil outcomes because social background and student abilities are not open to policy influence" (Connell, 2009, p. 225).However, research findings since the 1960s (Coleman, 1966) underscore the influence of socioeconomic background when addressing problems of inequality in education (Broer et al., 2019).Thus, political problems reflect how a society deploys its resources and how collective decisions are made (Connell, 2009).

Te in Norway
The issues outlined above provided the context for our investigation of TE as a governing tool for education and teachers as developed and disseminated in Norwegian policy proposals from 1988 onwards.In this section, we briefly describe some features of the Norwegian context that explain the particular relevance of investigating TE in Norway.
Norway's 356 municipalities and 11 counties 2 differ widely in terms of preconditions for systematic and comprehensive evaluation.Responsibility for developing, establishing and implementing TE systems is devolved to counties and municipalities as an element of local quality assurance.In many of the largest municipalities, information from examination results, national testing and student surveys is embedded to varying degrees in accountability practices.Although there is little systematic research, common TE practices include "school walkthrough" observation of teaching by superiors or internal or external "experts"; peer counselling; employee appraisals; self-assessment portfolios; evaluations based on student performance and student evaluations of teachers' instruction in upper secondary school surveys (e.g., Elstad et al., 2017;Lillejord et al., 2014).In addition, standards for "the good teacher", "the good class" as well as international observation protocols such as PLATO or CLASS seem to emerge within the decentralized competency development model (DEKOMP)a state initiative to promote collaboration between universities, colleges and municipalities at local level.
In the absence of explicit or comprehensive state frameworks for TE, it seemed relevant to investigate the subtler drivers of these developments in the associated policy discourses.The next section describes our theoretical framework, analytical approach and data collection procedure.

Theoretical framework and analytical approach
In studies of education policy, discourse frequently serves as an exploratory frame, but terms like discourse and policy are not always explicitly defined (Anderson & Holloway, 2020).The idea of policy as discourse is frequently applied in studies of policy sociology and education policy, as for example in the works of Ball (1990Ball ( , 1993Ball ( , 2003)).In describing policy as discourse, Ball highlights the ways in which global and national education policies intertwine and relate and how they "exercise power through a production of 'truth' and 'knowledge'" (Ball, 1993, p. 14).Discourses are seldom neutral; rather, they are shaped by policy agendas, scholarly traditions and networks and are often motivated by political interests, power relations, ideologies and rhetorical positioning (Anderson & Holloway, 2020).
Within the social sciences, established notions of policy as discourse have been invigorated by post-structuralist ideas about the relationships between power and knowledge, discourse, governmentality, subjectification and resistance (Bacchi & Goodwin, 2016).In particular, how policy workers and analysts approach specific tasks can be seen to contribute to shaping the social order, as policies, rules and regulations are informed by professional and "expert" knowledge that plays a significant role in how we are governed and in the kinds of "subjects" we are encouraged to become (Bacchi & Goodwin, 2016).Certain assumptions or "ontological politics" (Mol, 1999) shape the ways in which lived realities "are created by, rather than reflected in, social practices, including policy and research practices" (Bacchi & Goodwin, 2016, p. 6).
For present purposes, we view policy as discourse, represented by text in selected NOUs, and examine policy as the outcome of complex negotiation and contestation within languages and discourses (Goodwin, 2011).Our analysis of emerging TE policies in Norway exposes the underlying rationales and assumptions and reflects critically on how techniques of knowledge and strategies of power converge in discourse to form "local centres of power-knowledge" (Foucault, 1990;cited in Bacchi & Goodwin, 2016).To explicate the underlying assumptions of the political initiatives, we sought to reconstruct the associated problematizations of TE.To paraphrase Nicholas Rose (1999, p. 58), if TE is seen as the solution to political problems in Norwegian education, how are those problems to be concretely characterized?Adopting Bacchi's (2009) WPR approach, we unpacked the distinctive rationales and political objectives that motivated the introduction and retention of TE as a feature of governing education in Norway.The WPR approach is widely used to scrutinize the process of problematization in public policy-making in various sectors, including education.According to Foucault, a given problem representation produces a specific version of the phenomenon in question (Bacchi, 2009) in the present context, a particular version of TE.In this sense, problem representations are inherently political.In employing this approach to analyse green papers published by the Norwegian Ministry of Education and Research, we were able to explore discursive representations of the policy problems that TE is intended to solvethat is, the ways in which policy-makers express, frame and construct a particular issue and a proposed solution.
The WPR approach addresses six distinct questions to identify, reconstruct and interrogate problem representations in policy proposals, taking the proposal as a point of departure and then working backwards to identify the problem as represented.Because they are constructed, policy representations tend to highlight certain issues at the expense of others.The present WPR analysis of reports by government commissions of inquiry explicates the problems addressed by TE as an important aspect of ideological control policies "as becoming" (Ball et al., 2012).

Data and analysis
In the so-called "Nordic model of government", government commissions of inquiry play a key role in decision-making.Existing studies suggest that the comprehensive Norwegian ad hoc commission system exerts a significant influence on Norwegian policymaking (Christensen & Hesstvedt, 2019).A commission's work typically involves synthesizing existing knowledge and, in some cases, conducting or commissioning additional research (Christensen & Hesstvedt, 2019).After completing its work, the commission publishes its findings in a final report, preceded in some instances by an interim report, as part of the Official Norwegian Report Series (NOU).The principle of public access to official records in Norwegian public administration ensures that these reports are fully available to the public in the interests of democratic debate.NOUs are usually published annually and numbered sequentially, and we cite these sources in the native format (year:number: page).
The empirical material analysed here consists of commission of inquiry green papers published as part of the NOU series by the Ministry of Education and Research between 1988 and 2019.As NOUs typically present the first formal proposals for new or altered policies, they can be regarded as "raw material" in terms of new ideas or global discourses informing Norwegian policymaking.As a first step, we screened all available commission of inquiry green papers published by the Ministry of Education and Research in the NOU series.These were accessed through web databases at the Department of Knowledge and a database at the National Library containing scanned NOUs since 1972.To begin, we used the search terms laerervurdering [teacher evaluation] and laererevaluering [teacher assessment].As these produced relatively few hits, we extended our search to include related concepts that we considered relevant (see Table A1, Appendix 1).This search yielded 48 commission of inquiry reports published between 1988 and 2019.In a second screening, our qualitative reading of sections and pages to assess the relevance of these hits narrowed our data set to 11 reports (see Table A2, Appendix 1).
To guide our analysis of the 11 selected reports, we addressed four of the six WPR questions; as Bacchi ( 2009) noted, it may not be necessary to address all six questions in any given case.As a first step, we posed the following two research questions.
1. What problem is TE intended to target? 2. What underlying assumptions and arguments inform the framing of this problem?Question 1 sought to clarify the implicit representation of the TE problem within each NOU report.Question 2 sought to identify the assumptions underlying these representations and the contingent practices and processes that informed this understanding.Our analysis then proceeded to address two further questions.
3. How and/or where are these representations of the problem produced, disseminated and defended?4. What discursive effects and subjectification effects are produced?
According to Ball (1990a, p. 17), discourses are "about what can be said, and thought, but also about who can speak, when, where and with what authority".However, as meaning also arises from power relations and the associated concepts, categories and truth claims, question 4 refers to effects that emerge from what can be thought and said, and "subjectification effects" refer to how subjects and subjectivities are constituted by positioning those responsible, those with authority and those who must change (Bacchi, 2009).
While the study focuses on the problem(s) that TE was intended to solve, our findings are only meaningful within the prevailing political context, including the influence of PBA policies.The web of interconnections we identify aligns with Bacchi's view of the need to understand "the web of policies, both historical and contemporary, surrounding an issue" (Bacchi, 2009, pp. 20-21).

Findings and discussion
To structure our findings and analysis, we have contextualized the reform proposals from the late twentieth century in terms of contemporary policy discourse.We identify what was discursively at stake in committee reports at that time and where the respective problem representations were produced, disseminated and defended (questions 1 and 3).We also address the assumptions that underpin these "problem" representations during the respective periods of governing, and we discuss the contingent practices and discursive processes that emerged from these representations (questions 2 and 4).Period I (1988Period I ( -2001)).Paving the way: introducing and justifying the need for reform

Problem representations, producers and disseminators
The discourse analysis of committee proposals in the first period begins with NOU 1988:28 Med viten og vilje [With knowledge and will].The committee was tasked to assess goals, organization and future priorities for research and higher education.While this brief was not confined to compulsory education or teachers in particular, NOU 1988:28 clearly marked a point of departure for the new discourses of assessment and quality assurance in Norwegian education policy, paving the way for subsequent reforms that increasingly emphasized the significance of teachers and classroom practices for student outcomes.
In NOU 1988:28, the percieved problems of the existing system ("The Challenge", p. 7) were linked to accelerating global societal change as knowledge production and education became the key factors for a nation's success.One problem identified by the committee was that Norway " … does not get enough expertise from its population's talent" (p. 7).The committee also noted that "the results achieved are not at par with the skills that can be developed" (p. 7).Taking these issues together, the committee argued as follows.
(…) this is not just a matter of raising the achievements of those within higher education but of better utilizing the abilities of all.Without changes (…), [t]he national level will not reach the standard that must be maintained internationally.This is a pervasive problem for the entire system, from training in primary school to time spent in research (…) This message aligns with the conditions for economic growth: what we produce must measure up to what others are developing.Quality must be raised.(NOU, 1988:28, p. 7) This assertion that knowledge and competence production in the Norwegian education system was not "on par with" other nations set the tone for subsequent discursive problem representations.NOU, 1995:18 addressed how Norwegian education could achieve equal and high-quality education for all, emphasizing competition and market orientation, parents' right to choose schools for their children and issues of "user participation" (p.58).This report also contended that measures were needed to maintain and enhance systems for quality assurance from a user perspective.State financial and legal measures for quality assurance should combine goal and outcome management with local routines for evaluation and reporting to increase transparency and to assess the effectiveness of these measures using instruments, surveillance and access to systematic accumulated information from different practices (NOU, 1995:18, p. 58).
The problem representations described above were echoed in a subsequent report addressing how teacher training could be improved (NOU, 1996:22, p. 8).The report proposed a common curriculum, goal-and performance-based management reforms and local school-based assessment systems to "improve the quality of education for all students" and to contribute to "information and management data for state authorities, county municipalities, municipalities, schools and others" (NOU, 1996:22, p. 65).
The stated empirical basis for the proposed new "knowledge policy" included earlier OECD reviews (1987,1988) of quality and assessment in education, which had outlined future challenges and tasks for Norwegian education governing.These reviews referred to difficulties in assessing and evaluating Norwegian education development and outcomes, as in the following quote.
The examiners did not find it easy to comment on the use of educational resources because such data were almost completely absent … In examining international data, we were surprised by the absence of material that might prove useful to the Norwegian authorities for their own planning system.(OECD, 1987;cited in NOU, 1988:28, p. 25) The NOU reports referred repeatedly to the OECD reviews from 1987 and 1988.They noted that Norwegian education could not be detached from the international "knowledge-explosion" (NOU, 1988:28, p. 7) and the associated requirement for education policy reforms, including measures to evaluate the disseminators of knowledgein other words, teachers.

Underlying assumptions and discursive and subjectification effects
The arguments underpinning the problem representations outlined above seem to invoke increasing societal "risks", and the proposed solutions involve the implementation of systems for quality assessment and assurance at all levels.These reforms were considered necessary for personal and cultural growth, high and secure universal standards and the development of democracy and national identity to ensure Norway's status as "a solid and creative member of the world community" (NOU, 1988:28, p. 7).The reports also emphasized the ability to draw comparisons with other nations.This discursive competitive emphasis on knowledge production increases the pressure on the nation state, its inhabitants and, crucially, stewards of knowledge dissemination to "deliver" on higher international (and subsequent national) expectations.These arguments entailed proposals for TE (NOU, 1988:28, pp. 82-84) and improved teaching in higher education (pp.85-88) as related means of maintaining control and ensuring quality.Among other provisions, this could be achieved by "stricter requirements governing teachers' ways of working … and on the departments' work on study plans and evaluation of how they work" (p.84).
In this context, it is interesting that although the committees represented the problem as insufficient knowledge production, the main policy problem as stated in the background material was not the actual production of knowledge but the governing authorities' lack of instruments for gathering data on the knowledge produced.It was claimed that implementation of a new "knowledge policy" (NOU, 1988:28, p. 7) for evaluation and assessment would bridge this gap by addressing these deficits to put Norway "on par" with other nations as the OECD reviews suggested.Comprehensive systems for data gathering were seen as an appropriate means of making municipalities and counties responsible for meeting expectations of "raising quality" (NOU, 1988:28, p. 7).These proposals appeared in the 1988 report and throughout the analysed material.One discursive effect of what was thought and saidin this case, as communicated in the NOU textsrelated to the teacher's roles in enabling Norway to participate in the global competition around knowledge production.Teachers as subjects are constituted by the OECD and to varying degrees by the Norwegian commission reports, which hold them responsible for changing their practices in line with new national and international comparative and competitive expectations.In reflecting on practice and notions of teacher professionalism and the importance of subjectification effects, we note how the term professional was used to convey teachers' responsibility in this regard."The thinking teacher" would be aware of the autonomy and trust afforded to them by the community while remaining engaged and critically constructive in terms of education reforms and committed to the practice of selfcriticism (NOU, 1996:22, p. 265).
In portraying teachers as insufficiently open to scrutiny or to suggestions for improving their practice, existing notions of the autonomous individual practitioner underwent a discursive shift; teachers needed to understand that everything was changing and that they must do likewise.Continuous self-evaluation of teaching practices was seen as the "answer" to the problem of instability in a changing world, and this "reorienting and changing professional" in turn had to be more "open" and "flexible".The trust placed by the community in the teaching profession implicitly entailed an expectation of transparent professional practice and the fulfilment of any expectations arising from future societal change.In subsequent years, this requirement was communicated in proposals for performative accountability.
Period II: defining quality to make it (ac)countable (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013) Problem representations, producers and disseminators Having set the scene for education evaluation and assessment, the next period of development was characterized by more concrete proposals for addressing the problems in Norwegian education.In the aftermath of the shock of PISA 2001, two significant NOUs ramped up the PBA rhetoric in Norwegian education policy.NOU, 2002:10 opened with the following quote: "The world is not examining how one thing is put into action but how it turns out". 3The significance of this assertion in the immediate context is that the process of learning (and other aspects of schooling such as formation, democratic understanding or citizenship) may be less important than its concrete manifestation, represented in this NOU as results.
Here, the previous emphasis on a "knowledge society" gives way to a "competency society" that measures the outcomes of schooling in terms of competence rather than knowledge.NOU, 2002:10 highlighted the lack of systematic results data, systematic use of mapping and testing, evaluation tools for "school-based assessment" and improvement or follow-up routines.In addition to the emphasis on data-gathering for quality assessment, the concept of accountability was also introduced.Based on OECD advice, NOU, 2003:16 contended that future teachers should be more actively involved in research, with more basic education studies to identify "what works" and "evidence-based practices" (p.277).Regarding the state's role, the report emphasized the importance of explicit knowledge for research and evidence-based policymaking.
Another important feature that contributes to the increasing demand for educational research is that the authorities increasingly emphasize goals and results rather than rules.This increases the need for more explicit knowledge of the results of various processes locally, nationally and internationally.Research results are also used to improve education in many OECD countries.In the US, 'evidence-based' policy-making is used in primary school development.(NOU, 2003:16, p. 274) This led to the proposal for a "quality portal" [kvalitetsportalen] (NOU, 2003:16, p. 245)an online tool for comparing indicators of performance, process and structural quality in basic education.Assessment based on data and results as provided through the quality portal was seen to represent a solution to problems in Norwegian education.The report noted how societal changes created new expectations: "the competence society [demands] specialization and cutting-edge expertise, leading to schools that resemble other knowledge companies" (NOU, 2003:16, pp. 61-62).Accordingly, teachers would expect changes in working conditions, with rewards based on the evaluation of results produced.
… Teachers' working conditions will approach those of other groups, and they [teachers] will be required to provide documentation in relation to the results they achieve and rewards will be developed in relation to results.This will require more effort in relation to developing competence at level of the individual teacher and school leader, but also a systematic follow-up from the school owners.(NOU, 2003:16, pp. 61-62) 3 From the play Jean de France (1722) by the Danish-Norwegian writer Ludvig Holberg (1684-1754).
The reference to "rewards … in relation to results" implies that the implementation of comprehensive incentive-based TE frameworks was seen as a possible means of competence development.
Further, with reference to another OECD review from 2009, NOU, 2010:7 added minority and gender issues to the list of relevant problems.The OECD report noted that Norway's decentralized education system still lacked a "robust system of accountability" (p.66) and required additional measures to "strengthen the accountability model through clear goals, in particular with a view to achieving improvement in learning outcomes for students from an immigrant background" (p.66).The OECD report also highlighted the lack of clear teaching goals, noting that "evaluation results are used to a small extent to improve teaching and learning" (p.66).Expressing concern about the effects of increased segregation in Oslo on student learning outcomes, the OECD proposed supervision "to evaluate different pedagogical practices in schools with a diverse student body and to identify examples of good practice" (OECD, 2009a;in NOU, 2010:7;p. 66).
NOU, 2003:16 characterized teacher quality as a faceted term that cannot easily be specified by a set of measurable units (p.56).The report added that engagement and knowledge is not enough: "Researchers further noted that cognitive abilities are a key factor in several studies" (p.57).

Underlying assumptions and discursive and subjectification effects
The underlying assumptions identified in reports from period II build on discourses identified in period I.The need to control "inputs" (in terms of public spending on education) and "outcomes" (in terms of student performance and the proportion accessing higher education) prompted the turn to PBA as a means of making learning outcomes more transparent and accountable.To that end, new market-oriented instruments were introduced, emphasizing goal and result management and accountability based on surveys and user perspectives, including "systematic follow-up of quality development at the individual level" (NOU, 2003:16, p. 12).In this way, the NOU extended the focus to the quality of individual learning outcomes and teachers' individual responsibility to ensure that every student received "adapted training". 4This in turn paved the way for holding teachers accountable for student performance.
In operationalizing the concept of responsibility, the 2002 report was the first Norwegian green paper to use the term accountability (in English), building on previous green papers' efforts to contextualize global education discourse around transparency, accountability and user orientation.
Another argument for assessing quality, which is relatively little used in Norway, is the idea of a kind of societal accounting obligation or so-called accountability.Internationally, the term accountability is used in many contexts … both in a purely financial sense as 'value for money' and in the political sense … that governments have a democratic right to control systems that should serve a community mission.In this sense, modern school systems are often regarded as efficient and of good quality if they fulfil the users' wishes and needs.(NOU, 2002:10, p. 24) Used here to justify the government's "democratic right to control" and to provide "efficient" and "good quality" modern schooling (NOU, 2002:10, p. 24), the introduction of accountability generated a discursive shift, reframing schools and teachers as servantsnot necessarily to the more general democratic goals of schooling but rather to specified government goals and the "wishes and needs of individual 'users'" (NOU, 2002:10, p. 24).
The concepts of accountability and evidence-based practice underpinning the description of the role and significance of teachers, implicitly questioned teachers' abilities to raise educational outcomes in terms of "result quality", also holding them responsible for unintended consequences such as teaching to the test: the assessment scheme.Thus, any major investment in quality improvement must be thoroughly documented and well planned.(NOU, 2003:16, p. 282) This reference to "professional compliance" implicitly assigned responsibility for the unintended consequences of education policies to the teacher's lack of professionalism.Regarding the role of teachers and the problems that TE might help to solve, the 2003 report indicated that teachers would increasingly be expected to compensate for other societal developments such as increasing inequality (also referred to in NOU, 2010:7).Rather than problematizing recent societal developments, these NOUs framed changing demands and increased expectations as inevitable: "Schools and teachers must therefore get more help to comply with "what works" and to translate expectations locally" (NOU, 2003:16, p. 61).The latter part of that statement underscores the discursive replacement of local(ized) knowledge by an expectation "to comply with what works" and a move towards decontextualized evidence-based practices to enhance teaching and learning.
Making teachers accountable for how they generate "value for money" produces a persuasive subjectification effect.In order to compensate for societal problems such as inequality, gender and minority/language issues, teachers were to be held to account for student performance outcomes in order to comply with the needs of the government, society and the individual "user".This message was emphasized in the following period (III) by a discursive shift from "teaching quality" to "teacher quality", implying that teachers' professional identity and competence must not only be developed but also justified (NOU, 2003:16, p. 63).
Period III: from teaching quality to teacher quality; justification of practice (2014)(2015)(2016)(2017)(2018)(2019) Problem representations, producers and disseminators Our main findings from this period relate to reports by the Ludvigsen commission (NOU, 2014:7 and NOU, 2015:8) and the later Stoltenberg commission (NOU, 2019:3).The two Ludvigsen reports marked a rhetorical shift from the earlier focus on testing and results that would "give the school's broad educational mission a clearer place in everyday school life".According to NOU, 2014:7, [testing and results] do not provide a comprehensive picture of the situation in a school or municipality or across the whole country.There is therefore a need to produce supplementary qualitative and quantitative knowledge for assessment and development purposes, both locally and nationally.(p.28) While such statements imply a broader approach to the purposes of teaching, our analysis revealed that the proposals and justifications for new governing tools confirm that TE itself entails certain discursive and subjectification effects.
Although not explicitly related to TE, the most notable finding from this period is the OECD's pervasive influence on quality assessment premises in general and the definition and assessment of teacher quality in particular.Education quality was still largely defined by how "it turns out" (NOU, 2002:10)in other words, by the quality of the results.In this context, the problem was now identified as teachers' and school leaders' lack of competence and capacity to interpret results or to use them systematically to improve their work.Pointing to a lack of systematic TE and "school-based assessment and development", justification for these arguments was based on the OECD's Evaluation and Assessment Frameworks to Improve School Outcomes project, which analysed assessment frameworks in 28 countries (OECD, 2013).
Education systems are more oriented than before towards the use of evaluation results for evidence-based decisions, measuring student learning outcomes and comparing achievements over time (…) emphasizing student results and assessment of school and teachers.(…) Common challenges in many OECD countries include ensuring that assessment systems are based on student learning outcomes and that different assessments and evaluations are well linked.(OECD, 2013;in NOU, 2014:7, p. 109) As this quote suggests, the prevailing view was that assessment systems did not adequately reflect student learning and that the connection between external and internal supervision was inadequate (NOU, 2014:7, p. 109).The report problematized the existing decentralized model, which delegated responsibility for designing feedback systemsincluding TEto school owners and schools: "Despite … positive development in recent years, both Norway, Sweden and Denmark face challenges in terms of the quality and systematics of schools' self-assessment processes (OECD, 2013(OECD, , 2011a(OECD, , 2011b(OECD, , 2011c)" (NOU, 2014:7, p. 109).The OECD's advice was to assess more components of schooling, including teachers, using status reports to ensure better accountability.
In Period II, NOU, 2010:7 proposed accountability as a solution to the newly added minority and gender issues.Similarly, NOU, 2015:2 noted the need for "clearer accountability for the school owner school" and a "strengthened support system with clear responsibilities and division of tasks" (p.20) to solve problems related to students' psychosocial environment.For students with high learning potential, NOU, 2016:14 framed the problem as teachers lacking the knowledge, capacity and competence to change classroom practices.
In addressing gender differences in educational achievement and attainment, NOU, 2019:3 represented the most explicit attempt to position TE as a solution to the perceived problems in Norwegian education.
As the teacher is the key input factor in schools, the characteristics of individual teachers are central to achievement and learning.Existing research indicates that these characteristics are slightly affected by the teacher's education, admission requirements or gender.However, as they may also relate to the teacher's cognitive abilities, it seems important to recruit motivated and talented people to the teaching profession.Research findings suggest that much can be done to provide teachers with tools for self-improvement.(NOU, 2019:3, p. 20) That report suggested that school quality should be assessed in terms of results attributed to variables including "teacher quality", "characteristics of good teachers" and accountability measures related to student performance.Drawing on findings from a Norwegian study of accountability systems in Oslo in 2005 (Gjefsen & Gunnes, 2016), the report suggested that "(…) the introduction of school contribution indicators and the accountability of school leaders in the upper secondary school in Oslo in 2005 led to an increase in teacher drop-out" (NOU, 2019:3, p. 144).Concluding that "Most teachers who chose to quit left the teaching profession and were replaced by teachers who managed to deliver good student results" (NOU, 2019:3, p. 145), the report underscored the potential "value added" benefit of accountability policies.

Underlying assumptions and discursive and subjectification effects
In terms of discursive effects that emerged in Period III, knowledge-based management and soft governance stand out as proposed solutions to the perceived lack of systematic quality assessment: "The OECD pointed out that teacher assessment systems are the least developed area in Norway" (NOU, 2014:7, p. 108).This was the first Norwegian green paper to introduce "criteria for good teaching": "One of the OECD's general recommendations to the Norwegian authorities was to clarify goals and criteria for assessment and evaluationfor example, criteria for good teaching" (NOU, 2014:7, p. 108).The 2011 OECD report referred specifically to Norway's lack of systematic and comprehensive TE as part of the national quality assessment framework, and this proposal offered a solution to those problems, including frameworks for assessing teachers' work and, more specifically, making teaching transparent.
NOU, 2019:3 offered a clearly defined approach to knowledge, research and evidence-based teaching practices.The proposed solutions included reference to effect-based research and quantitative methods for "measuring teaching".To address the lack of control over the contribution made by Norwegian teachers, the report proposed laererbidragsindikator [teacher contribution indicator] (p.146), more commonly known as value-added models.The suggested combination of tools included teacher contribution indicators and judicial review of the relationship between professional judgment and the employer's right to control.The proposed measures and the selected supporting research revealed a "trust in numbers" (Porter, 2020) and a conviction that PBA could improve education quality.In addition, the proposed "soft governance" was based on the assumption that the professional community would be more likely to respond positively to "knowledge-based management", "competence development" or "voluntary standards for best practice" than to "objective and detailed rules or individual rights" (p.162).
The findings from Period III suggest that the discourse was less competitive than in the two previous periods, with less focus on how to "keep track" or "be on par".It also seems clear that there were fewer references to PISA than in the previous period (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014).However, we also detected a more individual focus at teacher level, including "cognitive skills" (NOU, 2019:3), which portrayed teachers as the main school input to student performance.With regard to subjectification effects, this part of the analysis demonstrated how teachers are constituted by problem representations that position them as responsible for change.Expectations regarding the delivery of education quality were redefined, as teachers were required to justify their daily practices in relation to predefined standards or "value added" to students' learning outcomes.Holding teachers to account for how they generate "value for money" reframes teachers as ethical subjects and modifies what is seen to constitute teacher quality and professionalism.

Conclusions, implications and suggestions for future research
The findings reported here highlight the role of discursive processes in the "making" of Norwegian teachers, including the problem representations, arguments and assumptions underpinning assessment reforms and new tools for TE.Over time, and regardless of the Norwegian governments' political left, centre or right persuasions (see Table A3, Appendix 1), we have identified the OECD as the most prominent contributor to discourses linking quality assurance and "teacher quality".In so doing, this article clarifies how techniques of knowledge and strategies of power converge in global discoursein this case, the discourses around TE as a means of solving political problems in Norwegian education.The underlying assumptions reflect the increasing "risks" associated with sustaining the welfare state, equality, efficiency, effectiveness and transparency in a globalized world by engaging with user needs and securing Norwegian knowledge production.These "risks" prompted a commitment to goal and result management and "evidence based practices" to increase transparency within a decentralized education system.Images of the "self-evaluating" and "flexible" teacher (NOU, 1996:22) were grounded in arguments that teacher professional identity and competence depend on continuous development and justification (NOU, 1996:22, p. 52;NOU, 2003:16, p. 63).Gathering evaluation data and TE were thought to increase transparency, making it possible to hold teachers and school leaders accountable for results.
New policy ideas and proposals for TE were imported from other contexts, as for example in the proposal to implement value-added models (NOU, 2019:3).The research selected to support these arguments revealed a "trust in numbers" (Porter, 2020), based on the underlying assumption that you cannot control what you cannot measure.TE would enable government to introduce controls at the level of the individual teacher.On that basis, we contend that attempts to "make" Norwegian teachers by implementing TE have been influenced by knowledge decontextualized of local classroom practices.
The OECD's prominent role in disseminating knowledge to "justify" teaching practices seems to have had less practical impact in the Norwegian context than in other OECD countries.We contend that discursive acknowledgements make Norway rather unique in international terms as a setting in which to investigate TE practices, as the country's policy-making processes partially seem to accommodate the agency of teachers and school leaders.In this regard, certain distinguishing features of the Norwegian education system seem to strengthen its resilience to high-stakes managerial approaches to TE.As Norway is among the few OECD countries with no comprehensive centrally initiated TE frameworks, further investigation seems warranted to identify the distinctive features of this apparent resilience.
As TE in Norway emerges under the umbrella of school-based assessment and development based on soft governance rather than high-stakes management, it is especially important to take account of discursive and subjectification effects, such as those pointed out and elaborated in each of the three periods analysed here.Coupled with TE and logics of PBA, the teacher's role in "knowledge production" and performance outcomes can be viewed as political reconstruction and disciplining of teachers as ethical subjects (Ball, 1990), increasing the pressure to "deliver" on ever higher expectations defined by measures decontextualized of Norwegian classrooms.
Among the limitations of this study, the selection of documents concentrates on NOUs that potentially guided new proposals and policy initiatives in Norwegian education.An understanding of actual policies implemented and the breadth of the policy context is beyond the scope of this article, as such an understanding would necessarily have required other sources of information, such as white papers, interviews with school professionals and/or observations of local professional practices.Additionally, search terms necessarily narrow the field of study.Coding, categorization, interpretation and analysis reflect the analytical and theoretical frameworks applied as well as the researchers' positionality, all of which are limiting factors.Nevertheless, we hope our findings will invigorate debate on the implications of discourses surrounding teacher agency and notions of teachers' professionalism when future policies are to be developed.Further research on TE in Norwegian education might usefully investigate the nature of NOU hearings, with particular reference to committee compositions and the involvement of Norwegian teachers and school leaders in the shaping of policies directly related to their daily work.We would also welcome further research on TE enactment processes in Norway, including teachers' and school leaders' perceptions of TE as a tool for professional development.

Table A2 .
List of included NOUs.Title of NOU, Norwegian [English]

Table A3 .
List of Norwegian governments from 1986 through October 2021.