A Discretionary Toolkit: Reasoning When Teaching Controversial Issues in Norwegian Upper Secondary School

Using a toolkit approach in combination with the concept of street-level bureaucracy and theories of discretion, this article has empirically investigated the resources that influence teachers’ discretionary reasoning when teaching controversial issues. The analysis has been based on 32 classroom observations at two upper secondary schools in Oslo, Norway, in one Religion and Ethics and one Social Science class, and interviews with 16 teachers who taught the same subjects. The results have shown that professional competence, professional and personal values, and relationships with pupils worked as a toolkit of resources that teachers could draw upon when making discretionary judgments in different contexts. A better understanding of teachers’ use of discretionary reasoning may enable curriculum developers and policymakers to support teachers in the complex social landscape of teaching controversial issues.

This article empirically investigates which resources influence teachers' discretionary reasoning when teaching controversial issues. Teachers' work is filled with planned and unexpected situations where they can choose among several courses of action. They must handle subjects such as racism, immigration, or religion. These issues can stir controversy and emotionally-charged responses in their classrooms, and thus require discretionary judgment in terms of how to present the subject matter and how to respond to different kinds of reactions among pupils.
Several studies have researched how teachers handle controversial issues in the classroom (Hess, 2004;Oulton, Day, Dillon, & Grace 2004;Stray & Saetra, 2016). However, little is known about which resources teachers draw upon when facing challenging teaching situations or how discretionary reasoning plays a role when choosing among different courses of action. Teachers are responsible for pupils' upbringing and learning. As professionals, teachers have specialist knowledge that distinguishes them from the unskilled. A core element of professional work is the use of discretion (Freidson, 2001;Lipsky, 2010). Lipsky (2010) termed the professionals who are responsible for implementing public policy in close contact with client's street-level bureaucrats. Teachers are tasked with implementing broad educational policies as they are stated in curricula through direct interaction with pupils and can therefore be described as street-level bureaucrats. Besides teachers, typical street-level bureaucrats are police officers, social service workers, or health workers-professions that are representatives of the welfare state.
According to Lipsky (2010), street-level bureaucrats' work is both (1) highly scripted and controlled and, at the same time, (2) dominated by improvisation and responsiveness. Molander (2016), describes the tasks they carry out as being "such that political authorities cannot specify exactly what the problems are, exactly what to do and exactly how much resources to deploy in particular cases" (p. 2). Ideally, street-level bureaucrats respond to the needs of individuals they meet. They are expected to use discretionary judgment because they have little direct supervision, and their work is too complex to be reduced to established guidelines.
A wealth of studies focuses on the use and limitations of discretion among professionals responsible for implementing public policy in close contact with clients. Common threads in the literature on teachers and discretion are how large the discretionary space is after applying the rules and regulations, and how teachers handle the discretionary space available to them (Hagelund, 2010;Karseth & Møller, 2018;Maynard-Moody & Musheno, 2003;Svensson, 2019). Researchers debate whether teachers, as street-level bureaucrats, have the same space for discretionary judgment now as when Lipsky wrote his book in the 1980s (Boote, 2006;Taylor, 2007). In a study from the United Kingdom, Taylor and Kelly (2006) claim that teacher's discretionary judgment has changed, and that one must analytically divide discretion into parts to understand whether it still plays a role in teachers' practices. They argue that discretionary judgment based on rules is used less often than 3 judgment based on values or tasks. A South African study concludes that teachers exercise a high degree of discretion in interpreting school rules, prioritizing what to teach, and emphasizing guidelines (Mutereko & Chitakunye, 2015).
Studies have shown that discretion is a slippery, hard-to-capture term (Brodkin, 2012;Evans & Harris, 2004). Overall, one can say that "to have discretion is to have authority to decide in cases where rules and standards do not prescribe determinate results" (Molander, 2016, p. 10). Researchers such as Hupe (2013) and Molander (2016) distinguish between two understandings of discretion: on the one hand, how much freedom rule-makers give to the actors who comply with the rule, and on the other, the ways in which actors actually use this freedom in specific circumstances. The concept of discretionary reasoning is related to this second understanding. Despite the fact that discretion is seen as an inevitable part of professionals' work, Grimen and Molander (2008) and Molander (2016) claim that the literature on professions does not fully address the concept of discretion and that the actual exercise of discretion-the discretionary reasoning-often remains an unanalyzed residual category.
Further, Watkins-Hayes (2009) argues that researchers and policymakers emphasize the outcomes of bureaucrats' decision-making rather than focusing on the complex social process of making discretionary decisions. It is, however, important to understand how discretionary reasoning works: If discretionary judgment is to be trusted, then it must appear reasonable to people other than the person exercising the discretion. There is a need for research that can provide a better understanding of teachers' use of discretion by unpacking what influence their discretionary reasoning. This article analyzes how teachers draw upon different resources as a toolkit for discretionary judgments using teaching of controversial issues as a case.
Von der Lippe (2019, p. 2) argues that what should be understood as controversial depends on both the social context and the political climate and may differ in both time and place. What are perceived as controversial issues depend on context and vary between country, city, school, and classroom. In one classroom, discussing immigration can stir controversy; in another, climate change may be perceived as controversial. Studies on controversial issues, both internationally and in Norway, focus mainly on if or how teachers handle these issues in the classroom (e.g., Anker & von der Lippe, 2016;Hess, 2005;Stray & Saetra, 2016). However, to my knowledge, none combine that focus with theories of discretion. In this study, I apply a broad understanding of controversial issues, understood as issues that teachers perceive to be particularly challenging or sensitive to teach because there was no scripted way to handle the situation and thereby demand teachers' use of discretion. Exploring what influences teachers' discretionary reasoning through the lens of controversial issues can contribute to understanding teachers' practices and help unpack how teachers exercise discretion. Thus, the research question is: How do teachers draw 4 upon different resources to make discretionary judgments when teaching controversial issues?
The framework I use for exploring this question is inspired by theories of discretion and street-level bureaucracy combined with Swindler's (1986) toolkit approach. The analysis is based on interviews with 16 teachers who taught Religion and Ethics or Social Science at three upper secondary schools in Norway's capital city of Oslo and 32 observations of one Religion and Ethics class and one Social Science class at two of the schools where I conducted the interviews.

Perspectives on discretion
Embedded in being a teacher is the uncertain nature of teaching. Helsing (2007) notes that teachers "experience uncertainties due to the complex nature of their work, which is centered on human relationships and involves predicting, interpreting and assessing others ' thoughts, emotions, and behavior" (p. 1317-1318). Dealing with uncertainties requires discretionary judgment. My understanding of discretion follows previous research, defining discretion as "an area where one can choose between permitted alternatives of action on the basis of one's own judgment" (Wallander & Molander, 2014, p. 1). Molander and colleagues (Grimen & Molander, 2008;Molander, 2016;Molander et al., 2012;Wallander & Molander, 2014) separate the interlinked structural and epistemic aspects of discretion to distinguish rules from reasoning in decision-making. Delegating discretionary power assumes the actor performing the discretion has good judgment and can decide based on reasoning (Grimen & Molander, 2008;Molander, 2016). When explaining structural discretion, Molander (2016) refers to Dworkin's metaphor: Discretion is a doughnut hole surrounded by a belt of restriction. The discretionary space for judgment-the hole-can be bigger or smaller, and discretion is the "freedom of choice delegated by authority" (p. 21). Previous research shows that several institutional or contextual factors influence teachers' discretionary space and, thereby, action in the classroom (Andresen, 2020;Karseth & Møller, 2018;Svensson, 2019).
Epistemic discretion-discretion as reasoning-assumes discretion is based on reasoned judgment and decisions, regardless of how large or small the discretionary space is. Under conditions of indeterminacy, where the available reasons do not warrant one outcome, actors must rely on their own discretionary judgement to make a conclusion (Molander, 2016). Zacka (2017) stresses that discretion is "not doing as one pleases" (p. 34) within regulatory boundaries but is internally constrained by a reasonableness standard. For example, street-level bureaucrats must explain the grounds upon which they based their chosen course of action (Lipsky, 2010;Zacka, 2017). Grimen and Molander (2008) define epistemic discretion as reaching a conclusion from a situational description in combination with a norm, and the norm justifies the steps between the situational description and the conclusion. Following Molander (2016), such action norms are similar to what Toulmin (2003) calls warrants. In a practical reasoning context, the warrants' strength is the most important variable. Warrants can be strong or weak, and discretion can be a form of reasoning when the warrant is weak. Thus, although people may be conscientious and thorough in their reasoning, they can reach different conclusions about the same problem (Grimen & Molander, 2008;Molander, 2016).
Much extant literature views discretion as influenced by different forms of reasoning. For example, Karseth and Møller (2018) understand professional competence as central for discretionary reasoning. Others see street-level bureaucrats as moral agents (Zacka, 2017) or as influenced by ethics codes or values (Taylor & Kelly, 2006). Maynard-Moody and Musheno (2000), in turn, see discretionary judgment as embedded within the relationships between street-level bureaucrats and citizens. A recent study on discretionary reasoning in Norwegian welfare offices shows how an institutional logic or a "norm of action" guide the frontline workers reasoning when concluding a case (Håvold, 2019). In a similar vignette study, Møller (2016) demonstrates how Danish caseworkers use different forms of categorization to evaluate their clients. Both studies highlight certain factors that influences the discretionary reasoning of the frontline workers decision-making regarding their clients.
Teachers as professionals exercise discretion in a different manner. During a classroom session, they continuously make discretionary judgments based on the context of that particular subject and the class they teach. They must choose what teaching strategy will give the best learning outcome, how to address a certain issue given the composition of pupils, and whether they should prioritize a coming exam or the school system's more general educational ideal. Harrits (2016, p. 13) argues that professionals such as teachers who are in close contact with citizens combine a logic based on formal training and knowledge with a more personal, relational, and emotional-based logic when making discretionary judgments.
Inspired by Swidler (1986), these different resources can be conceptualized as cultural resources, which teachers can incorporate into their daily work. Due to the complex nature of teachers' work, it may be useful to view these resources as a repertoire that teachers can draw upon when making discretionary judgments. Swidler views people as competent users of such toolkits and differentiates between situations in which people follow established action strategies without much reflection or resistance (settled lives) and those in which they test new strategies (unsettled lives). I argue that teaching potentially controversial issues can be seen as unsettled situations, where teachers must adapt teaching strategies to the current theme and context using discretionary judgment. Watkins-Hayes (2009), inspired by Swidler, had developed the term discretionary toolkits in her study of welfare officials who are in close contact with their clients. She writes, "Discretionary toolkits denote and organize the capabilities, perceptions, resources, and choices that organizational actors have at their disposal to shape institutional actions and outcomes" (p. 56). Focusing on the discretionary reasoning of teachers in situations where they teach potentially controversial issues at two different upper secondary high schools in Norway, this article will explore what kind of discretionary toolkits teachers draw upon when teaching controversial issues.

Data and methods
Data for this analysis were drawn primarily from 16 interviews with seven Religion and Ethics teachers and nine Social Science teachers at three upper secondary schools-one on the west side and two on the east side of Oslo, Norway (anonymized as Western High, Eastern High, and Eastern High 2). I also conducted 32 classroom observations in one Religion and Ethics class and one Social Science class at Western and Eastern High.
The analysis used the case of Norway. In Norway, 93% of 16-to 18-year-olds attended upper secondary schools, and there has been a strong ideological tradition for educational institutions preparing children to take part in a democratic society (Møller & Skedsmo, 2013; Union of Education Norway, 2019). A central aspect of the Norwegian school system has been to teach democratic citizenship, cultural diversity, and critical thinking (The Education Act of 1998, §1-1) in a society that has become increasingly diverse through modern immigration. Regardless of social background, gender, religion, or ethnicity, everyone has the right to be included in what is often called the Norwegian unitary school. The public educational system thus exposes individuals to a common frame of reference (Kjeldstadli, 2014).
In Oslo, 40% of pupils were immigrants or children of immigrants. However, the ethnic diversity vastly differed among Oslo schools. The two schools where I conducted classroom observations were purposively selected to vary in pupils' ethnicity, socioeconomic backgrounds, and grade-point average needed for acceptance. All but two interviewed teachers worked in those two schools (the other two worked at "Eastern High 2"). I did not observe lessons in Eastern High 2 due to conflicting teaching schedules but the two teachers I had been in contact with were interested in participating in an interview. At Eastern High and Eastern High 2, most pupils in the Educational Program for Specialization in General Studies were immigrants or children of immigrants. (General Studies prepares pupils for higher education; their other option is a vocational education program.) At Western High, pupils in the same program were mainly of Norwegian descent.
The interviews were conducted during and after 3 months of observation. While observing, I sat in the front or back of the classroom writing field notes throughout the lesson. I gained access to the schools through the principal at each school. The teachers and pupils were informed about the project and gave informed consent. Although data from the interviews form the main basis for the analysis, my observations in the four classrooms were discussed with the informants during the interviews and helped shape the interview guide. For instance, I used examples from the classroom observations to start conversations around different topics and to encourage teachers to elaborate on situations I observed. The interviews were conducted in meeting rooms or empty classrooms at the school and lasted 7 about an hour each. I recorded the interviews and had them transcribed afterwards. All informants and schools are anonymized in this article.
The 16 teachers' educational backgrounds varied. Three had specific teaching education from a university; 13 had general academic backgrounds in Norwegian literature, sociology, religion, history, social anthropology, political science, or economics before completing a 1year undergraduate teacher-training program required to teach at upper secondary schools in Norway. One teacher was from another Nordic country; the rest were of Norwegian descent. I chose to observe the subjects of Religion and Ethics and Social Science because the curricula specified that the subjects necessitated discussion and reflection (The Norwegian Directorate for Education and Training, 2006;2013). They exemplified issues that, in the interviews, teachers described as demanding their sensitivity and discretion. Both subjects were obligatory for all pupils who attended upper secondary school in the General Studies program.
I analyzed my observation fieldnotes and interview transcripts inspired by principles of thematic analysis-a method to identify, analyze, and report key patterns or themes in a dataset (Braun & Clarke, 2006). The first coding round for this article was data driven, and the concept of teachers' discretionary reasoning was present in many codes. In the second round, informed by the previously presented perspectives on discretion, I looked for excerpts in which teachers talked about what had influenced their discretionary reasoning when teaching controversial issues. Three themes became prominent in the material: professional competence, professional and personal values, and relationships with pupils. I viewed these themes as resources in the teachers' discretionary toolkits. In the following analysis, I use quotes from selected teachers, which illustrates broader patterns in the empirical material.

Results
I present the three resources-professional competence, professional and personal values, and relationships with pupils-separately to illustrate their differences and to show how teachers used parts of their discretionary toolkit in different contexts. Inspired by Molander (2016), I address the resources that teachers said shaped their everyday discretionary reasoning when teaching controversial issues. Discretion as reasoning emphasizes how discretion allows tailoring decisions to each case. The analysis showed that the interviewed teachers drew upon different resources from their toolkits interchangeably, depending on the topic and context, to make their discretionary judgments

Professional competence
The analysis showed that teachers' professional competence and education guided their classroom work and served as a resource for discretionary reasoning. The teachers' professional competence combined knowledge of the subjects they taught, pedagogy, subject didactics, practice, and the educational system's goals to promote democratic citizenship, tolerance, and inclusion. Teaching controversial issues gave teachers an opportunity to encourage pupils to think critically, articulate their problems, and understand other pupils' opinions, but it demanded much from the teachers (Stray & Saetra, 2016). Having knowledge of different topics and the schools' core values-that is, having professional competence-is the main resource teachers could draw upon in these situations. Professional understanding and competence were important because the teachers saw themselves as professional agents due to their strong professional identity (Harrits & Møller, 2014).
The Religion and Ethics and the Social Science classes had only a few lessons during the week but many obligatory topics to cover. Thus, teachers had a discretionary space wherein they needed to make discretionary judgments of what to prioritize. The teachers I interviewed had different educational backgrounds and thus different competences to draw upon when judging if and how to teach certain topics. Some argued that their educational backgrounds provided competence in teaching controversial issues; others disagreed. For instance, one teacher (Nina) explained that her teacher education stressed the aim of giving pupils knowledge about different religions to understand the world in which they lived and, through that knowledge, become more tolerant: The whole point with Religion as a subject was that [pupils] should be able to live as religious people or in a religious society and understand what that means. So, it is possible I have been influenced by that. And that's what I see they need. They need tools to understand and handle the situation they are in.
Nina's religion class was characterized by a diverse pupil composition. The pupils came from different cultural and religious backgrounds. Nina explained that she made discretionary decisions to expose her pupils to religious critique and prejudiced views on religion. These topics were perceived as controversial. However, embedded in Nina's professional identity was her mandate to teach her pupils how to handle these situations in the real world. She stated that felt she had the competence to do so.
Teachers who claimed they did not have enough competence to teach a subject made discretionary judgments to minimize the topics that could be controversial. This seemed to follow disciplinary distinctions: Both interviewed Social Science teachers with backgrounds in social economics stressed they did not always feel competent to teach controversial issues such as immigration or racism. Gina, a Social Science teacher at Western High, stated, "I think all the Social Science teachers in this school are better teachers than me. . . . I am the worst among the good." She laughed when she spoke but, during the interview, explained that she often experienced not having enough knowledge about sociological topics. Another teacher with an economics background at the same school had a similar experience; he stated that he felt the need to rely on the textbook, which he described as not nuanced enough: "Of course, I went through . . . prejudice and socialization and stuff like that. But I notice that it has something to do with my own skills. I am aware that I am not very good at it." Complex topics such as racism, religion, or sexual orientation demanded authority or confidence from teachers. The teachers had to feel able to use their discretion to handle possible prejudiced outbursts from pupils and present different sides of controversial issues. Being a relatively new teacher with an educational background in social economics is an example that emphasizes how a lack of competence leads to teachers' decisions to avoid certain issues. When teachers handled uncertain teaching situations created by controversial issues, they needed to draw upon their professional competence to make discretionary decisions on how to answer outbursts, how to present a topic, and what strategy would work best to create a constructive learning environment. This finding aligned with a prior study of how teachers used discretion to decide which pupils would receive special education in Norway: The teachers' educational background is understood as an important epistemic mechanism and being a professional with the proper background is central to discretional reasoning (Karseth & Møller, 2018).

Professional and personal values
In Norway, teachers are relatively free to choose how much time to spend on each topic and which teaching strategy to use in the classroom. Thus, professional competencies played a significant role when teachers chose a course of action in specific situations. However, when talking about their use of discretion, the teachers I interviewed also mentioned personal values as a resource for their discretionary judgments. This finding agreed with that of Zacka (2017), who suggests that being a street-level bureaucrat calls for people who can make normative considerations that pull in different directions because they are exposed to vague mandates, competing demands, and unforeseen dilemmas. In my interviews, the teachers addressed situations where they relied on values as a resource to make discretionary judgments. Espen, a Social Science teacher at Western High, explained: Some [decisions] are a professional assessment-what do we need to cover and how are you going to do it-the professional bit. When you choose topics, you need to do some evaluations: What is important when they will be examined? That must be covered. . .. During a discussion, I think that if something is unacceptable, then you just must feel it in your bones; I do not think this is okay. You also must think about the consequences it will have for fellow pupils.
Espen described how he based some discretionary decisions on what he felt was acceptable and other decisions on his professional competence. His reasons for choosing a course of action depended on the situation at hand.
At the same time, all the teachers were mindful about imposing their own values and beliefs. Vegard, a teacher at Eastern High 2, noted that it was not in a teacher's job description to produce political or religious clones. He tried to be conscious about separating his role in teaching Social Science curriculum from his personal values and opinions. That many teachers felt this way might reflect an ideal of "impartiality," as a basic principle of professional ethics, as well as for the exercise of public authority more generally in liberal democracies (Rothstein, 1998). In this sense, the norm of impartiality implies a limitation of the room for discretion and of moral agency. However, the teachers had different thoughts on how to ensure that pupils did not perceive them as partial. While some teachers were careful to keep their opinions private, others chose to be open about their own personal standpoints in various political and moral issues. Both sides argued that their choices benefitted the pupils.
The classroom context played a big part in whether teachers chose to draw upon values as a resource. My observations of the two Religion and Ethics classrooms and my interviews with those teachers showed a between-school difference in the strategies teachers applied to share (or not share) their own values. This possibly stemmed from pupil demographics. At Eastern High, it was difficult for most teachers to be neutral because there was no middle ground. There, pupils talked about their own or their parents' immigrant backgrounds, and religious beliefs were more visible via symbols (e.g., some girls wearing headscarves) and verbal expressions (e.g., pupils stressing their own religions). Pupils who are open about their values and beliefs demanded more openness from the teacher to create a good learning environment. At Western High, where most pupils had a majority background, both the teachers' and the pupils' values were implicitly understood as more similar than at Eastern High.
At Eastern High, Nina chose to be open about being an atheist (common in Norway). During my first day of observing her Religion and Ethics classroom, she asked the pupils to send her a letter about their religious beliefs (if they were comfortable doing so). Later, she told me that all the pupils in her classroom said they were religious-Muslim, Hindu, and (one) Christian-and that she had told the pupils she was atheist. In the interview, we discussed her views on sharing her lack of religious belief:

Do you tell the pupils you do not believe?
I told them that the first lesson.

Why? You answered it a little before, but-
To show them I can treat everything in the same way-neutral. I basically do not believe that one religion is more important than another. . .. Had I been a Christian, I would have said that, too. And, of course, I had told them that it was not going to affect me as a teacher. I am an atheist. I do not believe in God or any gods . . . [but] they will not get me to say that religion is nonsense. . .. I am just saying I don't believe. But [being an atheist] might also give me the strength to teach and highlight important aspects of religions and treat them equally. So, when a pupil expresses a statement or comment-saying that this sounds very strange in one religion-it is easier for me to say, "Yes, what do you think that [religious] person would say about your religion?" After all, they must familiarize themselves with different point of views.
Caroline, a teacher at Western High, taught a class in which most pupils had a majority background. Although she chose not to tell her pupils of her religious beliefs, the goal in her neutrality was the same as Nina's.
I do not think it really matters whether they know. They probably have guessed that I am not a Buddhist. I think they will get that I have a Christian background, so I guess they will understand that I have more knowledge about Christianity.

Is it because you do not want to influence your lessons?
The only reason I have not told them is that I want to teach [all religions] in the same way; that I do not prioritize one form of Christianity. I guess I would rather be looked at as a Religion and Ethics teacher than a representative for one religion.
Both Nina and Caroline argued that their motives for telling or not telling their religious belonging was concern for the pupils-so that the pupils would experience a neutral presentation of all religions. Nina chose to tell her pupils, predominantly Muslims and Hindus, that she was atheist. Caroline, a Christian in a classroom where all but three pupils had a majority background, did not announce her religious belief. She wanted the pupils to perceive her presentation as neutral but said they might guess her religion because she knew more about Christianity. Both teachers' reasonings stemmed from the same influence-their own values-but led to different discretionary judgments.

Relationship with pupils
I found that good social relations is a teacher resource for making discretionary judgments for two reasons. First, good relationships gave teachers valuable information about the pupils. They could then choose a teaching strategy based on their knowledge of the pupils or class dynamics. Teachers who had a relationship with the pupils knew which pupils struggled at home or had psychological issues that required their sensitivity. They also knew which pupils to push to be active during classroom discussions and which did not feel comfortable expressing opinions in public. Saetra (2020) argues that good social relationships constructed by the teacher and pupils is a core element to create a learning environment where controversial issues can be discussed. Second, good relations reduced the teachers' risk in trying new approaches. When the pupils and teachers had mutual trust based on good relationships, the pupils knew the teachers meant well. (2000) draw attention to relationships between street-level bureaucrats and their clients as a context for exercising discretion in what they call the citizen-agent narrative. They argue that street-level bureaucrats define their work in terms of relationships and not only rules and regulations. Teachers' everyday practices differ from other street-level bureaucrats' encounters because teachers meet with pupils regularly and for long periods. Thus, teachers may develop relationships with pupils that influence their discretionary judgment more than for street-level bureaucrats in other professions. Some of the interviewed Religion and Ethics and Social Science teachers taught the pupils in additional subjects; some were head teachers responsible for the class environment and individual development talks with all pupils each semester. Heidi at Eastern High claimed the pupils in her class-mostly immigrants or children of immigrants-were open to having a relationship with the teacher:

Maynard-Moody and Musheno
[It] gives teaching a new dimension. I know them very well; I know who they are. It is quite unique with Eastern High-you have a group of pupils that lets you get very close to them. Other pupils . . . have a lot of adults that care; parents, coaches, a team that is ready to help. Many of our pupils have a bunch of siblings and do not have that contact with their parents. And to be able to be there for a pupil, to be able to guide them for a few years-it is quite a cool thing to be able to take part in.
When observing Heidi's class, it was evident that she knew her pupils. She called them by name, made jokes, and challenged their views. However, not all the teachers had such close relationships with their pupils. For example, Anne taught Social Science at Western High for an average of 3 hours each week. I conducted my observations during the first semester, when she had known the pupils for only a month. With more than 30 pupils in the classroom, little space, and much noise, it was challenging for her to get to know the pupils. The lack of relationships made it harder for Anne to use discretionary judgment and to teach. She explained that she did not know which pupils to push, and which needed more space. Good relationships with pupils shaped the exercise of discretion not only because of having information, but also because a trusting relationship made teaching controversial issues less risky.

A toolkit for discretionary reasoning
In Religion and Ethics and Social Science, the topics that could be challenging to handle required teachers' discretionary reasoning. In the preceding sections, I illustrated how the teachers in my study drew upon different resources in their toolkits when making discretionary judgments. Looking at different resources as a discretionary toolkit was a useful analytical tool to understand the complex social landscape of discretionary judgments when teaching controversial issues. The following example from an observed Social Science lesson at Eastern High shows how one teacher used two resources in a particular teaching situation.

13
The topic of the lesson was causes of crime and covered both individual and structural explanations. The topic could be controversial because the pupils lived in an area where the crime rate was higher than in other parts of Oslo. In the first period, Heidi, the teacher, lectured on the topic and showed videoclips the class could discuss. The pupils seemed tired and unengaged. In the second period, Heidi assigned the pupils to "Vote with Your Feet." Heidi presented statements such as, "Is the age of criminal responsibility, 15, too low?", "Are you for the death penalty?", or "Do you think Norwegian prisons are too comfortable?" The pupils would walk to a corner of the classroom representing "Agree," "Disagree," or "I don't know." After choosing their standpoints, they debated. The pupils could move to a different corner if other pupils' arguments changed their views. Heidi challenged pupils on why they agreed or disagreed with the statements. The assignment created lively debate and more enthusiasm among these pupils than it had in the first period, when those pupils sat at their desks. In the first period, pupils had been passive; when the second period started and the pupils participated in the assignment "Vote with Your Feet," the entire room buzzed.
Teaching a controversial issue, such as reasons for criminality, could be understood as an unsettled situation, using Swindler's (1986) term, and required the teacher's discretionary judgment. Heidi based such a judgment on her professional competence and relationships with the pupils to choose a more dynamic assignment and awaken engagement. From her relationships with the pupils, she knew they could engage and discuss reasons for criminality, but she had to choose the right forum in which to do so. Her professional competence had taught her that assignments such as "Vote with Your Feet" were a good way to create debate and engage pupils. In a later interview, Heidi elaborated how she drew upon her professional competence and relationships with her pupils to make a discretionary judgment on what teaching strategy worked best: I always try to think that everyone learns differently. Some need something visual, something to look at; some need to hear something; some need to be provoked. So, I try to have a number of different elements in a teaching session to try to awaken as many as possible. I researched it myself in my master's thesis, that to participate in the classroom dialogue is very central for a learning outcome.
In this lesson, Heidi challenged her pupils' opinions but did not express her own views. The pupils presented contrasting opinions, so there was no need for Heidi to disturb that discussion by expressing her values. Something that could be perceived as the "right" opinion (because it came from the teacher) might limit pupils from voicing their own opinions. Several teachers explained that in other situations, if a group of pupils were onesided in an argument, then the teachers might challenge the pupils' standpoint with their own opinions or play devil's advocate. Heidi's example illustrated that to understand teachers' practices when teaching complex and controversial issues, one must look at available resources as a toolkit for teachers to draw upon when making discretionary judgments. What works in one context or situation may not work in another. Some teaching situations require a combination of all three resources-professional competence, professional and personal values, and relationships with pupils-and possibly others to make a good discretionary judgment, whereas other situations would demand only one resource.

Discussion and concluding remarks
In Norway, where the school system is one of the few institutions that meets almost all young citizens, it is important to understand what influences teachers' discretionary reasoning. This is particularly the case when dealing with potentially controversial issues in an increasingly diverse society. Using a framework drawn from street-level bureaucracy and theories of discretion (Lipsky, 2010;Molander, 2016), this article has delved into teachers' discretionary space and looked at how they draw upon different resources when several courses of action were available. I found that three forms of resources-teachers' professional competence, professional and personal values, and relationships with pupilsshaped the teachers' discretionary reasoning when teaching about controversial issues. I argue that teachers use this repertoire of resources as a discretionary toolkit that can be activated in different contexts.
The first resource to be highlighted was teachers' educational backgrounds, or lack of education, which influenced their feelings of competence and confidence when teaching controversial issues. The teachers perceived having the right competence as a resource for making discretionary judgments of if or how they would handle topics that might create debate or conflict. Karseth and Møller (2018) argue that teachers with formal educational backgrounds rely on that background when making discretionary judgments concerning special-education needs. I found that although teachers may have had proper formal qualifications, teaching controversial issues required competence that was more specific. Without sufficient professional competence on certain topics, teachers could not draw upon it as a resource to make discretionary judgments in challenging situations.
The second resource was the teachers' own professional and personal values. The interviewed teachers felt it important to expose pupils to different perspectives and beliefs, especially when teaching controversial issues, and the teachers' own values could be a resource in choosing if and how to present an issue. However, the two observed school contexts-where one was ethnically and religiously diverse while the other was far more homogeneous-differed in terms of how teachers talked about drawing on values as a resource. At the diverse Eastern High, flagging one's own beliefs was perceived as important in order to show that one did not have a hidden agenda, while at the homogeneous and wealthy Western High, the teachers' own beliefs were more often taken for granted as being neutral.
The third resource was the teachers' relationships with the pupils. This was a resource for discretionary judgment in two ways. First, information about pupils allowed teachers to judge what strategies worked with different pupils. Second, good relations made discussing controversial issues less risky because the pupils and teacher knew each other and thus allowed a larger space to say "wrong" things without assuming bad intentions. All interviewed teachers talked about good relationships as a resource. However, many taught pupils for only a few hours per week and did not have time to build relationships. For many teachers, this was a major challenge for developing trust and a classroom environment for discussing controversial issues in a constructive way.
Using the street-level bureaucracy perspective and discretion perspectives complemented the literature on controversial issues by unpacking what might influence teachers' discretion and what resources they drew upon in difficult teaching situations. This aided the understanding of discretionary reasoning by showing how the different resources that the teachers drew upon-competence, values, and relationships-led to different discretionary judgments, and how such discretionary reasoning depended on both the topic taught and its context. The present findings largely support previous studies in the field, such as Karseth and Møller (2018), who emphasize the role of professional competence, Taylor and Kelly (2006) and Zacka (2017), who focus on the values and morality of street-level bureaucrats, and Maynard-Moody and Musheno (2000) who view discretionary judgement as embedded in the relationship between citizens and the street-level bureaucrat. However, this article's main contribution lies in combining different understandings of what influences discretionary reasoning with Swidler's (1986) toolkit term, thus providing an analytical framework that captured teachers' complex everyday work. Looking at the different resources as a discretionary toolkit showed how these different entities informed and framed social interactions (Watkins-Hayes, 2009). This study had some limitations. I observed four classrooms in two Oslo schools and interviewed 16 teachers at three schools. However, although this relatively limited empirical material does not allow us to generalize findings directly, I argue that the framework developed for the present study is applicable for researchers looking at similar cases in other contexts. For example, future research could investigate forms of resources that other street-level bureaucrats and teachers in other contexts draw upon when making discretionary judgments.
In terms of implications for practice, I argue that if schools are to fulfil their democratic ideals by providing a conductive space where pupils with different backgrounds can engage with controversial issues, one needs to focus not only on teachers' competence, but also on facilitating trusting social relationships between teachers and pupils. This may help to provide a broad set of resources, which teachers may draw upon when using their discretionary reasoning in complex and demanding situations.