“She isn’t Someone I Associate with Pension”—a Vignette Study of Professional Reasoning

What drives frontline workers’ categorization of clients in rule-based settings with a large room for discretion? The literature on street-level bureaucracy offers a structural description of discretion that emphasizes working conditions, policy goals, personal preferences, client pressure and professional norms. However, in order to explain why frontline workers with the same room for discretion categorize clients differently, a theory of an epistemic understanding of discretion may contribute to this literature. Based on a vignette study of 24 interviews with Danish caseworkers, the analysis shows how professional reasoning, rules, and social stereotypes inform categorization and discretion. The findings indicate that caseworkers’ categorizations of clients are less responsive to clients’ needs and more sensitive to administrative reasoning when clients are associated with stereotypes of need. In addition, the analysis contributes to the theory of categorization and discretion in lower levels of government Keywords : Categorization, discretion, professionalism, stereotypes of need

This article studies what it is that drives frontline workers' discretion when they work in rule-bound settings. Do they bend the rules to keep intact what they see as their professional identity, or do they internalize administrative rules in ways that compromise how they think about themselves as professional frontline workers?
Discretion designates the space of reasoning used to transform a general rule into an assessment of a particular individual (Lipsky, 2010). According to the theory of street-level bureaucracy, the task of discretion poses an individual dilemma in public service delivery organizations, because frontline workers are cross-pressured by conflicting policy goals, professional standards, organizational goals, management and requirements of the target group with which they interact (Hupe & Hill, 2007;Lipsky, 2010;Maynard-Moody & Musheno, 2003;Winter & Nielsen, 2008). Furthermore, studies on discretion in lower-level government emphasize how "policy fidelity" can be challenged because frontline workers' working conditions alienate them from clients and potentially from the law, leading to lower responsiveness and quality in their decision-making (Tummers, 2012). The aim of these studies is to explain how these different sources of influence shape frontline worker's assessments of which clients should be granted what, when and why? So far, background factors, such as gender, experience, ethnicity, education and demography, have been given priority in studies of discretion, contributing to our knowledge about the impact of both structural and individual constraints and capacities on discretion. Adding to this knowledge, the present article examines the impact of "target group characteristics" on frontline workers' discretion of clients' needs, by using the theory of categorization to strengthen the study of discretion as a social process of reasoning about clients in rule-bound settings.

Theoretical framework: Discretion and categorization
In Lipsky's seminal work on the dilemmas of the individual in public services (2010), discretion is described as the space left over for frontline workers' as "a matter of some subjectivity" within existing rules and regulations. According to Lipsky (2010), frontline workers are expected to exercise and be accountable for their discretion as law-abiding bureaucrats, but sometimes also as knowledge-based professionals.
Among scholars, there is agreement that room for discretion is both desirable and inevitable, due to the simple fact that it is impossible to anticipate, and thus regulate, all the possible events and circumstances a client might experience (Evans, 2011;Høybye-Mortensen, 2015;Lipsky, 2010;Maynard-Moody & Musheno, 2000;Meyers & Vorsanger, 2012). Dworkin (1978) has described discretion as being "like the hole in a donut [that] does not exist except as an area left open by a surrounding belt of restrictions". His idea of discretion is that it is always subject to evaluation against rules. Grimen and Molander (2008) distinguish between structural and epistemic discretion and thereby develop this structural understanding of discretion as www.professionsandprofessionalism.com Page 3 "a room of maneuver". They argue that discretion is not in contrast to structure, but has both a structural and an epistemic dimension. Rules, knowledge, and social conventions are always part of a discretionary practice. They define discretion as a mode of reasoning (see also Molander, 2013) and emphasize that discretion is always expressed as reasoning, regardless of the "size" of the room of discretion. Following a similar line of thought, Wageenar (2004) points out that general rules should not be perceived as simple guidelines but work as sources of abstract thinking that must be converted into meaningful actions. According to this view, discretion is something both highly improvisational and creative. Frontline workers have to make things up as they go along, and they "make things up" by drawing on repertoires available in the situation, i.e. they draw on different sources of social, professional, and rulebased categories to describe and make sense of what they encounter at the moment (Wageenar, 2004). Maynard-Moody & Musheno (2000) also see discretion as both epistemic and practical. However, in contrast to Dworkin (1978) and Grimen and Molander (2008), who see rules as the constituents of discretion, they associate quality discretion with rule-bending, as, they argue, there is an inherent conflict between client responsiveness (client agency) and rule-abidance (state agency) (Maynard-Moody & Musheno, 2003). However, even though this distinction between citizen agency and state agency has provided some interesting empirical insights, primarily from the US context, the distinction blurs the third option of seeing rule-following as a way of being fair. Rules can also be a means to protect citizens from bias in individual encounters with the state, as with caseworkers in the Danish context being studied here, for instance. Therefore, the question of whether rules are used to protect clients from bias in encounters or to protect caseworkers from clients' requests will always be, first and foremost, an empirical question, which should, and will be analyzed, in the particular context.
As initially pointed out by Lipsky (2010), frontline workers are pressured by conflicting requirements from policy rules, their clients' needs, their professional codes, their personal values and sometimes by being over-committed to their job (Dahler-Larsen & Pihl-Thingvad, 2014;Tummers, 2012). As a result, frontline workers regularly experience stress when delivering public services to clients, and consequently, struggle to cope with their working conditions. In a systematic review of coping strategy studies, Tummers et al. (2015) identify three predominant ways of coping: 1) moving towards clients (the use of personal resources), 2) moving away from clients (rationing resources) and 3) going against clients (rigid rule following). Frontline workers lack control over the demand for and supply of public service, which means they need to cope to survive experienced work pressure.
Hence, coping strategies constitute one way of examining how "clients' needs" put pressure on frontline workers, though this is clearly not the only way. Categorization is another way of investigating how frontline workers respond, act and reflect about clients.

Categorization and its sources of influence
The concept of categorization has to do with how the final process of policy delivery orders clients into separate categories with distinct political rights. Stone (2002) argues that this social process is rarely unambiguous and that a situational assessment is needed to determine who belongs to a certain category and where to draw the line between the various categories. Even the way we construe an apparently clear-cut category, such as "age", depends on how we perceive at least two social groups: the young and the old. Or to take another example: a child suffering from ADHD can be classified as disabled or as a product of bad parenting, resulting in two very different sets of rights with regard to welfare support. According to Stone (2002), such a categorization depends on the way in which the process of categorization is organized, www.professionsandprofessionalism.com Page 4 but also on the particular group of comparison. Is the child being compared to other disabled children, for example, children with learning disabilities, or to naughty children in general? It is the basis for comparison rather than the specific case that determines the outcome?
Previous studies of how people "put a fix on other people" state that identity, shared norms, and feelings are strong sources of influence on categorization (Murphy-Berman, Cukur, & Berman, 2002;Van Dijk, Ouwerkerk, Goslinga, Nieweg, & Gallucci, 2006;Weiner, 1995). Other studies point to dissociation and aversion as sources for assessment of other people's values, motives and worthiness (Møller & Stone, 2013). In terms of how these associations play out in a rule-bound context, such as frontline work, it is a matter of studying how the target group is socially constructed and interpreted by society at large, and how these layers of symbolic meaning influence the policy design implemented by frontline workers (Kallio & Kouvo, 2015;Schneider & Ingram, 1993). This source of influence is symbolic in the sense that it draws more on social stereotypes than concrete experiences. When people categorize other people from a symbolic informed context, they compare their appearance, actions, and perceived values to abstract perceptions (Jenkins, 2000;Lamont, 1992;Skeggs, 2005).
Regarding identifying caseworkers' professionalism as another source of influence, Schott, van Kleef and Noordegraaf's (2016) distinction between "organizational professionalism" and "occupational professionalism" is used. Hybrid professionalism includes fields of "education," "social work," and "policing" even though they when compared to the "purified" forms of professionalism such as medicine and law, lack the same degree of occupational content and institutional control. As was originally pointed out by Freidson (2001), these professionals hold the key to better and more accountable control of public service delivery. They represent a "third logic," overcoming both consumerism and bureaucracy in legitimate and effective ways to standardize working procedures and motivate frontline responsiveness (Freidson, 2001). In "hybrid professionalism," professionalism is seen as a coproduct of both occupational and organizational principles and values (Evetts, 1999;Schott et al., 2016). "Occupational professionalism" refers to a bottom-up approach to professionalism, whereas "organizational professionalism" describes a top-down approach to professional control. Here, the source of influence is not occupational criteria, but the interest of the organization as defined by the manager (see also the study of lower level management by Evans, 2010).

Area of study, methodology, and material collection
In 2001, the Danish Government reformed the social insurance system. The aim was to enhance the cooperation between the labor market and public administration by redefining the aim, the means, and the target groups of social insurance (Ministry of Social Affairs, 2001). The letter of the law states that the policy intention is to "contribute to an efficient labor market" (Ministry of Employment, 2008). This emphasizes the client's right and obligations to contribute to the society. As a part of the Active Labor Market Policy (aktiv arbejdsmarkedspolitik) reform, a range of decision-making tools aimed at supporting caseworkers' discretion of needs were implemented. The mandatory use of the work capacity method (arbejdsevnemetode) includes a resource-profiling (ressourceprofilering), where caseworkers collect specific information about clients regarding social, cultural, economic, and health resources (Ministry of Social Affairs, 2001).
Caseworkers working under the active labor market policy exercise professional skills that include both social and legislative dimensions (Ministry of Social Affairs, 2001). This means that caseworkers use their discretion in accordance with legal standards. Caseworkers are expected to assess clients objectively, leaving subjective www.professionsandprofessionalism.com Page 5 attitudes aside. In the carrying out of their work, the caseworkers must make sure that the discretions made in the casework are not due to personal attitudes and values, but to professional explanations and frames of understanding, as well as reflected experience from practice (Ministry of Social Affairs, 2001). A professional assessment of a client's capacity to work is hence defined as: A professional discretion is a caseworker's assessment of the information and documentation collected to evaluate the client's resources in relation to the labor market. The assessment is not casual. It is based on the information provided by the client and the collaborator and is to be analyzed and compared with the caseworker's social professional knowledge and experience from practice (Ministry of Social Affairs, 2001).
In Denmark, active labor market policy is designed and implemented in line with many other welfare states, such as Holland, France, Sweden, and the UK, emphasizing national control with local unemployment management through process and case regulation (Bonoli, 2010). However, in contrast, to at least the UK and Sweden, Danish frontline workers are generally professionally trained caseworkers with both occupational and organizational expertise (Baadsgaard, Jørgensn, Nørup, & Olesen, 2014). Danish caseworkers working under active labor market policy constitute a relatively homogeneous group (the length of their education varies from one year to three and a half years), which allows for more systematic qualitative studies of how they draw on their professional background when categorizing clients. In addition, this policy program includes a wide range of decision-making tools intended to support (influence) discretion. This provides the opportunity to study how caseworkers use decision-making tools in their discretion of citizens' needs, as well as their categorization of clients. In addition, caseworkers, working under active labor market policy have a considerable amount of client contact, which makes eye-to-eye categorization and discretion part of their daily work routine.

Selection of informants
A theoretical sampling strategy was used to select municipalities and interviewees from job centers (Weiss, 1994). The empirical basis is 24 interviews with caseworkers from all areas of Denmark. To ensure socio-demographic diversity, caseworkers were selected from municipalities with more than 50.000 inhabitants. Furthermore, they were selected based on the criterion of task similarity. All the caseworkers administer the work capacity method (arbejdsevnemetode) and exercise the discretion of unemployed clients with disability issues in need of support from the state. The individual characteristics of caseworkers, in terms of, for instance, work experience, gender, and educational background, were not part of the selection criteria. Rather, they were classified according to these characteristics afterward, as part of the analysis (Table 1). The group of interviewees consists of both men and women, caseworkers with extensive and limited work experience and caseworkers trained as social workers and as social counselors, which are two comparable education programs that focus on equipping caseworkers with social pedagogical, administrative, and economic knowledge. The most common caseworker among the 24 interviewees is a woman with more than ten years of work experience who is trained as a social counselor. However, the differences between them suggest that patterns between categorization, discretion, and social stereotypes identified in the material are unlikely to be explainable by such individuals' characteristics.

Material collection: Semi-structured interviews and vignettes
The data were collected as qualitative interviews in 2007. The interview guide was structured in three main sections. The first section contains questions about work routines, use of clarification tools, and professional identity; the second part includes questions about specific cases and use of clarification tools, and the third part has questions about the interviewees' private-collective orientations. The specific cases were constructed as vignette cases resembling particular client problems of theoretical relevance. Three vignettes resembling stereotypes of need were constructed. The diagnostic content of the vignettes was developed with the help from a medical doctor in the field of psychiatry and pain (Dr. Med. Lise Gormsen, The Pain Clinic, Aarhus University Hospital). Three comparable pain conditions with different diagnostic profiles were selected and used to describe the health problems of the fictive cases. This ensured that any differences in the discretion of needs could not be www.professionsandprofessionalism.com Page 7 explained by objective differences in pain profile. To ensure variation in the symbolic context informing the fictive cases in the vignettes, a medical sociologist was consulted (Professor Peter Conrad, Department of Sociology, Brandeis University). All background information was standardized, and the vignettes were randomly assigned to interviewees as follows: The respondents were split into two groups, and half of them were presented to vignette A and C and the other half to vignettes B and C. Even though interviewees were theoretically selected, the vignette-combinations were randomly assigned by organizing envelopes containing vignettes A and C and vignettes B and C respectively in advance of the interview. Vignettes were designed as narratives about contested and non-contested stereotypes of need. Two vignettes were designed as stereotypes with a positive and a negative reputation, respectively, expressed as the difference between a contested and a noncontested pain condition. The third vignette was designed as a social stereotype leaning toward a mix of negative and positive reputation. The contested diagnosis selected was fibromyalgia (vignette A), and the non-contested diagnosis selected was multiple sclerosis (vignette B), and the control case that could go either way (vignette C) described phantom pains from the loss of an arm (see Table 2). This provided the option of comparing how the symbolic meaning of these diagnoses influenced discretions and categorizations, and how interviewees reasoned about their casework when addressing the fictive vignette cases. Table 2 Differences between salient features in the three vignettes Vignette A Imagine a 34-year-old woman with fibromyalgia. She is married and has two children living at home. She has been on sick leave for six months from her job as a social and health care assistant, mainly because of chronic pain in her joints and muscles. She wishes to apply for an early retirement pension because she does not see herself as being capable of doing her job properly. She now uses support bandages almost all the time, and she has tried all kinds of treatments without getting any better. In addition to her pain, she has trouble sleeping, along with memory and concentration problems. Her situation now is that if she goes to work or does housework, she ends up in bed for several days.
B Imagine a 34-year-old woman with multiple sclerosis (MS). She is married and has two children living at home. She has been on sick leave for six months from her job as a social and health care assistant, mainly because of chronic pain in her joints and muscles. She wishes to apply for an early retirement pension because she does not see herself as being capable of doing her job properly. She now uses a wheelchair almost all the time, and she has tried all kinds of treatments without getting any better. In addition to her pain, she has trouble sleeping, along with memory and concentration problems. Her situation now is that if she goes to work or does housework, she ends up in bed for several days.
C Imagine a 35-year-old woman, who lost an arm in a traffic accident. She is married and has three children living at home. She has been on sick leave since the accident 1½ years ago from her job as a childcare worker, mainly because of chronic pain in her back and head, as well as severe phantom pain in her missing arm. She wishes to apply for an early retirement pension because of her handicap. Since the accident, she no longer sees herself as being capable of doing her job properly, since she generally has a lot of trouble just trying to handle the extra pain and extra difficulties in her everyday routines stemming from her loss of an arm. In addition to her pain, she has trouble sleeping, along with memory and concentration problems. This case selection applies a Most Similar System Design (Landman, 2008). All respondents encounter the same target group, work under similar work conditions in similar organizations (job centers), and have a similar educational background. This means that any patterns in the material are less likely to be due to these identical or similar factors. The use of vignettes ensures similarity in objective client characteristics, such as gender, education, marital status, and variation in symbolic context by exposing the interviewees systematically to different stereotypes of need. In addition, the vignettes provide the opportunity to talk about specific subjects and not least to compare their responses across the material (Ejrnaes & Monrad, 2012).
There are obvious limitations to a vignette study, as they force caseworkers to reason about and categorize fictive clients. A way of addressing this has been to ask questions about their own experience with clients, in order to ascertain whether their reasoning differs and how it differs between real client cases and fictive cases.

Data analysis
Data analysis was carried out as a combination of inductive and deductive coding of interviews to ensure saturation and avoid forcing data into predefined categories (Charmaz, 2006;Gibbs, 2007;Glaser, 1998). Finally, and as the basis of the analysis, all material was combed according to a closed code list developed from the initial coding (Lofland, 2006). Data was systematically coded using both within-case and cross-case analysis, to gain an understanding of internal causality in each interview, as were patterns and correspondences across interviews (Miles, Huberman, & Saldaña, 2014). In the following analysis, both condensed analysis and in-depth interpretive analysis are presented. All quotes have been translated from Danish to English, and interpretations and coding summaries have been inter-reliability tested.

Professionals' reasoning about clients' needs
The first analysis of the material examines how the caseworkers reason about their clients in their everyday work. The analysis is based on the entire interview material, and it will be made clear whether quotes are responses to questions about the fictive vignette cases or caseworkers' own experience.

Administrative reasoning
A significant characteristic of caseworkers' administrative reasoning is the use of references to administrative categories, that is, to match groups or target groups. An extract from interview 16, in which the caseworker predominately used administrative reasoning, is presented below. Here the caseworker explains her general approach to clients: Basically, we're the promoters of a system, where we say: 'You have these options to choose between.' Obviously, we have to be able to present this in a decent way. And then there's the possibility that you don't choose. And if you don't choose, then I'll choose for you … because this is what I'm hired to do. (Interview  16) The quote illustrates a high degree of accountability to "the system," as well as a commitment to management.
The caseworker perceives the willingness of clients to internalize the values of assuming responsibility for oneself and active citizenship based on a desire to be responsible as a precondition for her practicing of what Evetts (1999) describes as organizational professionalism. The caseworker appears to be more concerned about upholding the rules than occupational criteria for exploring the client's reasons for requesting assistance.
She presents the client with a number of concrete activation opportunities and evaluation strategies. In doing this, she expresses both loyalty with regard to upholding the intention of the policy as well as loyalty to the concrete rules. However, she does not mention any occupational perspectives on the client's capacity to work. This is interpreted as an example of a caseworker who identifies more with being part of a professional organization than with the role of being an occupational professional.
The following quote provides an example from another interview of how administrative reasoning entails identification with the existing rules, instead of with the particular problems of the client. Here the caseworker responds to the vignette A describing a woman with fibromyalgia: She wants an early retirement pension … but she will be refused. No [laughing], it's just because it's so difficult today to get a pension. So, even before we get to the clarification. Of course, she must know the criteria for early retirement pension. And then we need to find out if she can return as a healthcare assistant. (Interview 9) The above quote is selected to illustrate how accountability sometimes comes before client responsiveness when caseworkers draw on organizational professionalism to reason about clients' needs. Here, the caseworker chooses to initiate the interaction with the client by clarifying the criteria for granting an early retirement pension instead of starting by clarifying the extent of the problem. The following quote provides a similar example of rule-based reasoning. The caseworker responds to a question about how they deal with chronic pain patients in general and not to the specific contested vignette case in question. She sees herself as administering a set of rules instead of utilizing a set of tools for client-assistance: Especially the people who have been here for many years and don't understand that the rules have changed. Now you have to [work]. A lot of them have received social welfare without ever showing their faces at the municipality [office] because they didn't have to. Now, they have to come … at least every three months, right? … So it's difficult to make them realize. 'So, this is the way things are now. The rules are different.' (Interview 11) This caseworker emphasizes how clients first and foremost have to adapt to the new rules in order to continue to be eligible for social welfare, rather than this being based on their capacity to work. This client approach also demonstrates an example of coping with working conditions by "moving away from clients" and hiding behind rules instead of using them to be client-responsive (as described in Tummers et al., 2015).
As exemplified by the quote above, there were often hints of resistance against clients when caseworkers used an administrative reasoning about casework. The reason seems to be that when a caseworker draws on professional principles concerning how well (s)he follows the letter of the law and the principles of management and organization, (s)he then sees any case in which (s)he has to deviate from management policy as an attack on her professional integrity. (S)he copes by moving away from or going against clients to avoid what she interprets as bending the rules in favor of client responsiveness. This seems to cause a lot of frustration towards clients, who are perceived as not showing the required level of cooperation and motivation to get better. In this sense, the more ardently the caseworker endeavors to follow the rules, the less room there is for exercising comprehensive discretion. The following analysis illustrates an example of categorization, where administrative reasoning shapes a curtailed discretion of clients. The quote is from a caseworker responding to a question about what she imagines she would do in the case of a www.professionsandprofessionalism.com Page 10 woman with fibromyalgia described in vignette A. The quote exemplifies how the rules for documentation reduce discretion as a room of maneuver for the caseworker, though not the act of actually making an assessment. The first line is also the quote used in the title of the article.
[S]he isn't someone I associate with a pension in any way. [Interviewer: Why not?] Simply because, well… exactly because there has to be something medical. There must be something medical documenting that she really can't move herself-her arms or her legs at all. In other words, really not capable of doing anything. And I simply just don't think this is the case. There must be something she's capable of doing. Yes. (Interview 24) If you follow the letter of the law, there are no exceptions as to when treatments and work testing should stop, because the law states that all options must be exhausted before a supportive effort is even considered. In practice, there will always be another job and another treatment to try. The initiation of supportive efforts, therefore, depends on the caseworker's discretion as to when "enough is enough." Determining when that is, however, is far from an objective, clear-cut decision. On the contrary, and as explained by Stone's (2002) notion of categorization as an ambiguous process, there are different interpretations of where this boundary lies, and one of these differences seems to be reflected in different symbolic contexts, as we will see in the next analysis.
It appears that when caseworkers draw on the administrative reasoning in a categorization, they let "common sense knowledge" influence their discretion regarding when to make exceptions from the rule. The categorization thus becomes a matter of identification with clients based on personal feelings and commonalities regarding social stereotypes rather than a comprehensive discretion of the specific case. In other words, the material suggests that when caseworkers primarily draw on organizational professionalism and use administrative reasoning they lack principles for a responsive client approach, and discretion as to when "enough is enough" is based on what associations they have with the particular client. What ultimately becomes the defining source of influence in discretion is the symbolic context of the client rather than a professional occupational principle.
When clients are categorized based on administrative reasoning, caseworkers tend to associate the client with a stereotype even before the assessment of ability to work begins. This exemplifies a curtailed discretion because clients are perceived as not meeting the criteria for assistance beforehand. Moreover, those who do not meet such criteria are typically regarded as the main target group of a more strict evaluation compared to a more lenient evaluation, namely those believed mainly to have an attitude problem. This contrast between strict and lenient evaluation is the center of attention in the next analysis.

Comprehensive reasoning
As opposed to the administrative reasoning about casework, some caseworkers draw on more occupational knowledge as a source of reasoning. One important aspect was when clients' actions and motives were interpreted without adding an administrative layer to the discretion. However, the strongest signifier of occupational reasoning was the tendency to use an assistance-based principle to justify the actions taken in relation to clients. The following quote from a caseworker responding to one of the opening questions about her main work tasks illustrates this: "They may have a hard time returning to something similar. And then you have to help them get started with something else" (Interview 14). In contrast to caseworkers' administrative reasoning, she refers to "help" instead of to a "rule." These differences in types of reasoning resonate with Grimen and Molander's (2008) emphasis on discretion as also epistemic and never just fixed by the structure. Even though some caseworkers insist on understanding their discretion as fixed by rules, the differences between administrative and comprehensive reasoning demonstrate why this is not the case.
The question of who is perceived as being responsible for solving "the situation" is another aspect that distinguishes organizational from occupational professionalism (Evetts, 1999). In the former, the tendency for the caseworker was to adapt to a discourse of "self-responsibility" embedded in decision-making. Within this discursive framing of social problems, the client was perceived as being responsible for solving the problem of disability and unemployment with the exception of highly specific situations in which clients have a detailed diagnosis and prognosis for the development of their disability (e.g. terminal cancer patients or patients with specified recovery plans). Caseworkers drawing on occupational principles about social work appear to hold an expanded view of who and what can be responsible for "solving the case," in the sense of helping a client return to the labor market or apply for appropriate social insurance. They see the client from a bottom-up perspective in the sense that they prioritize "treating" the client before complying with management goals, as described both by Schott et al. (2016), and also in Maynard-Moody and Musheno's description of what constitutes the difference between state-agency and citizen-agency (2003).
Resistance was also a significant trait of occupational reasoning, which is when a caseworker expressed resistance against the formal rule of social welfare. The resistance against the rule was against the purpose of assessing only labor-related aspects of clients. This kind of resistance is not surprising because ignoring the client's non-labor-related conditions beforehand contradicts the perception embedded in social work professionalism. One caseworker, when describing his work in general terms in relation to one of the opening questions, expresses his resistance as follows: Or they'll get social problems precisely of the disease.… So it's hard to stay away from, for example, guiding advice about relationships and everything else and economy without that we need to talk about their cash benefits.… However, financial problems, family problems, audit issues … housing problems.… So the property is indeed essential for all people to hold jobs.… So, it's hard to get around. (Interview 15) Here, the caseworker expresses a comprehensive reasoning regarding what (s)he believes influences the ability of the client to perform in the labor market. Moreover, the quote illustrates a categorization where the functional reasons for unemployment are examined before any conclusion about eligibility for social insurance is made.
In general, the pattern in the material shows how occupational reasoning leads to comprehensive categorization. In the following quote, a caseworker presents his reasoning when asked to compare the non-contested vignette describing a woman with Multiple Sclerosis (vignette B) and the vignette portraying a woman who has lost her arm in a car accident (vignette C): [M]any women-or some women-if they're involved in a divorce or something like that, where the woman becomes a single provider with one, two or three children, then obviously it matters that you're a single mother with three children … in relation to what she thinks she can manage.… So, we don't approach it so concretely and say: 'Well, you have three children, so you can't manage.' … We do actually have the same requirements. But still, you can have an understanding of this because you have three children, and you're alone with them, and then you have plenty to do ... including on the home front, but also in your free time.… Under such conditions, I think this could matter. (Interview 6) For this caseworker, the relationship between professional reasoning and discretion is not a matter of making an exception from the general rules for some deserving www.professionsandprofessionalism.com Page 12 clients as much as it is a question of treating individuals in relation to their problems, in order to be able to have the same requirements. In this case, (s)he is bound to the greater purpose of the rule and not to its technocratic dimensions that characterize caseworkers' administrative reasoning.
Moreover, this way of reasoning about having the same requirements, while simultaneously extending individual consideration, also defines what can be interpreted as "comprehensive categorization", which the caseworker demonstrates by stating that: "Under such conditions, I think this could matter." Here the client's "other" conditions become part of the categorization though these are usually excluded from administrative reasoning. "Conditions" are perceived as being circumstances that remove focus from the labor-directed effort, which causes the caseworker to focus on the client's problems instead of "rigid rule-following." Another caseworker explains, with reference to her own client experience how she justifies making an extra effort even though it contradicts the exact letter-but, according to her, not the spirit-of the law: I had someone with urinary tract problems. She had had this inflammatory condition in her body for almost a year and a half. Well, in her case, we know that she'll recover and that it won't be permanent.… But she's simply dead tired after this year and a half, so her body can't cope with me saying 'Go sign up for unemployment benefit'… Instead, I say: 'Well, let's do some easy training to get back into the labor market,' because after such a long sick period, not because I think it's permanent because she'll get better.… So, this is not a permanent case. Therefore, I also described how her employability is not permanently reduced but is reduced right now due to her long-term sickness. So, a case does not have to be completely stationary before we make an effort. (Interview 14) This quote provides an example of how the caseworker follows the "spirit" of the law rather than following it literally. In the quote, there is no trace of resistance against the spirit of the law. On the contrary, the caseworker uses the law as a tool to empower the client to re-enter the labor market, though does not follow the terms of it literally. This exemplifies how categorizing clients comprehensively draws on reasoning grounded in occupational professionalism about social work, as also emphasized by Evetts (1999). In other words, the crucial justification in this type of argument is based on a long-term perspective of satisfying the intention of the policy instead of a short-term perspective. This is crucial because, in the short-term perspective, the exemption from, for example, the "duration rule" is considered rulebending, while in a long-term perspective, the exception work is regarded as a precondition for actually empowering a client to get back into the labor market.
In general, the analysis shows how caseworkers think professionally about their discretions. However, they do so in very different ways. Some caseworkers curtail discretion by reducing what Grimen and Molander (2008) refer to as the epistemic dimension of discretion to rigid rule following while others exercise discretion based on comprehensive assessments of needs. When caseworkers explained their categorization during the interview, they tended to prefer either to reason according to management and the principles of their organization or according to occupational knowledge about social work. In the case of the former, they used a technocratic and administrative reasoning about clients being more accountable "upwards" than "downwards" than in the case of the latter. Here, they also moved towards the clients using personal resources of engagement and responsiveness, whereas they moved away from, or against, clients to align with management and organizational professionalism. Table 3 shows that all 24 caseworkers use professional reasoning in their categorization of clients and that many use both organizational and occupational reasoning. However, even though both types of professional reasoning appeared in almost all the interviews, the dominance of these varied, organizational professionalism being the most dominant. Here clients were categorized according to administrative categories, and caseworkers primarily see themselves as gatekeepers of "the system." In contrast, caseworkers who reasoned according to occupational principles of social work were more inclined to categorize clients according to which economic and social resources conditioned clients' general well-being, and they perceived themselves more as the citizens' advocates than promoters of a set of administrative rules. In total 413 24* * All 24 cases were coded according to professional reasoning.
In the following, the analysis will focus on the caseworkers' responses to the fictive cases presented as stereotypes of needs. The analysis seeks to specify how the symbolic context of clients shapes how caseworkers reason about clients, and whether this influences their discretion of needs.

Caseworkers' use of clarification tools and discretion of needs
To specify caseworkers' categorization and discretion of clients' needs, the analysis focuses on their use of mandatory tools to clarify clients' ability to work. First, the range of tools used by caseworkers will be presented, along with the analysis of the purposes that caseworkers use as arguments for choosing to use a particular tool.
Although it is mandatory for the caseworkers to use the work capacity method (arbejdsevnemetode), they can choose among a range of clarification tools. A clarification tool is an activity or treatment that can help the caseworker to assess clients' need of assistance. In the following, the distribution of their uses of clarification tools is presented, ordered according to a systematic interpretation of whether the tool is used with a "hard" or "soft" purpose. A soft use of a tool corresponds to a positive response to the client's request for a pension, whereas a hard use of a tool corresponds to a negative response. In the case of the former, the tool is used to document the request's validity and in the latter to question it. As Table 4 shows, caseworkers may use the same tools, but with different purposes. When a caseworker uses a tool for a "hard" purpose, the tool is used to question a client's request for a pension. Here the tool becomes part of the caseworker's strategy to cope with a client's claims regarding needs. In general, the coding of what purposes caseworkers use the various tools for shows that "hard" use often corresponds to an attempt to hide behind rules to protect the caseworker from more client contact than desired.
In contrast to what I classify as a "hard" use, some caseworkers use tools, even the same tools, in a "soft" manner. A tool being used in a "soft manner" refers to use in which the purpose is to demonstrate how much a client is in need of service, that is, here the tool is used to document a lack of ability to work rather than to question the client's claim of inability to work. Based on the coding of "soft use," the material shows that when caseworkers use tools "softly" they bend the rules by using them strategically to document, for instance, no work capacity. This resembles the difference in coping strategies identified by Tummers et al. (2015) between going against and moving towards clients. In situations where caseworkers use tools with a "soft" purpose, they also move towards clients and their claims by drawing on personal resources, such as empathy.
One issue is the notion that caseworkers reason in particular ways when they talk about their experiences, another issue is whether the flavor and the tendency of professional reasoning are influenced by the symbolic context of the client encountered. In response to the contested stereotype of need (vignette A), all caseworkers preferred a hard use of clarification tools, whereas they preferred a soft use of the noncontested stereotype (vignette B). This pattern supports the expectation that symbolic context matters in the discretion of needs because all cases were comparable on all other dimensions than the reputation of the diagnosis of fibromyalgia and Multiple Sclerosis, respectively. In relation to their preferred use of clarification tools towards the vignette describing phantom pain from a lost arm, an interesting pattern appears, namely that this preference is primed by the first vignette. In Table, five quotes from the typical relationships are displayed. In response to the question of what she would do with a client portrayed through a non-contested stereotype, the caseworker from interview 7 narrows it down to being "a diagnosis, which can give a pension." In contrast, in relation to the one-armed woman portrayed in vignette C, the same caseworker states that: "you don't get a pension for a one-armed disorder." Even though both phantom pains and multiple sclerosis involve considerable pain, clarification tools are used very differently. In the case of the former, tools are used for the purpose of documenting the client's inability to work, and in the latter situation, the caseworker uses the tools to question the client's service needs. These patterns characterize 24 out of the 48 vignette responses (each interview contains www.professionsandprofessionalism.com Page 16 two vignette cases). However, the other majority (8 out of 11) came to a different conclusion regarding this client. They emphasized the loss of the arm as something that goes beyond the loss of the actual arm, as one caseworker puts it: "She has lost a lot, so she may end up receiving a pension" (Interview 11). This group all responded in this way after explaining why they preferred a hard use of clarification in relation to the contested stereotype of need (A), for instance in saying "but she will be refused. No [laughing], it's just because it's so difficult today to get a pension. So, even before we get to the clarification" (Interview 9).
So, in terms of examining the effect of symbolic context, the reactions show how socially shared stereotypes of need wander into professional discretion of needs.
As was argued for in the theoretical framework, discretion can be analyzed as an outcome of a social process of categorization, which highlights how caseworkers are influenced not only by rules and professional context but also by client characteristics. The analysis indicates that reasoning and categorization may be conditioned by symbolic context. This points to an impact of symbolic categories, as the vignettes used were comparative functionally but different symbolically. Client characteristics function as a source of influence, the flavor of which depends on which symbolic context the caseworker associated with a particular client.
Caseworkers who responded to the contested stereotype of need (vignette A portraying fibromyalgia) were inclined to cope by going against the client. They preferred using an administrative reasoning to "hide behind rules" to protect them from clients' claims about being in need of service. The dominant way of using clarification tools were "hard," as they used them to question clients' service needs. Caseworkers who responded to a non-contested stereotype (vignette B portraying multiple sclerosis) were inclined to cope by moving towards the client. They preferred using comprehensive reasoning to follow the spirit, rather than the letter, of the law. The dominant way of using clarification tools was "soft", as they used them to document discretion of needs. Finally, caseworkers who responded to vignette C were inclined to move towards or go against the client, depending on whether they were primed by a contested or a non-contested stereotype. Their interpretation of a case involving the loss of one arm followed the opposite pattern of the priming vignette, which indicates how the symbolic context matters even in cases where the client in question does not represent a typical stereotype.

Conclusions
I find that caseworkers' discretion of clients' needs is systematically related to categorization practice, either as comprehensive practice-as in reflective and empirically based categorization-or as rule-bound categorization practice, as in routinized and stereotyped categorization. Even though the interviewed caseworkers share educational status and administrative tasks, and are part of comparable organizations and demographic surroundings, the data show that some are more driven towards stereotyped categorization practices than others. The analysis finds that the stereotypes of needs associated with client characteristics contribute important insights into this relationship.
One implication of the analysis can be related to the theory of "hybrid professionalism" (Schott et al., 2016). Almost all caseworkers used both organizational and occupational reasoning in the interviews, and this supports the tenet that frontline professionalism is indeed different from purified professionalism. The analysis adds to this theory by linking curtailed discretion to organizational professionalism, and client-responsive discretion to occupational professionalism.
A second implication of the analysis confirms classic studies of categorization (Jenkins, 2000;Lamont & Molnar, 2002;Stone, 2002;Yanow, 2003). They add knowledge about the significance of categorizing in a rule-bound setting and indicate www.professionsandprofessionalism.com Page 17 why curtailed discretion is associated with client characteristic loaded with contested stereotypes of need.
A third implication of the analysis is that even when the symbolic context is weak, clients are compared to other strong stereotypes, which primes the categorization and discretion. This ties in with the general wisdom in street-level bureaucracy, where discretion is described as inevitable and clients themselves as a source of influence on discretion (Lipsky, 2010).
Stereotypes work through over-determined associations of either a positive or negative nature and hence always trump the particular experience and specific evaluation of the client. This way of assessing contradicts fundamental principles of equal access to treatment in the political system, as well as an objective evaluation of-in this case-the client's ability to work. Many clients' complaints and reasons for unemployment are fuzzy, and their personal reasons for seeking assistance hard to measure. The potential consequences of being ascribed such negative values may give the client quite a different course through the system compared to the client that is associated with a positive stereotype. Seen in the light of the empirical findings presented here, stereotypes of needs affect both the use of clarification tools and the kind of professionalism that caseworkers draw on when categorizing clients' needs.
In sum, studies of discretion should not ignore its social process, but directly address the impact of categorization on discretion and its sources of influence, in order to qualify core dynamics at the frontline. This article finds that the symbolic context of clients' shapes how frontline workers reason about their discretion.
There are obvious limitations to this study, related to the constructed setup of using vignettes to control the influence of client characteristics. These force caseworkers to reason and categorize on 'unnatural' terms. This means that conclusions should not be made beyond analytically similar situations. This calls for further studies, perhaps applying a more quantifiable design across policy programs to examine the extent to which the conditional influence of symbolic context applies beyond the policy program of active labor market policy, in a larger empirical setting and outside of Denmark.