Two sides of the coin – Domestic violence survivors’ expectations of financial support and social workers’ expectations of survivors within the social assistance system

This article is based on an interview study with thirteen survivors of domestic violence and ten social workers within the social assistance system. This article aims to understand how applications for social assistance from survivors are handled between the discretionary powers of the social workers, the organization’s fixed categories and the survivors’ need for support. Theories of street-level bureaucracy and human service organizations was used in the analysis. The results show that social workers state that they treat survivors with respect and generosity. Survivors said they wanted to meet a committed social worker, which was not the case for many of the survivors interviewed. Categorizing the ‘right’ kind of survivor is of great significance for being eligible to social assistance.


Introduction
Survivors of domestic violence who leave violent partners are very vulnerable financially, emotionally and psychologically. This is because of reduced capacity to work, security issues, or psychological and physical injuries caused by violence (Sanders, Weaver, and Schnabel 2007;Lindhorst, Oxford, and Gillmore 2007;Anderberg and Rainer 2013;Haeseler 2013a;Roschelle 2008), but also because violent men often use economic means to control women and these women may therefore lack knowledge of and access to economic resources (Lindhorst, Oxford, and Gillmore 2007;Lundgren et al. 2001;Adams et al. 2012). Survivors often have financial difficulties for a long time after a separation (Stylianou et al. 2013;Trygged, Hedlund, and Kåreholt 2013). For a separation to be successful, women require both financial security and emotional support (Haeseler 2013b;Enander 2008). Postmus et al. (2012), demonstrate in a study that the professionals who encounter abused women often prioritize psychosocial and legal support while the survivors instead want more tangible support in the form of housing and work. People who have survived domestic violence are referred to as survivors, an established way of referring to those who have survived what can be a deadly situation (e.g. Stylianou et al. 2013;Postmus et al. 2012;Kelly, Burton, and Regan 1996). The term does not imply that the survivors are out of danger after the separation.
Earlier research on domestic violence has focused financial support systems (e.g. Roschelle 2008;Hetling 2011;Adams et al. 2012). Aspects such as the outcome of the support system, description of the categories of women within the system or how to identify abused women are often the focus of such studies. However, policies are implemented within the power relations between service users Backgroundsocial assistance in the swedish welfare state Social assistance has a special function and organizational context in a universal welfare state. National and universal welfare systems mainly meet the needs of people who are or have been active in the labour market when they get sick, injured, unemployed or when they need financial support such as child allowance, old age or disability pension (Harsløf and Ulmestig 2013;Hvinden and Johansson 2007). Thus social assistance in Sweden functions as a final safety net for the most marginalized in society, which increases stigmatization in comparison to countries where all citizens share the same duties and rights within a single system for all citizens Researchers have highlighted how an old 'poor-law' discourse still influences the social assistance system in Sweden, categorizing people as either 'deserving' or 'undeserving' (e.g. Panican and Ulmestig 2016; Köhler, Thorén, and Ulmestig 2008;Thorén 2008). However, Sweden is a country with high legitimacy for welfare and the universal welfare system also ensures that all citizens are offered, for example, free child care and schooling. It is in that context this study should be understood. The social assistance system is organized into 290 municipal welfare offices and is regulated through the Social Services Act (2001:453), which provides a broad legislative framework but few precise instructions. Therefore, both the social workers processing applications and the welfare office exercise considerable discretionary powers with regard to legislation. There are significant variations between municipalities in all aspects of how legislation is interpreted (Köhler, Thorén, and Ulmestig 2008;Thorén 2008) and also, to some degree, between different social workers in the same municipality (Byberg 2002;Stranz 2007). Social assistance applications are processed by social workers with a university degree, predominately in social work. In addition to addressing financial issues, they are also supposed to provide counselling.
Since 2007, social services are required by the Social Services Act (2001:453, 5:11) to show special consideration toward victims of crimes, with particular attention to survivors (Ljungwald 2011). The legislation does not require the welfare office to have special programmes or support systems for survivors. Many welfare offices have a small unit or social worker with specialist knowledge of domestic violence but this support is often more psychological than financial (see Skillmark et al. 2017;Lundberg 2017).

Theoretical framework
The municipal welfare offices can be understood as human service organizations whose main task is to determine whether an applicant can be allocated to one of its fixed categories of people that is the organization's 'raw material' (Hasenfeld 1999). If not, the office can reject the application or reframe the issues to match a fixed category. Lipsky (2010, 60) wrote: 'Clients seek service and benefits; street-level bureaucrats seek control over the process of providing them.' Day-to-day social work often becomes a matter of standardizing human problems into fixed categories that existed before the individual's application (see Lindhorst, Casey, and Meyers 2010). This raises questions about who is a survivor and for how long.
The Social Services Act highlights individual assessments of needs, yet categorization is the opposite of individual assessment. On the other hand, without the fixed category of survivors, their special needs and expectations would be defined solely at the discretion of individual social workers. Finding a balance between an efficient way of processing individuals and being able to meet individual needs is important for legitimacy of the organization (Gillingham 2015;Hasenfeld 2000). This balance often needs to be managed by street-level bureaucrats (Lipsky 2010;Brodkin and Marston 2013). The fixed categories also enable people to obtain suitable services and fulfil their expectations in line with the knowledge and moral principles of the human service organization (see Hetling 2011). Yet it is often hard to measure success or failure and human service organizations, for their part, are measured by how they fulfil the society's norms and expectations (Hasenfeld 2000;Hasenfeld, Ghose, and Larson 2004); they become 'moral entrepreneurs' (Hasenfeld 2000(Hasenfeld , 2010. Fulfilling the expectations of the survivors becomes a social process dependent both on social workers' discretionary powers and on the welfare office's fixed categories like 'survivor' or 'crime victim', where the organization has expectations of survivors applying for social assistance (e.g. Hetling 2011;Laakso and Drevdahl 2006). In this study, social workers are understood as streetlevel staff. The analysis does not make clear-cut distinctions between the organization and the individual social workers because the empirical material doesn't allow it. What the street-level staff does is, in an analytical sense, what the organization does.

Method
The analysis was carried out in five steps, using reflective dialogue between the researcher, available research, theory, and the empirical evidence, as suggested by Alvesson and Kärreman (2012). We began with 19 in-depth interviews with survivors (see Lucas 2014;Irvine 2011).
Two of the survivors were contacted by the researcher via newspaper advertisements; 17 were selected via women shelters where staff was requested to ask survivors they were in contact whether they would like to be contacted by researcher.
The interviews lasted approximately one hour. They are part of a larger research project focusing on the financial aspects of domestic violence, funded by the Swedish Crime Victim Compensation and Support Authority. It was not a selection criterion, but all the women who participated had been abused by a male intimate partner whom they had left. Most of the interviews were carried out in women's shelters and others in the women's homes. For the 17 survivors in shelters, approximately 18 months had at average elapsed since the separation from violent partners (between 1 month and 7 years) and 16 of them had children. The interviews focused on how financial issues had affected them during and after the separation, about economic abuse, what financial security meant for them and whether they were able to achieve this. For further description on the sample see Table 1. The interview themes were based on earlier research on the financial situation of survivors and influences upon this (e.g. Hetling 2011;Lindhorst, Oxford, and Gillmore 2007;Adams et al. 2012;Stylianou et al. 2013). From the 19 interviewed survivors, four had received social assistance during the separation, nine were receiving at the time of the interview and six were not receiving any social assistance and were thus excluded from further analysis. The first thematic analysis of these interviews was done by reflective dialogue and it became obvious that the social assistance system was important. The reflective dialogue approach revealed complexities in our perceptions and interpretations of the interview material via different understandings and meanings in the interviews, earlier research, theory, and in the researcher's interpretations (Alvesson, Hardy and Harley 2008;Alvesson 2003). The use of reflective dialogue made evident the importance of mutual expectations and how difficult it is.
The second step was to make an interview guide for social workers from the same municipalities as the survivors. This was based on themes from the survivors (offensive treatment, important persons, and unfulfilled expectations), supplemented with questions and two vignettes (that lack direct relevance for this study and will therefore not be analysed here) on the significance of background of the survivors for the application process. We carried out eight interviews with ten social workers. All were women with a university degree in social work with one exception, a social worker with a university degree in behavioural sciences. They had worked between 18 months and 20 years with social assistance. Approximately half worked with applications from new recipients and half with long-term recipients. They were chosen by the researchers because they had responded positively to requests for interview. The interviews were held in their offices. In the third step, the author of this article took a comparative standpoint (between the survivors and the social workers) in the reflective dialogue. All interviews were recorded, transcribed and coded after empirical themes regarding the mutual expectations. One theme was the effort the social workers felt they made to ensure everyone felt well-treated. The survivors described how they were sometimes treated well and sometimes disrespectfully. The survivors' and social workers' descriptions were related but the theme of being well treated was understood in different ways. Earlier research on the social assistance system were used to develop the analysis taking the special context in to consideration.. There is always a risk that a researcher codes quotations that support their understanding of a specific theme. The reflective dialogue led to different understandings of the interviews, through earlier research and theory, and aimed to avoid predetermined answers.
In the fourth step, the themes were retested against the theory of street-level bureaucracy (Lipsky 2010) and human service organizations (Hasenfeld 2010). The results of the analysis were then tested against earlier research on social assistance in Sweden (e.g. Van Aerschout 2011; Thorén 2008).
In the fifth step, the analysis results were then retested against the empirical material and, when this supported the analysis results, these were presented in the text with the theory and earlier research.
The study was based on the ethical principles for the humanities and social sciences stipulated by the Swedish Research Council (Swedish Research Council 2002). The research was approved by the Linköping regional ethics board for research involving people (No 2012/396/31). We gave written information to the informants about the research project before the interview, including ethical reflections on risks involved for survivors like memories and experiences of violence being revived. We informed the interviewees that they could end the interview at any time with no need for explanation. The interviews with survivors were conducted by a female researcher with experience and knowledge of working with survivors, and who could provide further support or refer survivors to a local support agency. Some of the survivors' and social workers' quotations and descriptions are deliberately vague and in some cases descriptions have been removed to ensure that individuals and geographical places cannot be identified. (2004), Haeseler (2013a) and Sanders and Schnabel (2006) stress the importance of survivors who leave and do not return to violent men being able to support themselves. This is also reflected in this study. For women on a low income, supporting themselves and their children often depends on mean-tested social assistance. It is therefore very important to understand how social assistance applications from survivors are dealt with, given the discretionary powers of social workers, the organization's fixed categories and the survivors' need for support.

Social workers treating survivors well
The social workers described how difficult it was to process applications from survivors. Survivors are a prioritized category in social assistance, according to descriptions by social workers as well as national guidelines and legislation, which affect social workers' discretionary powers. One social worker said: 'First, these kinds of cases often take a lot more time than others. But it is also different. For example, before we grant financial support to other service users, we require them to submit a lot of documents from the bank etc. We completely ignore this at first here, until you know . . . it's calmed down a bit.' It may seem like a contradiction, but using the discretionary powers to make exceptions to bureaucratic routines may take more time than following the routines, for example when the exception involves waiting and building a relationship. One social worker, asked whether being a survivor changes anything in the processing of social assistance, answered: 'It makes it easier for us, that we are more . . . what can I say, you shouldn't say generous, but then we don't have to get so much information, so it will be easier for us if we are to manage it. Because we know that we're going to help here. ' Responding to a question about specific issues when working with survivors, one of the social workers said: 'They're pretty, what should I say, difficult to work with. Not in a negative way, but they are usually very . . . You don't really know where you are with them. It's not about being dishonest with the system in any way, but it's all about the relationship with this man, as it usually is [. . .] That they are unreliable in that you never know if it is a definite break or if they will get together again.' Another social worker described the process many survivors go through when leaving a violent partner: 'You always feel frustration, but it's actually your own frustration you feel when they go back.' Despite the difficulties, the social workers say that they show respect and try to use their discretionary powers to help survivors.
It is difficult for social workers to work with applicants who are survivors, but these applications are easier to process within the organization, and this affects the treatment of the survivors. Some of the difficulties inherent in a human service organization like the social services can be bypassed for survivors, due to the moral consensus that this is a very vulnerable group in society. Social assistance processing is described by the social workers as more generous and flexible compared to other categories of applicants and that they can use their discretionary powers. A social worker said: 'We have guidelines for housing costs for all applicants, but when a female victim of violence, for example, has to leave a shelter then there is a greater tolerance for higher housing costs, so that you can grant the costs.' This special effort in favour of survivors is supported by guidelines and legislation but also by the strong political and moral consensus about supporting victims of crimes.
This is an example of how survivors receive greater discretionary assistance compared other social assistance applicants. This is directly connected to the treatment of crime victims (Peter 2006;Ljungwald 2011); since 2007, according to the Social Services Act (2001:453), social workers are obligated to show special concern to them. It is reasonable to assume that these changes to the Social Services Act have affected what social workers expect of survivors.
Individual means-tested assessment supports the discretionary powers of social workers. One said: '. . . we can really make any assessments because we have an individual means test.' For individual assessments to be considered legitimate, the social service organizations need to approve and give individual social workers the discretionary power they need. These individualized means-tested assessments are used by the social service organization to legitimize decisions concerning almost all aspects of social assistance applications. Human service organizations must adhere to available resources and existing structures, yet individuals are far too complex to be processed in an identical way within a social services organization (Hasenfeld 2010). It is reasonable to presume that organizations are affected by the morality connected to the discourse of domestic violence and its great political import.
There are, however, limits to the social workers' discretionary powers, for example, the difference between formal discretionary powers and how these are applied. Even if the social workers sympathize with service users, the organization normally limits the number of exceptions allowed. Service users are under obligation to try and become economically independent and this seems to apply less to survivors than other applicants. Survivors are a privileged category due to this willingness and ability to make exceptions to the requirement to provide data to support decision-making, like bank statements, rent payments or other bills. Social workers do not normally start processing applications if service users have not provided these documents. One social worker said it was not always possible for survivors to provide the required documentation. 'So, based on her story about the urgent assessment, you trust the service user.' This moral choice is explicitly supported by the organization and legislation.
Lipsky describes discretionary powers as a relative concept and argues that 'The most common structural device for managing the consequence of routines is emergencies. ' (2010,136). Emergencies solve major service dilemmas for street-level bureaucrats. One social worker said: 'First you solve what is urgent, and then you handle the rest.' Events judged to be emergencies are very convenient for the organization and, of course, for service users whose needs are defined as urgent. This can solve problems when dealing with survivors, since action can be taken that is otherwise hard in a bureaucratic organization.
Another limitation of social workers' discretionary powers is how management and colleagues, through support and monitoring, affect decisions made by social workers. One social worker spoke about a difficult case: 'I would not decide this for myself, although I have the delegated authority for decision-making about social assistance. This case would be discussed at every team meeting.' This can be understood as a limitation of discretionary powers. Another social worker says that she wished that her colleagues dared to use their discretionary powers and make more decisions on their own. She said: 'You are scared. You reject the application because you went to the boss and the boss has told you to reject it. But actually you have full authority to decide for yourself anyway.' Social workers are in an ambivalent position within social service organizations. Processing social assistance applications is difficult, not only because the organization limits their discretionary powers, but also because of survivors' expectations.
Being treated wellmeeting an engaged social worker who supports survivors in a respectful way The survivors interviewed described how they expected to be given financial support because they knew that it was important to become independent. One survivor said: 'The financial situation is very important, that it works, because life doesn't work without it. It was important for the separation in the sense that I know I have something. That I don't need him. I don't need him to decide what food to buy, or if I'm allowed to eat that day, or . . . there are others that can . . . help me.' Many of the survivors have been controlled by their ex-partners trough financial matters.
In the study, we noted that high expectations are placed on the social assistance system to safeguard survivors' independence. Survivors may not receive everything they applied for, but noone said that social assistance had not covered their most basic needs.
There are several descriptions of how the role of social workers is very important, over and above ensuring financial support. This is often exemplified in terms of 'doing that little extra' and being generous. One example was a social worker who immediately took the survivor to a shelter and accepted that she would not be able to provide the answers normally required for approval of a social assistance application for a long time. The social worker used her discretionary powers to fulfil expectations of fair treatment and to build trust with the survivor who said: 'I called the social worker every time I needed something and that felt good. She didn't make me tell her everything, because you don't know who to trust.' This is one of several examples where the social worker supported the survivors with needs that were not financial. Another survivor described her social worker as follows: 'She was there for me. I am so grateful that I had that social worker.' These descriptions, however, are only one side of the coin. It is more common to hear accounts of how survivors are met with disrespect. Discretionary powers can also be used to not make exceptions or not fulfil expectations. With regard to her experience as a victim of male violence, one survivor said: 'I have a lot of respect for men. I dare not say what I think and feel. When I called and it was a man and he was very . . . what can I say? Mean. But as if he . . . well, he was mean when I spoke with him and he had a really bad attitude.' Another survivor said: 'I have spoken to my social worker here in municipality X. I have never met her, only talked on the phone. I have never spoken to such an unpleasant social worker in my entire life. She shouldn't even be working as a social worker.' The interviewer asked: 'Can you describe what an unpleasant social worker is like?' The reply was: 'She asked a lot, about how my life was and I explained. It was then, when she said that about the rent, "Now you have to move out," and I said "Where am I supposed to live?" She replied, "I don't care at all where you stay afterwards [after the women's shelter]." [. . .] She was really rude and impolite.' This is an example where the social worker did not fulfil the survivor's expectations of receiving financial support and being treated well. According to the survivor's description, the social worker chose being rude and unfriendly and to not use her discretionary power.
Social workers work within a system that stigmatizes the most marginalized people in Swedish society (Köhler, Thorén, and Ulmestig 2008;Van Aerschot 2011). Social workers have great discretionary powers but also scarce resources and high workloads (Thorén 2008;Van Aerschot 2011). This can make it difficult to fulfil survivors' expectations. From the perspective of the individual survivors, some welfare offices are better at dealing with this than others. One survivor said: 'Your luck depends on the social worker you get at the beginning. If there is a social worker who has her heart in the right place, then that's fine.' Most survivors described differences between social workers, and that there may also be variations between municipalities. This was supported by one of the survivors who said: 'If I hadn't had them [the social workers from her home town], then I would have been in the same situation as most of the other women here in municipality X. It feels a bit like, up here nobody wants to pay. You need a woman's shelter, but the municipality doesn't think you need it because of the cost.' This describes variations between the municipality where she used to live and where she now lives.
The survivors said they wanted to be treated as people with individual needs, which was not always the case. There were several descriptions from survivors that deviated from the social workers' descriptions of how survivors are treated by the organization. Yet, the first step in being treated with relative respect and having the service individualized is to be categorized as a survivor and as the 'right' kind of survivor in the eyes of the welfare office.

Categorizing survivors with the 'right' problem that matches the human services organization
The organization views social workers using their discretionary powers to make exceptions being legitimate if the applicant can be categorized as a survivor or, even better, as the 'right' kind of survivor. To be eligible for the relative financial security as a survivor within the social assistance system, survivors are expected by social workers to act in an appropriate way in order to be assigned to the right category. In a discussion of how social workers can convince survivors to make a police report One social worker said: 'It feels like everything gets a little bit easier if they report it. Then the police conduct an investigation. Hopefully the man will be sentenced if he is guilty and then she has not done anything wrong, or how should I put it.' Interviewer: 'When you say "a little bit easier", is this in the sense of the well-being of the woman or about yourself as a social worker?' Social worker: 'Both, I think.' Being categorized as someone that 'has not done anything wrong' makes the case easier to manage, and if the court found her ex-partner guilty, this was one way of ensuring this. It became a 'symbol' of 'doing the best they can,' and of being treated as a survivor. When making the police report, the applicant became a survivor in the social worker's eyes, in the sense that she was trustworthy because she had left the abusive partner and was not someone that would go back to him or try to receive support without really wanting to leave him. Making such an assessment and deciding whether women are serious about leaving partners is a difficult task. A great deal depends on the survivor fulfilling the social worker's expectations and if the survivor can prove that she is a victim of violence, she has better chance of being categorized as a survivor. One example was a social worker who said: 'I'm thinking that if a woman came into the welfare office for the fourth time in a year and said "Now I have left him again," then you could perhaps begin to thinkis it just that you need extra money and that you will go back to him again after two days?' Decisions, based on judging whether they are telling the truth, and survivors' stories, are necessarily very complex.
One could ask whether it is even possible to assess the authenticity of reported violence and the survivor's desire to leave her partner. Assessments about lack of motivation can also be internalized by survivors (see Lipsky 2010;Järvinen and Mik-Meyer 2003). Lipsky (2010, 153) argues: 'If the service user is to blame, street-level bureaucrats are shielded from having to confront their own failure or the failure of the agencies for which they work.' The applicant must not only be a survivor but also preferably be the 'right' kind of survivor according to the categories of the organization. This can be exemplified by one social worker who said, in a discussion about a middle-class survivor that deviated from more typical situations and that they had to bring in an accountant to look at her financial situation: 'Yes, it's not so common, and when you have such a case, it's not . . . usually you have to ask for help from an economist or the like.' She is not the kind of survivor that the organization is accustomed or organized to dealing with. Another social worker said: ' I must say that it is, it's a very complicated case . . .
[laughter]. That's not an easy one you gave me there. But as I said, the case has to be looked at. If for example it's pointed out by the National Board of Health and Welfare, the Social Services Act, that you can ignore her husband's responsibility to support his family and the duties of the wife when there has been violence in a relationship. She lives in a shelter, and therefore there can be special reasons to take into account.' The social worker reframed the situation of the middle-class woman as being a survivor living in a shelter and thus the organization was able to help her. The social worker also referred to the Social Services Act and the special treatment of survivors. However, it is quite possible that she used legislation to support her assessment rather than it directly affecting her decision (see Lipsky 2010). But in both ways the law would have helped the survivor to get support.
Furthermore, being categorized as a survivor does not mean staying in this category forever; there is a 'best-by date' for being categorized as the 'right' kind of survivor. The category is temporary and time-dependent and can change, develop and disappear because of changes in society and the organization (see Hasenfeld 1999Hasenfeld , 2010.
A social worker said: 'Yes, absolutely, then maybe you don't [make exceptions] for ever. You might set up a deadline when there has to be a division of property. They must make a decision on how to go ahead.' The social workers were aware that survivors often attempt to leave partners several times, but also that it was difficult to decide whether to believe the survivor or not.
The interviewer asked: 'Is it relevant to the case if it is the fourth time the woman has left her partner?' The social worker answered: 'Actually it should not be, one would expect. It could be as urgent and the women are just as exposed anyway. But then, it's hard to say. For it depends on whether it affects what really happens orwhat to say?if it's real or not. There are many here who have some form of mental illness, and there are delusions and things that may not quite match up and it becomes a cry for help. Attention-seeking and stuff. Then perhaps it's more about this kind of assessment you have to make.' The needs of the 'wrong' kind of survivors are less likely to be met, and this is not necessarily a consequence of less sympathy for survivors but rather the lack of suitable techniques (Lipsky 2010). The organization needs to process applications into a fixed category that it can manage in an effective and legitimate way (Hasenfeld 2010;Lipsky 2010). Legitimacy for the social assistance system means that no person obtains support if they have other sources of income or if they own something they can sell. The desire to help survivors can conflict with the need to attain legitimacy for the social assistance system (see Hasenfeld 1999).
Being categorized not only as poor, which is the organization's appropriate category, but also as a survivor, and, preferably as the 'right' kind, provides access to resources (see Hasenfeld 2000). However, categories are not stable and survivors are easier to manage if they can prove themselves deserving.
Being categorized: advantages of being the 'right' kind of survivor There were several examples of survivors describing how they did not fit into the 'right' category in order to receive support. One survivor who had lived in Sweden for many years said that she had needed support to pay the application fee for Swedish citizenship, which would have allowed her to move abroad and get away from her violent ex-partner and still be able to move back to Sweden. Her application was declined because this cost was not a responsibility of the welfare office. She said: 'Applications for citizenship cost 1500 SEK [approximately 150 euro] and they thought that I could save or find the money somewhere else. I promised them that I had really tried. I certainly don't have anyone else who could give me money, otherwise I would never in my life have called them because I feel like an idiot every time I call or submit an application there.' Regardless of how 'deserving' a service user is, the poor-law logic of the social assistance system makes it impossible to make financial support a right (see Van Aerschot 2011). Not even when such a small amount, that also would save money for the municipality because she moves abroad, is easy to grant. It is hard for human service organizations to adapt support to the needs of the applicants and therefore individuals often have to adapt to the kind of support they are offered (Hasenfeld 2010).
The organization wants to work with the 'right' problems and people. The case of a survivor with protected personal data, due to threats to her life, challenged bureaucratic routines regarding job applications: 'I feel like they don't take it seriously. For example, you can't really go and leave your CV with photo when you have protected personal data. The cover letter and CV can end up anywhere and it's not quite right. They may not have experienced anything like this, so they don't really understand. ' As argued previously, being categorized as a survivor makes the system somewhat more individualized and generous, but the system is greatly influenced by means testing, the powerlessness of service users and the 'poor law' tradition. Not even survivors with high legal priority within the organization can completely escape the 'poor law' logic of the social assistance system.
However, according to the descriptions of the survivors in our study, there are large advantages to being categorized as a survivor, even though categories are instable and are continually being renegotiated. One of the survivors, when asked about her expectations of the welfare office meeting her needs, answered: 'I did not have a positive view of the social services before receiving social assistance. The people I have had contact with after I left him have been good. It feels like they help me more now [after leaving her abusive partner].' She is worried that coming to a new municipality could mean that she will not be viewed as a survivor in the first instance but as a single mother who cannot support herself, which may affect the service.
Leaving a violent relationship is a process (Enander 2008;McLaren 2013). One woman said: 'I don't know how many times I left and went back.' This makes it difficult for the welfare office, as a human service organization, to process the application and fit the woman into the category of survivor, which would make it easier to offer her support. When survivors leave and return repeatedly, they describe how this affects the possibility of being viewed as 'deserving'. One survivor said: 'I felt it the second time too. When I had problems after that, it felt like [. . .] they didn't quite help me the same way as the first time, and I also believe that it's common that you don't get the same sort of help a second time when you go back. [. . .] And then I was threatened a lot by my ex-husband and then I went back again, and now it's the third time. Then quite a lot happened. There was my daughter who has been exposed to a lot of sexual and all other kinds of violence.' When the violence was directed towards her daughter, the welfare office and other authorities acted more decisively and the woman was offered more services. The violence towards the daughter changed her position and it became more important for the organization to take action than in the case of a woman with a history of failed attempts to leave a violent partner. It was impossible for the social workers not to act to prevent the sexual abuse of children, regardless of the survivor's 'best-by date'.

Concluding discussion
This article started in a mystery (Alvesson and Kärreman 2012) about the way applications for social assistance from survivors are processed using the discretionary powers of social workers, the human service organization's fixed categories and the survivors' need for support. The reflective dialogue between the researcher, available research, theory and the empirical evidence, as suggested by Alvesson and Kärreman (2012) unfolded a complex power game far removed from policy statements about service provision for survivors. According to Alvesson and Kärreman (2012), a reflective dialogue cannot be fully described in an objective and rational manner. The core of this analysis is that adaption and negotiation of results come from repeating the reflective dialogue and its different possible interpretations. At the end of the process, the researcher must make informed choices about how to present results, which can only be understood within methodological limitations and it should also take into consideration that another researcher could have made other choices about which empirical patterns or results to prioritize in the analysis.
There is a difference between the way social workers describe how they fulfil expectations and how the survivors themselves describe this. The social workers in this study described how the survivors often are categorized as being 'deserving'. As such, and, in the best-case scenario, as a survivor with urgent needs, being treated with relative flexibility and generosity. The social workers, on one hand, describe how they fulfil these expectations. Yet, on the other hand, they are working with survivors within the confines of a social service organization and its poor law history.
Survivors of domestic violence are highly prioritized in legislation as well as in public discourse in Sweden (Ljungwald 2011). Yet different sections of the welfare systems have different functions. The social assistance system in Sweden has a special position regarding universal welfare state, providing support only for the most marginalized in society, and this is therefore more stigmatizing than in countries with the same systems for financial support for all citizens. The dominant opinion is that the state should only spend tax revenue on financial support if this is essential and should only be allocated to those lacking all other options; the incentive for social workers is to reduce costs and the period of benefit support (Purvin 2007). The social assistance system is characterized by more emphasis towards duties and less towards rights than other financial support systems (Thorén 2008;Van Aerschot 2011). However this competes with the provision to take care of crime victims, and especially victims of domestic violence. Individual assessments provide social workers with a discourse and legal framework where they can be flexible, 'bend the rules' and take the time required to support survivors (see Laakso and Drevdahl 2006). Human service organizations have always had to adapt to these discourses (see Hasenfeld 2000;Lipsky 2010). The prioritized everyday goal of the organization is to find a balance (see Lindhorst, Casey, and Meyers 2010), but this is difficult. Control of service users, for example, verifying whether they have resources of their own or deviate from societal norms, has always been an element of poor relief (Geremek 1994). No one applying for social assistance in Sweden can avoid such investigations, survivor or not. Simmel (Simmel and Jacobson 1965) argued more than a hundred years ago, that all systems of poor relief, aim to discipline people using individualized assessments. Survivors receive greater support from social assistance applications than most other applicants. Within this context of powerlessness, stigmatization and control they become, relatively speaking, the VIPs of the poor law logic. The system is slightly more flexible and generous towards the 'deserving' poor, but still does not fulfil the expectations of the service users.
This study is not without its limitations. There are not a great number of interviews and these were carried out in a small number of municipalities in southern Sweden. Caution is required before any conclusions are generalized to other settings. Future research in other municipalities and other countries could serve to confirm or refine the results. Results could be then used to develop policies that assure financial security for survivors. Recommendations for social workers could also be drawn from the results.
The study shows, within its limitations, that it is important to discuss expectations with survivors, to build relationships that support survivors when they are trying to leave violent partners. The way that social workers define survivors is less relevant here. It is equally important to recognize that survivors' difficulties can last a long time after leaving a violent partner and that going back to them can be a part of the process of a final separation.