The Policy Level
At the policy level documents are collected from the different discussions related to social work and in particular to issues concerning unemployment in the Nordic countries. We ask questions such as the following: To what extent is social work confined within steering documents? Who are the dominant actors in this field? We believe that the data may reveal many levels of conflict between the social worker and the steering documents, a state that makes it is important to trace these significant documents. The principal methodological approach is based on Norman Fairclough's model for critical discourse analysis (Fairclough 1992), in which discourse is about how language is used, written and spoken as a social practice. In order to be able to identify the actors behind and the contextual construction of dimensions with political and ideological characteristics we need analyses of text and context. Special attention is paid to how problems are defined, represented and legitimated and how solutions are described and justified. There is an ambition to understand how positions are constituted through discourses and how the subject is created and shaped by the practices (Nicoll and Fejes 2008, 14-15).
Reflecting on Real-life Cases
On the basis of real cases the challenges of young unemployed people are discussed with teams of frontline practitioners, including both social workers and other professionals. The social workers select critical cases and reflect on the cases with their co-workers in the practice unit. The methodological approach is based on the Reflective Mirror Model (Yliruka 2011). The theoretical basis of the model includes ideas on reflective evaluation, which emphasizes that knowledge arises from action and exists for action, and that knowledge is tested in real-life situations. The model requires mutual trust between team members and an interest in common knowledge production. The objective of the reflective discussions is to enable the social workers and the team to analyze the core of their work and their primary tasks while they reflect on the client situation and the operating environment. This can also be a tool for handling the complexity and multilayered nature of social work (Yliruka and Karvinen-Niinikoski 2013).
This mirror model provides a comparative frame for analysing cases in different countries. It involves the use of forms designed to support the documentation of the professional's progress in client work. The analyses of the cases are performed in national teams in real-life welfare settings where the research team also participates in the reflective discussions. Users are involved in the study, either as participants in the reflective sessions or through interviews. These reflective sessions also provided the basis for a learning network where practices, emerging practices, organizational issues and theories are discussed.
Tuning the Theoretical Perspectives
Surrounding the daily activities of social work, the perspectives indicate areas of tension between the different levels. The practice as a whole contains discursive, organizational and individual levels, which indicates the significance of distinguishing between different domains when studying organizations (Danermark and Kullberg 1999). The identification of the various domains implies that events and activities that at first glance would appear to affect only one professional group's way of defining and explaining a problem in reality could prove to affect other issues within other domains. Individual persons, authorities and organizations contribute actively in shaping the phenomena we study in different societies (Emirbayer and Mische 1998; Hacking 1999).
We emphasize the importance of looking at social work practice as it is shaped by context and structural conditions as well as how it shapes itself and its surrounding contexts. A number of theoretical perspectives at different levels are addressed in the study, including the sociology of structure and agency (Emirbayer and Mische 1998), theories of human services organizations (Hasenfeld  2009), street-level bureaucracy (Lipsky 1980), public encounters (Goodsell 1981), theories of neo-institutionalism (Ahrne and Hedström 1999) and domain theory
(Layder 1993; 2006). In the design and analytical framing, the project draws on theories of practice research (Miettinen, Samra-Fredericks and Yanow 2009; Marthinsen and Julkunen 2011; Julkunen 2011; Uggerhöj 2011), institutional interaction (Drew and Heritage 1992; Sarangi and Roberts 1999), actor-network theory (Latour 2007) and critical discourse analysis (Fairclough 1992; 1995; 2003).
Even though these theoretical perspectives seem massive, we use them as a loose methodological and theoretical frame so that it can provide space for emergent methodological and theoretical perspectives. Pickering (1995, 21) uses the term tuning as a perceptive metaphor when researchers tentatively construct their actions in research. Research processes in emerging practices requires that there is room for counter-arguments during the research process. We set out with the understanding that this form of tuning will occur throughout the multilevel and context-sensitive process.
Cross-national comparison is an important tool for welfare research, but too often important international differences within the organization of welfare services and in the relationship between welfare practices and experiences are not considered. In a comparative perspective it is still common to approach the issue from a separate perspective or examine the issue at a single level, such as from the point of view of the users, the profession, the organization or the general policy. A more comprehensive perspective is needed.
Still, the context is by no means restricted to a local area, but could concern even internationally distributed activities, which creates a contextual robustness that is very challenging. When we avoid abandoning the comparative approach and instead it enrich it with the dialogic processes of knowledge production, we might allow more sensitivity to the contexts in order to overcome the dissociation between the subjects of welfare and the expectations of objectivity in the prevailing ideas of evidence for policy-making.