Being Professional and Being Human. Professional’s Sensemaking in the Context of Close and Frequent Interactions with Citizens

Conceptions and critiques of the assumption of functional specificity

The literature on professions and professionalism points to a fundamental asymmetry between professionals and citizens. For both Parsons (1939, 1954a, 1954b) and Freidson (2010), this asymmetry is structural in nature, meaning that it is given by the difference in formal knowledge and expertise inherent in the different roles as either “professional” or “client.” Further, the asymmetry is intimately linked to what Parsons calls the functional specificity of the relation (Parsons, 1939, p. 460; Parsons, 1951/1991), p. 305).

This means that even though the professional-citizen relation is personal, in the sense of two people meeting and interacting with each other, it is also rational, universal and specific, in the sense that all professionals are expected to treat citizen-clients “the same,” based solely on information relevant to the specific problem in question and not on personal characteristics (Parsons, 1939, p. 462). As mentioned above, it is this functional specificity, meaning that it is the focusing of the relationship to problems that can be understood qua the formal knowledge and expertise of the professional that constitutes the legitimacy of the asymmetric relationship and thus of professional autonomy. In other words, professional authority is inherently functionally specific in the Parsonian conception:

The functional specificity of the professional-client relationship is similar to commercial relationships in the marketplace, Parsons explains. However, whereas the specificity of the professional-client relationship is confined by formal, scientific knowledge, the specificity of market relationships is confined by the contract. Both these types of functionally specific relationships are, then, different from the diffuse relationships characteristic of for example the family and friendship. Here, the relationship is not based on “universal bases of classification” (Parsons 1939, p. 462), but on the singular, particularistic and personal relationship between people:

To be sure, Parsons’ understanding of the professional-client relationship is structural, or as he calls it, “institutional” (Parsons 1939, p. 459). Another way of seeing this is that the asymmetrical relationship between professionals and clients is inherent to the status-role of each of these parties, and not necessarily to their concrete interactions. Parsons defines status-roles as the bundle of the actor’s social position in the system (status), and the actions that she performs in her “relations with others seen in the context of its functional significance for the social system” (Parsons, 1951/1991, pp. 15-16). Also, he includes the notion of the functionally specific and asymmetrical relationship in his discussion of the sick role (i.e. the role of the patient) vis-à-vis the role of the physician (Parsons, 1975).

This also means that in the concrete professional practices and concrete interactions between professionals and clients, it may be difficult to uphold the distinction between specific and diffuse relations (Parsons, 1939, p. 466). Even so, the concrete blurring of relationship logics is considered as a deviance from the universal and institutionalized model of professionalism, and not as pointing toward new types or variants of professionalism as such.

The functionalist conception of professions and professionalism has been thoroughly criticized. However, this critique is mainly directed at the knowledge base of professional closure, the disinterestedness (or altruism) of professionals and the ideological status of the service ethic (e.g., Friedman, 1962; Larson, 2013; Saks, 2010, 2012), and in many classical conceptions, the implicit assumption of a functionally specific relationship between professionals and citizen-clients is upheld. This includes, for example, Abbott’s (1988) well-known account of professional work, where he delimits the task of professionals as “human problems amenable to expert service” (Abbott, 1988, p. 35). Here, Abbott portrays professional work and the professional-citizen relationship as functionally specific, in particularly when he distinguishes between colligation and classification. Colligation, he explains, is “the assembly of a picture” consisting of rules declaring “what kinds of evidence are relevant and irrelevant,” whereas classification is “referring the colligated picture to the dictionary of professionally legitimate problems” (Abbott, 1988, p. 41). Thus, although Abbott admits some element of subjectivity and ambiguity in colligation, the diagnosis clearly is portrayed as including only functionally specific information.

Further, also Lipsky’s (2010) seminal account of the professional as a street-level bureaucrat assumes the relationship to citizen-clients to be functionally specific. Albeit Lipsky’s concept of discretion actually recognizes the adaption of policies to the “the human dimension of the situation” (Lipsky, 2010, p. 15), this adaption is mostly done by transforming citizens from “unique individuals with different life experiences, personalities, and current circumstances” (Lipsky, 2010, p. 59) into clients. One key element in street-level work is thus making sure that citizens are approached as clients only with regard to the functionally specific problems contained in policies.

Although it is rather widespread within classical accounts of professionalism, existing empirical studies actually demonstrate severe limitations to the assumption of a functionally specific relationship between professionals and citizen-clients. For example, studies of medicine and nursing have found that professionals legitimize their professional boundaries with reference to both expertise, competence, efficiency, and patient-centeredness (Sanders & Harrison, 2008), and that they maintain a discourse of holism even in the context of organizational change (Checkland, Harrison, McDonald, Grant, Campbell, & Guthrie, 2008). Also, studies of both teachers and caregivers (or early-education teachers) show how the relationship between professionals and children are characterized by a professionalism based on formal knowledge, emotions, and personal relationships (Hargraves, 1998; Isenbarger & Zembylas, 2006; Malm, 2009; Manning-Morton, 2006; O’Connor, 2008; Quan-McGimpsey, Kuczynski, & Brophy, 2011). However, many of these studies focus mainly on specific professions, or on professions within specific areas (health, care or education), and they seldom discuss the implications for a theoretical understanding of professionalism.

Furthermore, most current empirical studies of professions and professionalism question the Parsonian assumption of functional specificity by demonstrating how organizational and managerial changes affect professional identities and professional logics, shaping professionalism into new hybrid forms (Gulbrandsen, Thune, Borlaug, & Hanson, 2015; McGivern, Currie, Ferlie, Fitzgerald, & Waring, 2015; Nordegraaf, 2007, 2015; Numerato, Salvatore, & Fattore, 2012; Spyridondis, Hendy, & Barlow 2015). Thus, there are not many studies of professionalism in the context of close citizen contact (for one exception see Weicher & Laursen, 2003).

A comprehensive critique of the assumption of functional specificity in professional-citizen relations and the implications for professional work is thus primarily to be found outside the study of professions. Here, for example, the literature on representative bureaucracy (e.g., Meier & Capers, 2012; Watkins-Hayes, 2011; Wise, 2012) claim that a bureaucracy representative of the population in terms of certain characteristics will also be more perceptive and responsive towards the needs of groups with similar characteristics, as they will make their decisions taking special characteristics and needs into account. Also, studies of street-level work (e.g., Brodkin, 2012; Meyers & Nielsen, 2012; Ellis, 2011; Hupe, 2013; Maynard-Moody, & Portillo, 2010; Therum, 2003) demonstrate how discretion making is often based on street-level professional’s own preferences and personal habitus. Thus, whereas Lipsky assumes that citizens will be adapted into functionally specific roles or categories of “clients,” the literature on street-level work has shown how the “client-role,” as well as the concrete discretion making, are often based beyond a functionally specific foundation in formal knowledge or formal rules. This is particularly well demonstrated in studies of encounters between frontline professionals and citizens (Dubois, 2010; Epp, Maynard-Moody, & Haider-Markel, 2014; Harrits & Møller, 2014; Maynard-Moody & Musheno, 2003, 2012). These studies demonstrate how professional work must be grasped as everyday interactions between professionals and clients, including the conflicts, meaning-making and identity construction that we would normally include in our understanding of everyday life.

In sum, whereas most research within the studies of professions seem to uphold an assumption of a functionally specific relationship between professionals and clients, or focus more on hybrid professionalism resulting from changes in organization and management, studies on street-level and frontline work have pointed towards an understanding of professionals severely questioning the assumption of functional specificity for professions with close and frequent citizen contact. Unfortunately, however, there is no serious discussion in the literature of what such results can mean for our understanding of professionalism. Are we to understand the transcendence of a functionally specific relationship between professionals and citizen-clients as deviances from an otherwise solid model of professionalism and the institution of professions? And does this mean that professionals with frequent client interaction most likely fall “outside” the model of professionalism and are thus to be characterized as semi-professionals (Etzioni, 1969)? Or do we actually need to reconsider how we think about professionalism?

To get closer to answering these questions, I empirically explore how three different welfare professionals (from health care, child care, and education) make sense of their own professionalism and their relationship to citizen-clients, focusing explicitly on how this relationship may or may not be seen as functionally specific by the professionals themselves. Borrowing from recent studies on hybrid professionalism, I focus the analysis on the possible blending of logic based on formal knowledge and training with personal and emotional logics. Based on the empirical analysis, I then discuss how the results can contribute to a theoretical understanding of professionalism beyond functional specificity. First, however, I present the theoretical framework for the analysis as well as the methodological design and analytical strategies.

Theorizing sensemaking and hybridity

The analysis focuses on professional’s sensemaking; that is how professionals themselves see and describe who they are as professionals (identity), what they are supposed to do (role) and how they do it (work and discretion making). I thus study the possible limits of the assumption of functional specificity by studying how professionals see and describe their own professionalism, and not how they actually perform their work, make discretions or interact with citizen-clients. I return to the limitations of this perspective in the concluding discussion.

The focus on sensemaking is inspired by studies on hybrid professionalism (e.g., McGivern et al., 2015; Spyridonidis, Hendy, & Barlow, 2015), showing how professionals themselves make sense of their professional roles and identities in the context of managerial reforms. Following Skelcher and Smith (2015), I see hybridity as the presence and blending of multiple logics or rationalities. However, instead of focusing on hybrids of professionalism and managerialism, I focus on the possible combination of logic based on formal knowledge and training and logic based on personal and emotional relations. Referring to the Institutional Logics Approach (ILA) (Currie & Spuridonidis, 2015; Skelcher & Smith, 2015; Thornton, Ocasio, & Lounsbury, 2012), this is equivalent to focusing on what in ILA is called an institutional logic of professionalism and an institutional logic of family and civil society.

Also, whereas studies of professionalism and managerialism explicitly focus on developments and change due to recent managerial and organizational reforms, I do not focus on change. First, I do not have data to explore possible changes, and second, there is no theoretical ground for expecting change to be significant. Rather, we could expect combinations of logic based on formal knowledge and training with a logic of personal values, relations and emotions to be a generic trait for professions working in close and frequent contact with citizens.

I study the possible combinations or blending of logics across three dimensions of sensemaking mentioned above: sensemaking with regard to professional identities, professional roles, and professional work, or more specifically discretion making. Identity can be defined as an individual’s sense of self in relation to one’s social relations, groups, and contexts (Jenkins, 2008; McGivern et al., 2015). Professional identity is thus the professional’s immediate sense of him/herself as a professional in relation to his/her professional role and work: Who am I as a teacher? Compared to this, a role is a more stable, generalized, and institutionalized set of expectations related to a specific task or a specific position or function in society. Professional roles are thus the professional’s perception of what they are supposed to do, and the expectations to one’s tasks and functions as well as to the tasks and functions of colleagues (Dubois, 2010; Parsons 1951/1991).

Finally, I also study sensemaking in relation to what professionals do as professionals, or more specifically how they make discretions; that is how they reach a specific decision and how they reason in relation to such decisions (e.g., Grimen & Molander, 2008; Hupe, 2013).

Data and methods

Exploring professional’s sensemaking requires data where discourses and arguments are accessible, and where practices can be represented in a narrative but also condensed form. I, therefore, use semi-structured individual interviews with professionals (e.g., Charmaz, 2006; Kvale & Brinkmann, 2009). I study professionals from three different Danish welfare professions working with close and frequent interaction with citizens, namely health nurses, child care workers and primary school teachers. Studying three professions from three areas of work, and three different knowledge-bases helps reaching a point of saturation with regard to sensemaking. However, the three chosen professions also carry several similarities, making comparisons easier (but generalizations more difficult, see concluding discussion). The three professions are all held 3,5 to 4,5 years of education from a University College. Furthermore, they are all employed in a Danish municipality, and they all work with children and families whom they meet in rather informal settings (home, day-care center, and classroom). An obvious limitation of this selection of professions is, then, that no comparison can be made to professions without close and frequent interaction with citizens.

The data originates from the research project Categorization among Street-Level Bureaucrats, conducted in cooperation with Marie Østergaard Møller. Compared to Harrits and Møller (2014), the present analysis focuses on parts of the data not previously analyzed. ¹ In order to reach a minimum degree of analytical inference, the interviewees in this project were selected theoretically, focusing on a geographical spread as well as on a spread with regard to the social composition of the areas in which the professionals perform their work. Four municipalities were selected varying in size and urbanity (large+urban, small+urban, large+rural, small+rural), and within these four municipalities, two areas were selected, varying in the social composition of citizen (homogenous vs. heterogeneous).

Within each of the eight areas, local managers of the health nurse office, child care facilities, and schools were contacted, and they then made contact to interviewees that were able and willing to participate. This final selection via local managers can introduce a potential bias, however in many instances it was merely a question of making an arrangement that was practically feasible for both parties. Furthermore, with the diversity of backgrounds among the welfare professionals interviewed, with regard to age, years of experience and social background, no serious bias seems to be present. With one exception: Almost all interviewees are female (except for one child care worker and two teachers). Even though there is an overweight of women in all three professions, this bias makes generalizations and analysis with regard to gender somewhat problematic. A total of 58 interviews were collected by two interviewers (16 health nurses, 20 child care workers, 22 primary school teachers), lasting between approximately one to three hours. ²

The semi-structured interviews covered a variety of themes, but the present analysis is focused on questions regarding professional identity (“what do you like most about your job”; “who are you as a professional”) professional roles (“whose voice or what kind of voice are you,” “what is you role vis-à-vis society”) and professional discretion making (“when you make judgments like this, do you do that based on your professionalism or does it have more to do with a personal or inter-human understanding”) ³ . The question on discretion making is made after presenting interviewees with two vignettes describing the situation of two families. This design was made to test a hypothesis on social distance (see Harrits & Møller, 2014), which is not relevant for the present analysis. However, I use here the questions posed directly after the interviewees have told what they thought of the cases and what they would do, asking the interviewees to reflect on how they reached the decision they just made. These data have not been analyzed in Harrits & Møller 2014. ⁴

Several analyses have been performed by qualitative content analysis, coded systematically by the author in Nvivo, and afterwards condensed in different displays (Larsen, 2002; Miles & Huberman, 1994). The coding has focused on professional sensemaking and the exploration of the logic of formal knowledge and training vs. the personal, relational and emotion-based logic (see a coding frame in Table 1). The coding frame has been developed after an initial phase of open coding followed by a systematic, focused coding across all 58 interviews. During the open coding on professional roles, a third logic related to state authority became visible, and this was included in the coding frame. Also, during the open coding, the dimension related to the use of formal knowledge was supplemented to include the use of practical knowledge and experience, which is clearly seen by the professionals as being part of the knowledge-based logic. In contrast, coding for the personal, relational and emotion-based logic has been applied whenever professionals do not refer to their formal knowledge or professional experiences but instead to emotions, intuitions, common sense knowledge, everyday life or personal values. This distinction resonates with both Freidson’s understanding of formal and practical vs. everyday knowledge (Freidson, 2001, p. 34).

Professional sensemaking: Identities, roles, and discretions

Focusing, first, on professional identities, I find elaborate and widespread references to the use of formal and practical knowledge and to personal relations with clients. Many health nurses, for example, explain how the relationships they are able to form with families and especially the mothers when they visit the family homes after a new child has been born are crucial for their understanding of themselves as professionals:

Many health nurses also emphasize the use of formal and practical knowledge and expertise on for example nursing, nutrition and child development and the way they can use this knowledge to help families as an important part of their work identity. For a majority of health nurses, however, the relationship to families and the use of formal and practical knowledge is intertwined, and the one is almost portrayed as a prerequisite for the other.

Similar patterns are found among child care workers and teachers, with the exception that the relationships emphasized as crucial here are with the children and not with the families:

Also, many child care workers and teachers emphasize the use of formal and practical knowledge and the professional work they do, highlighting, for example, the ways in which they can help children develop and teach them important things. However, here I also find that the personal relationships and the use of knowledge is often combined and presented equally necessary:

Taken together, more than two-thirds of the child care workers and teachers present such a combined understanding of personal relationships and formal and practical knowledge as being part of their professional identity.

Moving on to sensemaking with regard to professional roles, I find a somewhat similar pattern, although role perceptions emphasizing the state, authority and enforcement of rules is also prevalent here, at least among health nurses.

Among the health nurses, half of the interviewees actually respond that they are “the voice of the child,” when asked directly what kind of voice they represent in their jobs and in society. This seems to indicate that health nurses emphasize the relationship to citizens in their role perceptions. When they further elaborate on this, however, only one health nurse emphasizes this aspect. Instead, all health nurses emphasize the use of knowledge as a primary role in society:

The data also displays a third role salient in the self-images of especially the health nurses, namely the role as a state agent (Maynard-Moody & Musheno, 2003). This role is not directly based on the use of specialized formal knowledge, but rather in the enforcement of state authority, rules, and regulations.

For the child care workers and the teachers, role perceptions are somewhat different. For the child care workers, it is clear that roles mirror the professional identities much more than for the health nurses. This means that many child care workers express the use of formal and practical knowledge and the establishment of close relationships with the children as their primary roles. Especially the role as professional caregivers is very salient among child care workers:

But this role as professionals is supplemented by a role perception emphasizing the establishment of relationships to the children, again not necessarily in opposition to but often combined with the use of knowledge:

Compared to the health nurses, however, the child care workers do not express a strong role perception regarding state authority.

A somewhat similar pattern is found among the teachers, with the exception that the conception of the relationship to the children is somewhat different. First, it is clear that many teachers see themselves as professional, knowledge-based experts teaching children the basic prerequisites for them to move on in life:

But again, this role as professionals is supplemented by a role perception emphasizing the relationship to the child. However, for the teachers, the primary aspect is not necessarily establishing an equal relationship with the children, but rather teaching them about values and how to become good human beings:

For the teachers, the primary role is thus to teach children both academic skills and how to become a good person, mixing a logic based on formal knowledge and training as well as practical knowledge and experience with a more personal and value-based logic. In sum, both child care workers and teachers express role perception combining a logic based on formal and practical knowledge with a personal logic and to a large degree downplay their roles as state authorities.

Finally, when asked about how professionals make sense of their own discretion making, I also find a quite strong mix of a logic based on formal knowledge, training and experience and a logic based on intuitions and a personal “sense” for the situation and the citizen-client. A minority of professionals explain how they use formal knowledge, expertise and professional experiences when making discretions in situations like the ones in the two case stories. Also, a few professionals explain how they mainly use their own personal experiences and intuitions. However, the majority of professionals explain how they do both, and many also explain how those two things cannot be separated:

This means that when exploring the ways in which professionals themselves conceive of and describe what it is they are doing, I find a widespread combination of the use of formal knowledge and professional experience with personal experience and intuition that is a combination of a logic based on formal knowledge and training and a personal and relational logic.

Summing up the analysis, I find across professions a widespread combination of a logic based on formal knowledge, training, and experience and a personal, relational and emotion-based logic when exploring professional sensemaking with regard to professional-citizen relations. Only very few professionals (seven out of 58) express a consistent sense of professionalism based solely on formal and practical knowledge. Also, taking into account the way in which the logic of formal and practical knowledge and the personal, relational and emotion-based logic are combined and integrated, I would categorize this as a blending of logics (e.g. Skelcher & Smith, 2015), that is as a “synergistic incorporation of elements of existing logics into new and contextually specific logics” (Skelcher & Smith, 2015, p. 440). Finally, I find rather strong indications that the assumption of functional specificity does not hold for the professions studied here. Below, I discuss the implications of this analysis for a theoretical understanding of professionalism among welfare professionals.

Concluding discussion

I started this article by pointing out how the assumption of functional specificity in the relationship between professionals and citizen clients is common across many theories on professions and professionalism. Contrary to this assumption, the analysis has demonstrated the presence of different logics beyond what can be characterized as functionally specific, and concluded that especially the integration of a logic based on formal and practical knowledge and a personal, relational, and emotion-based logic can be characterized as a blended, hybrid professionalism.

Obviously, no statistical inferences can be made, but the way in which many professionals make sense of their identities, roles and discretions in a very similar manner, using almost identical expressions, suggest a saturation of the data and analysis and thus the possibility of analytical inferences. In other words, I suggest that for professions in similar settings as the ones analysed here, that is professions working in close and frequent interactions with citizens-clients, in work focused on health, care or education, organized with a large autonomy for the individual professional, we could expect to see a similar forms of professionalism.

Also, the design of the study contains other limitations which should be taken into account. First, based on the ways in which professionals describe their work and the relationship to clients, focusing on the daily interaction and the establishment of a personal bond, it is reasonable to suggest that the closeness, frequency and length of interaction is the cause of the blending of logics. Professionals that do not interact with citizen-clients frequently or over a long period of time (e.g., specialized medical doctors or lawyers) may not display a similar professionalism. However, since this study cannot make any comparisons to professions without close or frequent client interactions, further studies are needed.

Second, the analysis focuses on professional’s sensemaking and not actual interactions or professional work nor sensemaking done by citizens. Future studies may thus explore further if a blended professionalism integrating a logic based on formal and practical knowledge, and a personal, relational and emotion-based logic can be found also in professional work and practice, and whether or not such professionalism is also salient in expectations and sensemaking among citizen-clients.

Regardless of these limitations, the study raises some questions for how we think about professionalism and the “peculiar sociological structure” of professional authority (Parsons, 1939, p. 460, see above). The blending of logics that has been demonstrated in this article are so widespread, explicit and integrated that they cannot be dismissed as “deviations.” Instead, they point to a more fundamental need to rethink the links, at an institutional level, between science/knowledge, professions, and civil society, as well as to rethink the normative foundations of professional authority and the legitimacy of professional autonomy.

Many theoretical discussions have questioned the conception of professional knowledge as formal knowledge, pointing out how some professionals have a more bodily and tacit knowledge developed in practice (e.g., Freidson, 2001; Pavlin, Svetlik, & Evetts, 2010; Schön, 1981). However, the present study point to hybrid professionalism beyond the use of practical knowledge, including also a personal, relational and emotion-based logic one would expect to find in interactions within families or among friends.

Further, especially scholars within child care and education have pointed to emotions and personal relations as an inherent part of teaching and care work (e.g., Hargraves, 1998; Manning-Morton, 2006; O’Connor, 2008; Quan-McGimpsey, Kuczynski, & Brophy, 2011) whereas others have pointed out the advantages of upholding a distinction between “being a teacher” and “being a parent” (Katz, 2000). From a sociology of the professions perspective, however, these scholars rarely discuss what the presence of a personal, relational and emotion-based logic means for our understanding of professionalism and professional authority.

One suggestion, then, would be to conceptualize professionalism broadly as based not only on formal and practical knowledge (in the broadest possible meaning of this concept) but also on personal qualities such as the ability to engage in personal and emotion-based relations to citizen clients. As argued by for example Pahuus and Eriksen (2012), welfare professionals need both professional discretion drawing on formal knowledge (and formal rules) and personal discretion, which they describe referring to both Aristotle (phronesis), Kant and Løgstrup and sum up as everyday knowledge, “common sense” or “le bon sens” (Pahuus & Eriksen, 2012, p. 63). Using this conception, the professionalism found in the analysis above would indeed be seen as non-surprising and suitable, building not only on science and professional experiences but also on shared (institutionalized) societal values.

Empirically, such an understanding seems more valid than the widespread understanding of professionalism as primarily based on science and formal knowledge. However, as a normative model for understanding professional authority and legitimacy, it points towards a deep-seated ambiguity in the notion of professionalism. On the one hand, a model of professional authority building on shared values and seeing close, personal and diffuse relationships with citizens as within the professional’s societal task, can be attractive. It may thus be a possible way of making encounters between professionals and citizen more perceptible to citizen needs (e.g., Maynard-Moody & Musheno, 2003; Pahuus & Eriksen, 2012), and caring and emphatic nurses, child care workers and teachers willing to get personally involved with the children and families they meet, indeed seem to be preferable compared to cold experts or strict rule-following bureaucrats.

However, normatively, there may be a “dark side” to such an understanding of professional authority as based on shared values. First, when professionals explain how they see the establishment of a personal relationship to the client as crucial for being able to change citizen behavior, or even for controlling citizens, the personal relationship to citizens seem to become somewhat instrumentalized and paternalistic. Second, when professionals explain how they consider it part of their job to engage in raising children and explicitly state that they do this on the basis of their own personal values, how can this be seen as a legitimate professional task in a differentiated, multi-cultural society where values and understandings of ‘the good life’ exist in multiple forms? And finally, when professionals seem to engage in discretion making that differs when confronted with differing social groups and different lifestyles and habits (Harrits & Møller, 2014; Nordhaug, 2013), how can this not raise raises serious questions of distributive justice and fair and equal treatment?

In sum, the analysis in this article suggests not only further analysis of the professional citizen relationship. It also suggests a broader discussion of our model of professionalism, both theoretically, empirically and normatively.

Acknowledgements

I would like to thank the Danish National Research Council for financially supporting the research in this article. Also, I would like to thank Marie Østergaard Møller for data collection and for cooperation on the research project. Finally, I would like to thank Lars Thorup Larsen, Annick Priuer, Morten Frederiksen, Oline Pedersen, Mathilde Cecchini and three anonymous reviewers for excellent comments on previous versions of this article.