As policy issues have increased in quantity and complexity, so has the necessity to include nonelected experts in policy-making (Kingdon 2014; Nalbandian 2006). New governing ideals (cf. Sørensen and Bentzen 2020), paired with difficulties in finding political candidates, have also pushed the need for civil servant involvement, especially in rural municipalities (Aars and Offerdal 1998; Erlingsson et al. 2022). In Sweden, where local autonomy is high and many policy areas are decentralised (Ladner 2019; Lidström 2016; Wollmann 2004), local civil servant influence has gradually increased as routine decision-making becomes administrative (Karlsson and Gilljam 2015). Niche politicians with special knowledge have been phased out in favour of fewer, more generalist candidates, while stronger emphasis has been placed on resource efficiency and pre-defined objectives (Karlsson and Gilljam 2015). The development has further made the borders between local politicians’ and civil servants’ tasks less clear, thus making the line between managerial and political power equally blurry and dependent on the relations between these actors rather than on a formal hierarchy and political or administrative mandate (Högberg 2007; Montin 2018).

As increasing complexity and asymmetric information are major drivers of civil servant influence, it should also be expected from processes with less established procedures, such as local citizen dialogues (Adenskog 2018; Smith 2009). This is especially likely when physical meetings or other resource- and labour-intensive formats are employed, as more taxing processes encourage more rational setups that, in turn, hamper the involvement of politicians (cf. Montin 2016; Tahvilzadeh 2015). The preconditions of rural municipalities can, further, be expected to amplify this. Here, resources tend to be more strained and full time politicians fewer, which obstructs politicians’ control of complex processes. While, in general, the ambition is to involve both politicians and civil servants throughout the process (SALAR 2019), technical and time-consuming tasks tend to leave most of the planning and realisation of these dialogues to civil servants (Hysing 2014; Tahvilzadeh 2015). However, how and the extent to which this discretion translates into tangible effects on policy is not clear. Both politicians’ and civil servants’ actions are guided by prevalent ideas about what governing should be and what roles involved actors should have (Pierre and Peters 2020; Stoker 1998; Svara 2006b; Sørensen and Bentzen 2020). Thus, high potential civil servant influence will not automatically result in bureaucratic rule (Stocker and Thompson-Fawcett 2014).

Just as politician–civil servant cooperation is an increasing necessity in everyday policy processes (Bobbio 2019), so is cooperation a deliberate and often emphasised ingredient of dialogues (SALAR 2019). Although information may be asymmetric and roles may be unclear, both civil servants and politicians tend to regard the involvement of elected representatives as crucial for maintaining legitimacy (Soneryd and Lindh 2019; Stocker and Thompson-Fawcett 2014; cf. Granberg and Åström 2007). The importance of the politician–civil servant relationship (Kingdon 2014; Overeem 2005; Poulsen and Koch 2018; Rutgers 2000; Svara 1985, 2006b) can therefore be expected to also appear in Swedish local citizen dialogues, despite potential civil servant dominance. While, in Swedish municipalities, such cooperation can be argued to reflect national policy as well as the general discourse (Montin 2018), studies have also suggested that local preconditions affect views and values (Denters et al. 2014; Olsson 2009; Syssner 2020). The basis for how politician–civil servant relations and their actions are formed may thus vary among municipalities. Although civil servants tend to be more influential in Swedish rural municipalities (Karlsson 2013), their influence in practice over complex local processes, such as dialogues, is more ambiguous.

To study civil servant influence related to prevalent ideas and politician–civil servant relations in this context, the two subsequent sections will present how these concepts are understood in this paper.

Traditionally, relations between politicians and civil servants have been treated as two opposite poles of a single dimension of public administration (Weber 2015; Wilson 1887). As many scholars found this and its competing model too simplistic and equivocal (cf. Sayre 1958; Stout 2010; Svara 1998), Svara (1999, 2006a, 2006b, 2008) attempted to congregate supplemental approaches into a more comprehensive framework. Thus, in addition to the dichotomy’s distinction between values and roles (Overeem 2005), this framework also involves the interdependence, reciprocity and overlap found in these relations.

Svara (2006b) illustrated this framework as a two-dimensional typology of political– administration relations models (Figure 1). While he mainly applied this typology as a theoretical backdrop for complementarity Svara (2006b, 2008), it can also help in understanding the differences between these relations. Its two scales determine where relations end up among the typology’s models: (i) the level of control of administrators by elected officials (henceforth, politicians), and (ii) the distance and differentiation between politicians and administrators in terms of roles and professional norms. In this context, roles are functions, or tasks, that are traditionally tied to either the political or administrative spheres (cf. Svara 1985). Norms, in relation to the typology, refer to values that are either political (e.g. advocating special interests and party or personal agendas) or administrative (e.g. rule of law, impartiality and consistency) (cf. Granberg and Åström 2007; Svara 2006b). As with a dichotomous model, the political and administrative spheres may be more or less distinct depending on the degree of role diversification and adoption of the opposite sphere’s norms (Aberbach, Putnam, and Rockman 1981). However, the interdependence and reciprocal aspects introduced in Svara’s framework can be expected to dampen norm changes and the effects of role overlaps. Thus, politicians refrain from ignoring the administration, even though they can, while civil servants seek politicians’ approval without being forced to (Stocker and Thompson-Fawcett 2014). Similar findings in Swedish studies (Montin 2016) also support these expectations for this study.

The typology consists of mainly four political-administrative ideal-type models that correspond to its two scales (Figure 1). Each model entails expectations of how these actors can and will act.

The separate roles model corresponds to the political–administration dichotomy. Here, both civil servants and politicians stay within their traditional roles (decision-making and implementation, respectively) while countering interference from the opposite sphere (Aberbach, Putnam, and Rockman 1981). The high control allows politicians to govern the level of civil servant involvement in policy-making (Svara 2006b).

A relation in the autonomous administration model is characterised by sphere separation and a self-directing, but not dominating administration. Political control and scrutiny of administrators are limited by knowledge and information asymmetry (cf. Putnam 1975), especially in complex policy areas (cf. Hysing 2014).

In the responsive administrators model, the borders between the spheres are weak, and administration integrity is low. Administrators adapt to political goals or favour political above administrative norms, as loyalty is expected and rewarded. However, Svara (2006b) argues that the relationship is not necessarily manipulative but may also be cooperative.

Cooperation is the main point of the overlapping roles model (cf. Alford et al. 2017; Bobbio 2019). Here, the border between politicians’ and civil servants’ roles may be even more blurred. However, in contrast to the responsive administrators model, civil servant autonomy is retained, and the two spheres are more clearly guided by their respective norms.

In practice, Svara (2006b, 2008) argues, relations encompass aspects of all four models (Figure 1). Nevertheless, the overlapping roles model is dominant, as it involves central aspects of interdependence, reciprocity and overlapping. This can also be expected at the Swedish local level, where formal hierarchy, norm integrity and unregulated role distribution are similarly prominent (Jonsson et al. 2012; Montin 2016). Manager civil servants in this context can be equally expected to gain influence during political standoffs (Högberg 2007; Jonsson et al. 2012; cf. Kingdon 2014).

In sum, applied to local rural citizen dialogue processes, this typology allows for analysing the nature of politician–civil servant relations and provides certain expectations for potential influence and effects on the processes. However, to analyse the practical implications of potential civil servant influence, this framework needs to be complemented by theory that addresses how (i.e. why or why not) potential influence translates into practice. Therefore, the following section outlines a way to understand the prevalent ideas, guiding civil servants’ perceived roles and actions.

Although the notion of governance paradigms has been debated, most scholars agree that different ideas about governing have been prominent at different points in time (Lo 2018; Osborne 2010; Stoker 1998). Drawing from Osborne (2010), Sørensen and Bentzen (2020) describe three distinct paradigms, presented as a succession of dominant sets of governance norms.¹ In the context of civil servant influence, these paradigms comprise guides for what roles to adopt, what priorities to make and how to utilise this influence.

The first of these paradigms, old public administration (OPA), emphasises a dichotomous separation of politics and administration. Thus, neutral administrations and the rule of law are central here (cf. Weber 2015). The second paradigm, new public management (NPM) (Hood 1991; Rhodes 1997), is characterised by private sector-inspired management, efficiency, governing output, citizen/customer satisfaction and evaluation. The last and most recent of the three is new public governance (NPG) (Osborne 2010; Torfing and Triantafillou 2013). Here, minimal hierarchy, co-governing and inclusion of stake holders are central. Public actors are ideally one of many equal stakeholders, although this may not be the case in practice (Blakeley 2010; Pierre and Peters 2000; Sibeon 2000).

Thus, each paradigm involves different ideas about the role of civil servants. Sørensen and Bentzen (2020) label these roles the policy advisor (OPA), the policy manager (NPM) and the policy facilitator (NPG). Policy advisors (OPA) are mainly experts who support policy makers. Their role is to ensure well-informed decision-making by providing accurate and neutral expertise. The policy manager (NPM) role involves audit and manager-like priorities.² Focusing on production efficiency, this role tends to emphasise policy output (Sørensen and Bentzen 2020), while soft goals, such as democratic values, are not prioritised (Aucoin 1990). As the NPG paradigm advocates network-like policy-making that involves different public and private actors (Kooiman 2003; Pierre and Peters 2000), this is reflected in the policy facilitator role. This role focuses on mediating and persuading actors to cooperate around mutual interests.

The succession of paradigms does not render them obsolete (Osborne 2010). Different contexts, issues and prevailing governance modes may involve aspects of all three paradigms and call for different approaches. This may require civil servants to adopt a mix of the three paradigm roles, depending on their position within the organisation and the nature of the process. This also depends on how they perceive their own role and what is expected of them (Sørensen and Bentzen 2020). Identifying traits of these paradigm mixes allows for a better understanding of how civil servants view their roles in the studied processes and, thus, how they can be expected to use their potential influence.

The two cases – the municipalities of Krokom and Gislaved – differ considerably in several aspects (Table A1). Krokom is situated in northern Sweden while Gislaved is in the more densely populated south. Krokom is geographically more than 5 times larger and has 1/11 of the population density of Gislaved. Both have a population growth below the national average, although Gislaved ranks high in the rural group and has twice the growth of Krokom. The economic differences are also clear. While Krokom has an average income and financial solidity below the national average, Gislaved is well above average in both measurements. These contextual differences are expected to affect the preconditions of the studied processes and thus civil servant influence. This is, in turn, expected to show in the cases’ placements in the typology (Figure 1). It further makes these cases strategic, not only in terms of diversity and transferability (Denscombe 2016), but also in relation to the study’s aim.

Although the cases are municipalities, the analysis was narrowed to specific processes within each case rather than to general policy-making. These dialogue processes qualify as democratic innovations (Smith 2009) to the extent that they were initiated by the council and were meant to (among other things) increase citizen participation and inclusion in policy-making (cf. Arnstein 1969). As these processes are optional and unregulated (Lidström 2016), they leave initiative and format to the municipalities themselves and can thus be expected to vary in design between contexts.⁵ The studied processes were aimed at community development and consisted of a combination of town meetings and deliberative workshops. Both cases followed a similar format, with meetings held in several smaller communities across the municipality. They differed somewhat in execution, number of meetings, time frame (which partially reaches beyond the scope of this study) and the various functions of the involved actors.

Two pairs of meetings in two of Gislaved’s seven dialogue areas had been arranged at the time of the empirical analysis.⁶ The remaining meetings were planned over the following years.⁷ In each community, two different meetings were arranged. At the first of these, department civil servants answered technical, legal and other administration-related questions. Two months after the first meeting, another followed during which politicians discussed issues with participants. Group discussions or town meetings were used, depending on the organisers’ expectations of central issues.

The Krokom dialogue involved a series of six meetings, several of which covered participants from two or more communities.⁸ These were held on three occasions, each with two meetings simultaneously. Meetings were planned as part group discussion and part town meeting. Unlike Gislaved’s, Krokom’s dialogue had an explicit theme – the municipality growth strategy – as a vantage point. Politicians attended each meeting, and technical civil servants were brought in when deemed relevant. Although not planned at the time of the study, this dialogue was meant to have a later follow-up process (Krokom Municipality 2023).

In Gislaved, two local development civil servants planned and executed the dialogue together with a small group of council members. These civil servants moderated the meetings, while the council members attended only the politician meetings. Questions and feasible proposals were planned to be sent directly to the relevant department.

The Krokom process was mainly civil servant driven with recurring follow-up meetings between the chief executive officer and the council. Planning and execution were largely done as part of the administration’s day-to-day work. No plan for follow-up meetings or input handling was made prior to starting the dialogue.

The empirical material consists of interviews with council members and civil servants involved in planning and realising the dialogue processes. Krokom’s four civil servants were managers (cf. Svara 2006b), in that they were either the chief executive officer or heads of departments. Their two Gislaved equivalents were lower ranking and especially designated towards rural development and democracy projects. Two politicians from each municipality were interviewed. The informants are presented in Table A2 of the Appendix. Informants were contacted via e-mail or telephone either directly, referred to by the administration as central to the process or referred to by other recruited informants following a snow ball selection strategy (Patton 2015). As the number of actors involved in the dialogues was limited and differed between the two cases, this produced two different sample sizes.

Interviews were semi-structured, approximately 60 minutes long and conducted either face to face or via telephone. They followed an interview guide based on the theoretical framework, which provided thematic and theoretical consistency while allowing for the necessary flexibility to address new questions raised during the interviews. Thus, for instance, questions about initial planning and allocation of tasks generated further questions about politician–civil servant communication and misunderstandings. I was responsible for conducting all interviews, transcriptions and translations from Swedish. Any quotes in this text are translated and reproduced verbatim, except for minor editorial adjustments.

While the case study design makes this study explorative rather than confirmatory (Denscombe 2016), its guiding research questions are drawn from theory. I therefore applied a thematic analysis of the data following the procedures presented by Braun and Clarke (2006, 2022; Clarke and Braun 2017), which allowed for the necessary part theory-centric, part data-driven approach (Denscombe 2016). In addition, analytical themes needed to be kept relevant, as they were neither given beforehand nor disconnected from earlier research. The method accomplished this throughout the analysis by forming new themes where vital data extracts failed to sort into existing themes. This method also allows themes to amalgamate or branch, which enables relevant distinctions, prevents bloated themes and is vital for making interpretations rather than merely summarising the data (Braun and Clarke 2006).

Braun and Clarke (2022) stress the need to become familiarised with the material before coding and thematising, in this case, by repeated reading of transcripts and listening to recorded interviews. This acted as preparation for the coding phase, in which codes were synthesised from analytically relevant data extracts (Terry and Hayfield 2021) in relation to the theoretical framework and the research questions. Potential themes were generated by grouping codes with common patterns. This initially produced candidate themes (Braun and Clarke 2022), such as “involuntarily political actors” and “flexibility vs. planning”. This process was repeatedly reviewed and revised to establish a set of three distinct contributing themes. The aforementioned example candidates were, for instance, amalgamated into Theme 1 and Theme 3’s subtheme 3a, respectively. The relevance of the final themes, individually and combined, ultimately depended on the answers they provided to the research questions. Overlapping or otherwise irrelevant candidates were either reworked or excluded. The analysis is reported theme-wise, partly presenting overall results and partly going into the specific results from each case. Data extracts were applied illustratively (Braun and Clarke 2022).

This first theme pays attention to the specific position, or rank, of the civil servants conducting each studied process. The analysis found contextual preconditions to have affected these civil servants in correspondence to their positions. These positions further proved to have implications for politician–civil servant relations as well as priorities and available options. Together, these aspects were found to have affected civil servant influence in the processes.

Based on the informants’ narratives, each case’s organising civil servants were understood to reflect their local context, in the analysis. This is in line with expectations regarding the impact of political and economic motives (Denters et al. 2014; Olsson 2009; Syssner 2020) and suggests that such aspects affected both the processes and their involved actors to a degree.

In Gislaved, this contextual impact was described as closely connected to the two civil servant informants’ positions as the administrator organisers of the dialogue. While they were lower ranking with no managerial responsibilities, they described having discretion within their areas of responsibility (Appendix A2). They further narrated how this had affected the design and focus of the dialogue, as their specialised positions made them the predetermined choice of organisers.

While the analysis found this civil servant position to have entailed a specific set of roles, both formal and self-imposed (Osborne 2010), it also found links between these roles and the politician–civil servant relation. Gislaved’s civil servants described their responsibilities and ambitions within the scope of the process as restricted to planning and realising the dialogue, while addressing participant input would be handled by the departments.¹⁰ From a governance perspective, this can be understood as an NPG mediator–, or facilitator– (Sørensen and Bentzen 2020), focused role and an absence of the other two paradigms (OPA and NPM). Part of this NPG focus could be argued to be intrinsic to this civil servant position, as they lacked the technical knowledge of law or science that Sørensen and Bentzen (2020) connects with OPA. However, this does not explain their lack of NPM-related concerns, such as efficiency or private sector-inspired flexible policy-making. The data further pointed to a broad support for the NPG focus, as the civil servants’ priorities were mirrored in the politician informants’ arguments. For example, a politician mentioned the importance of allowing time for response to participant input, despite the frustration this might generate:

“And [it is important] that you have a month between [the meetings] in order to give departments and committees enough time to look at these questions and issues that the citizens have brought [to the meetings], in order to be able to respond.” (Informant 3)

Given this NPG orientation, the civil servants conducting the Gislaved dialogue appeared to not be adopting multiple paradigmatic roles, as could be expected from the literature (Osborne 2010). Rather, they focused on and were supported by the politicians in facilitating cooperation between involved actors. The analysis linked part of this to the confinements of their responsibilities and the fact that specialised department administrators made additional advisor (OPA), manager and auditor (NPM) roles redundant. However, their overall narratives also suggested a general facilitator mindset, which, judging by the joint Gislaved narrative, also characterised politician–civil servant cooperation.

Gislaved informants continuously narrated relatively close cooperation between the two civil servants and an appointed group of three politician representatives (including Informants 3 & 4). Although previous studies have found tendencies among dialogue organisers to exaggerate consensus (Tahvilzadeh 2015), the analysis delineated the general planning as a joint venture with relatively clearly demarcated tasks. The narratives of both civil servants and politicians mentioned mutual contribution, with limited friction:

“… about the question of ‘politicians versus civil servants’: there was nothing, neither territory nor any competitive relationship. [Everyone] has a rather humble attitude towards each other’s roles there.” (Informant 4)

Thus, the data suggested that Gislaved’s organising civil servants maintained their existing relation to the politicians without gaining influence in the studied process. The data further showed that this was not simply due to their subordinate rank (cf. Hysing 2014; Kingdon 2014; Svara 2006b) but also to their chosen governance paradigm role (Osborne 2010; Sørensen and Bentzen 2020).

The expected impact from context appeared evident in Krokom. In the civil servants’ narratives, preconditions demanded that this dialogue process be planned with short notice and limited resources. Therefore, the natural candidates for organising administrators were heads of relevant departments (i.e. Informants 5–8). This civil servant position (Appendix A2) had broad sets of responsibilities, for example, for resource management connected to the dialogue but also largely for the mandate to see the process through. The executive officer (Informant 5) and the politician representing the opposition (Informant 10) further described how more responsibility had been temporarily moved at this time from the political to the administrative sphere (cf. Karlsson and Gilljam 2015).

While the politician informants claimed significant involvement in the dialogue, with references to council meetings prior to the process, civil servant informants emphasised occurrences of politician input. However, the civil servants otherwise described the process as a council-approved, administrative-sphere product with limited involvement of politicians outside the dialogue meetings. Knowledge and consensus about the potential problems regarding a process with little politician involvement (cf. Soneryd and Lindh 2019; Stocker and Thompson-Fawcett 2014) did not seem to have affected the course of the process. Thus, the analysis showed that contextual demands were either given higher priority than politician involvement or were not possible to overcome. This, in turn, suggested that local preconditions increased the influence of the organising administrators in the studied process.

The Krokom civil servants’ broader responsibilities seemed to necessitate a more diverse set of governance roles. However, the distribution was uneven. Their descriptions of roles and tasks and the dialogue’s overall setup, as well as the fashion in which all of this was narrated, largely corresponded to the NPM policy manager (Sørensen and Bentzen 2020). For example, they tended to speak in terms of efficiency when arguing for their dialogue’s more rational setup. Informant 7’s description of the financing for the municipality’s dialogues and how to cope with it illustrates the reasoning and incentives behind this approach:

“… in the actual planning process […] there is a given sum of money. But it must never cost anything […] We own the school, so there is no rent to pay for that, only [costs] for some material, and then there is [the costs for] working time. […] It’s low budget.”

Krokom’s civil servants further expressed concern with the lack of inclusion and engagement of politicians and the legitimacy problems this entails. Part of this can be understood as expected civil servant professional norms, which would suggest equally expected intact administration integrity and the ideal of high sphere separation (Svara 2006b). The concerns about politician exclusion from the process are also features of the OPA policy advisor role, as it focuses on aiding politicians’ decision-making. In the analysis, the Krokom civil servants therefore appear to have reluctantly overlapped the political sphere out of necessity. Moreover, despite these policy advisor role concerns, this overlap nevertheless made the administration highly autonomous (cf. Svara 2006b) and the organising civil servants increasingly influential. The process was consequently heavily affected by the largely NPM policy manager–focused role of this civil servant position.

Thus, although the Krokom process was affected by contextual preconditions, in terms of both the options available and who would be conducting it, the organising civil servant position had a decisive impact. While this group of managers could be expected to gain influence due to the council’s problems at the time, as this might have effects comparable to political standoffs (Högberg 2007; Jonsson et al. 2012; cf. Kingdon 2014), the influence was gained in a less anticipated fashion. The analysis suggests that the roles of this civil servant position emphasised rationality before politician inclusion, which, in turn, lowered politician control (Svara 2006b) and increased civil servant influence in the studied process.

The second theme focuses on how politicians and civil servants described the dialogue and the extent to which they had a shared understanding about its background, purpose and execution. This points to the level of interaction and what Svara’s (2006b) typology refers to as the distance between the two groups and is thus connected to civil servant influence.¹¹

The interpretation of the material showed indications of these aspects, mainly in the narratives around specific details of the processes. However, part of the results are ambiguous.

The narratives of both politicians and civil servants regarding the Gislaved dialogue, its background, purpose, problems and viable solutions were generally similar. While the politicians saw fewer obstacles, the civil servants shared their general optimism about the process. Both groups also presented themselves as the initiators of the process. On the one hand, this revealed some confusion and divergence in terms of the details, but on the other, it further suggested a shared positive view on the dialogue. An analysis of the material did not point to direct links between such shared views and mixing, or adoption, of opposite sphere norms, as can be found in extreme examples of politicised administration or bureaucratised politicians (Overeem 2005; Svara 2006b). However, the shared views suggested an element of adaptation, as civil servants described misunderstandings between themselves and the politicians early in the process:

“ … last spring, I felt a great frustration about the entire project when we, from time to time, got very contradictory information from […] different parts of the organisation about what had been decided about a possible working team. Who should be part of [this team]? Who should be attending the [dialogue meetings]? […] And it all led to that the invitation got really confused and the invitation for politicians got out pretty late. And then it was us [civil servant organisers] who took the hit. So we got criticism from the politician side about it being messy and late and… […] So, it’s important that there is an internal, shared view before a process like this – what should be done and what the purpose is and who should be involved.” (Informant 1)

This description was supported by the politicians’ narratives, which described an initial period of finding roles and outlining working processes. The analysis could therefore show a harmonisation through which initial diverging ideas between the two spheres somehow aligned along the course of the process. The data provided no clear image about what this harmonisation looked like, nor are they clear as to whether it affected hierarchy or distance (Svara 2006b) in the Gislaved process, although the narratives about the relation suggest equilibrium (see Traditionally, relations between politicians and civil servants have been treated as two opposite poles of a single dimension of public administration (Weber 2015; Wilson 1887). As many scholars found this and its competing model too simplistic and equivocal (cf. Sayre 1958; Stout 2010; Svara 1998), Svara (1999, 2006a, 2006b, 2008) attempted to congregate supplemental approaches into a more comprehensive framework. Thus, in addition to the dichotomy’s distinction between values and roles (Overeem 2005), this framework also involves the interdependence, reciprocity and overlap found in these relations.

Svara (2006b) illustrated this framework as a two-dimensional typology of political– administration relations models (Figure 1). While he mainly applied this typology as a theoretical backdrop for complementarity Svara (2006b, 2008), it can also help in understanding the differences between these relations. Its two scales determine where relations end up among the typology’s models: (i) the level of control of administrators by elected officials (henceforth, politicians), and (ii) the distance and differentiation between politicians and administrators in terms of roles and professional norms. In this context, roles are functions, or tasks, that are traditionally tied to either the political or administrative spheres (cf. Svara 1985). Norms, in relation to the typology, refer to values that are either political (e.g. advocating special interests and party or personal agendas) or administrative (e.g. rule of law, impartiality and consistency) (cf. Granberg and Åström 2007; Svara 2006b). As with a dichotomous model, the political and administrative spheres may be more or less distinct depending on the degree of role diversification and adoption of the opposite sphere’s norms (Aberbach, Putnam, and Rockman 1981). However, the interdependence and reciprocal aspects introduced in Svara’s framework can be expected to dampen norm changes and the effects of role overlaps. Thus, politicians refrain from ignoring the administration, even though they can, while civil servants seek politicians’ approval without being forced to (Stocker and Thompson-Fawcett 2014). Similar findings in Swedish studies (Montin 2016) also support these expectations for this study. ). The lack of inter-sphere norm adoption further supported this notion, although the adaptation pointed to relatively high sphere interaction.

The analysis could thus discern coherent views regarding the general aspects of Gislaved’s process, with some confusion around the details. It further observed that sphere distance appeared to have been affected in opposite directions but nevertheless displayed expected the traits of average Swedish local politician–civil servant relations (Montin 2016).

The Krokom politicians’ and civil servants’ narratives about the overall background and purpose of the dialogue were similar. Expected arguments and case-specific problems were brought up by both groups, further suggesting a degree of interaction and shared ideas about general aspects of the process. This was supported by the politicians’ and civil servants’ equally positive expectations. While the analysis was equivocal due to the relative vagueness of the politicians’ narratives, the similarities and apparent consensus could also be expected from the close proximity between politicians and managers (Högberg 2007; Kingdon 2014; Stocker and Thompson-Fawcett 2014). However, consensus was largely limited to the process’s overall background and purpose, while narratives about the process itself clearly diverged. The analysis attributed part of this to the low involvement of politicians in the process, especially in the planning stage. A civil servant summed up the current process and its yet-to-be planned continuation:

“Well, there was a requirement from the politicians that there should be some form of citizen dialogue, but then, how this was designed and realised, that was a pure civil servant product. But [the politicians] are happy that we carried it through, but now they say that we have to work on the next step where they are more involved and where they can take more of a lead.” (Informant 5)

This also exemplifies the diverging views on the details of the process, as the politician informants described the council as taking an active part in planning the dialogue. However, the overall narratives of both civil servants and politicians largely confirmed the abovementioned description. A tentative analysis could connect the diverging views on various details of the process to the low politician involvement and subsequent low politician–civil servant interaction in the dialogue.

Regardless of the impact of a lack of politician involvement, Krokom’s clear divergence regarding the specifics of the process suggested relatively high sphere distance in terms of interaction. This was unexpected given the managerial position of the organising civil servants, which tends to have closer relations to the political sphere (Högberg 2007; Kingdon 2014; Svara 1985).

This theme encompasses two related subthemes about (i) the impact of how a dialogue is planned and (ii) what political control would mean for the studied process.

An analysis of these aspects of the data pointed to a synergy between the two subthemes. While the planning could be argued to be shaped by the politician–civil servant relation, the data also present the design of the dialogue as potentially redefining the relation. The two subthemes would thus also define civil servant influence in the dialogue process.

The subthemes depicted Gislaved’s process as having a relatively long-term plan. This plan was characterised by shared traits of the dominant governance paradigm among both politicians and civil servants, as well as by active political support. Political control was expectedly high (Montin 2016) and consistent, as politicians both took part in planning and acted as the link between the two spheres.

In the Krokom case, the two subthemes jointly suggested a dialogue process in which politician involvement and influence were weakened and civil servants gained influence. The planning, informed by the dominant governance paradigm, was efficient and flexible but also short-term. This made the process less transparent, increased its threshold for politician involvement and kept civil servant–politician cooperation modest.