Increasing parental participation at school level: a ‘citizen to serve’ or a ‘customer to steer’?

ABSTRACT Collaboration between schools and parents has become increasingly prominent on the political agenda in Norway. Schools are obliged to promote parent–school cooperation in accordance with parents’ rights as stakeholders in education. This article explores the governing strategies of seven primary or lower-secondary schools that have taken initiatives to improve parent–school collaboration. The main intention is to explore how New Public Management (NPM) measures (such as market values, decentralization, competition, and output control) and New Public Service (NPS) tools (including coalition building and citizens’ involvement) are reproduced at the local level when parent–school collaboration is put on the agenda. The analysis shows that street-level discretion at school level implies considerable uncertainty around the achievement of policy objectives. Different opinions on parents as a target group seem prominent in explaining how frontline workers act and strategize. Two distinct collaboration strategies are identified: serving and steering. The serving strategy is based on a linear partnership by making use of local knowledge in order to reach parents and enable their participation. The steering strategy is characterized by non-linear relationships with parents and certain steering mechanisms by routinizing collaboration activities, modifying goals for parent–school collaboration and rationing school services to parents.

During the past few decades, the collaboration between schools and parents has become increasingly prominent on the political agenda in Norway. Schools are now obliged to promote parent-school cooperation in accordance with parents' rights as stakeholders in education. Parents are viewed as an important resource for teachers, for pupils' wellbeing, as well as for educational achievements (Catsambis, 2001;Epstein & Sanders, 2000;Sheldon & Epstein, 2005a, 2005bSimon, 2004). Although collaboration is considered crucial, there are certainly challenges in bringing the political intentions of establishing operative relationships between parents and schools into practice. This paper explores the realization of national authorities' intentions of improving parentschool collaboration by analysing the influence of prevalent governing principles on performance at the school level.
Education authorities, as well as schools, may rely on different steering principles when implementing the collaboration policy. The long and dominating tradition of New Public Management (NPM) currently seems to have been opposed by a new post-NPM steering regime, New Public Service (NPS), using tools such as coalition-building and citizens' involvement (see Christensen & Laegreid, 2011;Denhardt & Denhardt, 2015. While NPM elements, including market values, decentralization, competition and output control, have dominated education steering over the past decades, a new turn towards NPS in Norwegian education governance can be identified. Lack of policy control and inefficiency are considered negative effects of NPM. To address these negative effects, NPS elements such as network governance, local partnerships and soft steering instruments seem to be increasingly adopted by educational authorities (Helgøy & Homme, 2016). To structure our investigation of parent-school collaboration in practice, we have specified a two-fold research question. Thus, we explore whether the governing principles of NPM and NPS are reproduced at the policy frontline by asking how, and to what extent, the principles affect the schools' strategies, and what the impact is on parent-school collaboration. In Norway, municipalities own schools and are responsible for implementing both national and local education policies. Moreover, the principle of local discretion is built into the governing model. Several researchers have pointed to local expertise as significant for effective performance in governing systems characterized by local discretion (Elmore, 1979(Elmore, /1980Alford, 2009;deLeon & deLeon, 2002;Hill, 2003;Hupe & Hill, 2007;Lipsky, 1980). Only 3.5 per cent of the pupils in primary and lower-secondary school attend private schools (Directorate for Education and Training, 2016). However, the degree of municipal autonomy in schooling varies over time in accordance with different policy programmes (Homme, 2008). Furthermore, the acceptance of teacher autonomy as a major characteristic of Norwegian education is widely recognized (Helgøy & Homme, 2007). Additionally, parents are formally included in the local school organization through representation in the Parents' Working Committee (FAU), the Coordinating Committee (SU), and the School Environment Committee (SMU).
The paper is structured as follows: first, we elaborate on the theoretical ideal-type framework of parent and school roles within two steering models based on the NPM and NPS typologies of Denhardt (2015, 2000), and front-level work by Durose (2011). Then, in our two-part analysis we explore the realization of parent-school collaboration along crucial dimensions at the frontline; that is, (1) the role of the school, the head teacher/teacher and the parent, and the type of initiatives carried out, and (2) the key organization steering mechanisms. By analysing how the different steering principles impact the content and practices of local parent-school cooperation, we explore how the parents are involved and taken into account. Consequently, we expect NPM and NPS to ground different views of the parents as a partner and contrast the kinds of instruments used to engage parents for collaboration. Essentially, the two steering models differ in that one serves to empower parents and the other steers by holding parents accountable (see Denhardt & Denhardt, 2015.

Research on parent-school collaboration
Research on parent-school collaboration has shown that parental engagement and positive relations between parents and school are important factors in improving student performance, school attendance and well-being at school. Parental engagement is, however, influenced by several factors. Different class backgrounds are seen as explanations for variation in involvement (Epstein, 2001(Epstein, , 2002Hallgarten, 2000;Hanafin & Lynch, 2002;Lareau, 1997Lareau, , 2000Useem, 1992;Vincent, 1996;Vincent & Ball, 2006;Vincent & Martin, 2000). Parents with limited resources or with a negative view of school are more reluctant to participate, whereas parents with higher education are inclined to be more active in collaboration with school compared to parents with less formal education (see Baeck, 2010Baeck, , 2007Epstein, 2001Epstein, , 2002Hallgarten, 2000;Hanafin & Lynch, 2002;Lareau, 1997Lareau, , 2000Useem, 1992;Vincent, 1996;Vincent & Ball, 2006;Vincent & Martin, 2000). Furthermore, mothers tend to be more active in parent-school collaboration compared to fathers (see Baeck, 2010Baeck, , 2007Baeck, , 2005Cole, 2007;Nordahl, 2000). The research findings above indicate that parents with less formal education are insecure about their own competence regarding school, or that their experiences with the education system have been negative. Uncertainty and negative experiences may prevent parents from collaborating with schools. In their systematic review of parent-school collaboration research, Desforges and Abouchaar (2003) underlined four elements for good collaborative practice: (1) existence of strategies to engage all the parents, (2) awareness of the divergent groups of pupils and parents, (3) identifying and valuing the parents' contribution to pupils' learning, and (4) empowering parents. All four elements underline the vital role of the school in establishing good collaborative practice with parents. However, how and to what extent governing principles may influence good collaborative practices remains underexplored.

Theoretical framework
According to Michael Lipsky, frontline workers contribute significantly to policy making through exercising discretion in their everyday work (Lipsky, 1980). Based on Lipsky's notion, we argue that school head teachers and teachers play important roles in the schools' strategies to improve parent-school collaboration. Moreover, we claim that the school's frontline workers relate to the prevailing governing principles of public administration. This relates to the ongoing debate on the development and change in the wake of the yearlong dominance of NPM (Christensen & Laegreid, 2011). Some researchers have referred to a change in governing models from steering to serving, characterizing the different relations to, and involvement of, the target groups of the policy (Christensen & Laegreid, 2011). Thus, according to Denhardt and Denhardt (2015), the NPS model represents a contrast to NPM. While the huge NPM wave from the 1980s started by viewing the government from the standpoint of markets and customers, the NPS standpoint is that bureaucrats play an important role in policy making. Bureaucrats are urged not only to carry, but also to steer their organizations (see Peters, 2011). In NPM, the relations between bureaucracies and their customers are understood based on self-interest, bringing in accountability mechanisms in order to achieve performance goals. While managers are being held accountable, they are also granted more power to choose the direction on how to achieve organizational goals. In response to this intensified steering regime, NPS represents a new public administration model grounded in the argument that the government belongs to citizens (Denhardt & Denhardt, 2015). NPS bureaucrats, as they implement public policy, are expected to focus on their responsibility to serve and empower citizens (Denhardt & Denhardt, 2015). By highlighting the sustainable building of institutions and achievement of policy objectives through coalitions with public, non-profit and private agencies in order to meet agreed-upon needs, this model of public administration decreases the emphasis on steering and managing.
Catherine Durose argued that local government in the UK has undergone significant and continuing reform, including 'joined-up'-government and partnerships that have emerged with a growing acceptance at the local level as 'the most appropriate and effective scale for community engagement' (Durose, 2011, p. 979). Building on Lipsky, Durose introduced an interpretive framework for analysing frontline work as street-level bureaucracy. The framework indicates thematic changes in the frontline work on local governance over time regarding (1) the relationship between politics and administration, (2) the key organizational mechanism, and (3) the frontline role (Durose, 2011). Hence, we aim to combine elements from the framework for analysing frontline work by Durose with governing principles of NPM and NPS to analyse our research question on how parentschool collaboration strategies are reproduced at the policy frontline. In our analytical scheme, we connect the general developments of public administration to the role of the school in parent-school collaboration practices defining two ideal types, the steering model (NPM) and the serving model (NPS). This implies that the frontline role of the steering model is the street-level bureaucrat, which uses discretion in his/ her work and develops a series of techniques for administering policy, such as routinizing, modifying, goal setting and rationing services (Durose, 2011;Lipsky, 1980). The relationship between the streetlevel bureaucrat and the parents is non-linear. Within the steering model, the parents' role would be as the customer. The key organizational mechanisms are (1) shaping of collaboration policy at school level and (2) creating mechanisms to achieve policy objectives. Mechanisms to achieve policy objectives can be both incentive structures and accountability structures; that is, rewarding parents, or obligating parents to explain and justify their conduct (see Bovens, 2007, p. 450).
The frontline role of the civic entrepreneur relates to the serving model. Smith (2012) stressed that the strategy of serving includes stronger citizen involvement, both in local decision-making regarding public services and in the implementation of public policy. An important motivational factor for establishing collaboration based on mutuality between parents and frontline workers at school is that collaboration is an effective way to make progress (Smith, 2012). The civic entrepreneur's work is about using local knowledge in order to reach parents and enable good collaborative parent-school practice. The relations to parents are seen as linear partnerships and the parents' role is that of stakeholder. The key organizational mechanisms are (1) shaping of collaboration by dialogue, and (2) building coalitions to meet mutually agreed-upon needs.
We apply the theoretical framework elaborated above to the Norwegian educational setting in our analysis of parent-school cooperation. The framework posits ideal types of public administration as steering and serving, respectively, along the following dimensions: frontline role, type of frontline work, parent role and key organizational steering mechanisms (as summarized in Table 1).
In order to analyse the role-pattern dimension in parent-school collaboration, we highlight the aspect of linear versus non-linear relationships by examining how local initiatives define and involve parents. Further, we identify the roles of schools' frontline workers and parents. The dimension type of work points to how discretion is used by modifying, limiting and rationing the work as coping mechanisms. Discretion is also used to involve parents and local knowledge as input into collaboration projects. According to the analytical scheme, we discuss the degree to which we can characterize the parent role as either a 'customer to steer' or a 'citizen to serve'. Moreover, in order to reveal the school role, we look for which mechanisms the schools make use of to either steer or serve the parents. Thus, the mechanisms of organization for achieving improved parentschool cooperation are relevant. Do schools build incentive structures for disciplining parents, or do they build coalitions with parents to establish a dialogue on collaboration tools? We will relate the expectations of good practices (Desforges & Abouchaar, 2003) to the two ideal types of street-level bureaucrat (steering) and civic entrepreneur (serving), respectively. The intention is to investigate the extent to which the collaboration policy at school level is characterized by serving rather than steering. Serving implies that the relationship between parents and the school is relatively linear. Thus, we expect the collaboration to be shaped by schools approaching parents and inviting them to voluntarily cooperate. According to the serving model, schools should identify and value the parents' different abilities to support the pupils and develop strategies to empower that support. The ideal type of civic entrepreneur seems to lack individual interests, as well as professional norms and processes in his/her daily work routines (Meyers & Nielsen, 2012). These constructions are rather what we expect to be the basis of how the street-level bureaucrat operates within a steering model. Accordingly, we will reveal whether frontline workers use certain coping techniques and routines, and whether they create incentive structures or oblige parents to establish parentschool collaboration. Moreover, we will explore whether frontline workers tend to establish a dialogue-based collaboration with parents, or make topdown decisions in collaboration initiatives.

Method
The analysis presented in this article is based on a qualitative case study comparing local parent-school collaboration initiatives in seven primary/lower-secondary schools in three Norwegian municipalities. A comprehensive and more detailed version of the study is published in a report to the Norwegian Directorate for Education and Training (Helgøy & Homme, 2015). The schools have received financial support from the Norwegian Directorate for Education and Training for implementing initiatives to increase parent-school collaboration, from 2011-2014. All schools are owned and administered by the municipalities. The data were taken from documents, such as local parentschool collaboration plans and municipal reports to the county governor (Fylkesmannen), as well as semistructured interviews with representatives for the county governors, the municipality school administration, head teachers, teachers and parents. A first round of interviews was conducted when the schools were planning their initiatives to increase parent-school collaboration in 2011/2012. The interviewees were school administrative representatives in the three municipalities, head teachers and teacherstotalling 17 informants. At that point, parents were not involved in the initiatives, and were consequently not included in the sample. Two years later, in 2013/ 2014, we conducted a second collection of interview data to investigate the development of collaboration initiatives. In addition to school administration representatives, head teachers and teachers, the second round of data collection included parents and representatives for the county governors. The parents included were mainly elected representatives (in FAU, SU or SMU). Although we interviewed parents from each of the municipalities, we experienced that parents were hesitant to participate in the research project. This seems to reflect the challenges regarding parent-school collaboration at some of the schools. Altogether, 29 informants were interviewed in the second round, making the total number of interviewees 46 (see Table 2). The interviews were conducted either at the individual school, at the municipality administration centre or at the county governor's office. In addition, one interview was conducted via telephone. The informants were interviewed about the content of the local collaboration measures, their role in and their experiences with the local initiatives, and their views on the local parent-school collaboration practices in general. The documents were subject to qualitative document analysis in order to supply the interview data.
In the next section of the article, we present our analyses of different strategies and initiatives for parent-school cooperation in the seven schools.

Dialogue-based collaboration (cluster A schools)
The pattern of roles and type of work Taking the first two analytical dimensions of roles and frontline work into consideration, we find that one cluster of schools implemented measures by including all parents in the collaboration. The overall element characterizing these schools is that they tried to involve the parents as equal partners both in defining the local parent-school collaboration initiatives and in implementing the decided measures. The frontline work and the communication between the school and the parents were characterized by linearity. In dialogue with the parents, one of the main measures was restructuring compulsory meetings between the school and parents. According to the initiative's objective for improving parentschool collaboration, compulsory meetings were important opportunities for partners (parents, teachers and school leaders) to get to know each other.
The informants pointed out two main benefits from changing the structure of the meetings in order to include more parents and make the agenda more relevant and interesting to them. First, both the parents' and the schools' representatives agreed that a specific challenge related to the meetings was that only some of the parents, and repeatedly the same parents, participated in the meetings, and in school activities generally. The biased parent participation marginalized some of the parents and influenced the overall parent-school cooperation unfavourably. Therefore, some of the schools made it a prominent goal to increase parent participation more broadly, as the following quotation from a parent indicates: We faced challenges in getting everyone to join. Not [all parents] participate in the meetings, nor are all the pupils attending the events meant for them. We wish to get everyone to join in. There are great differences among the parents when it comes to engagement in their child's school day and activities in general. (Interview 16,parent) Second, changing the content of the parent meetings was deemed important in order to engage all parents. In line with the linearity mentioned above, this was implemented to make use of the parents' 'local' knowledge and resources, as well as to enable parents to engage more broadly. The intention had thus been to turn the established practice from informational meetings, with one-sided, hierarchical dialogue from teachers to parents, to dialogue meetings focused on activating all the parents. This shift was made in order to build stronger relations between teachers and parents, but also between the parents themselves. At one of the schools, the slogan 'Adults Create Friendships' was launched to promote the idea that friendship among parents would lead to friendship among pupils as well. At this school, the teachers partly handed over the planning and the implementation of the meetings to parent representatives or the teachers planned the meetings together with the parents. All parents were invited to suggest themes for the agenda or for common social activities. The meetings often opened with a relatively short informational part, and continued with different themes of common interest to be discussed in smaller groups. The meetings ended with a social event, such as sports, a meal, etc. We also found examples of meetings taking place outside the ordinary classroom/meeting room. Some meetings even took place outside the school building. One example is that the parents were invited to visit the pupils at an outdoor overnight excursion in the neighbourhood. This was seen as an informal setting, thus increasing the parents' participation and engagement. In addition to the annual compulsory meetings, some parents regularly arranged social activities where pupils, parents and teachers would come together in order to get to know each other. This engaged all parents and built trust, as illustrated by a teacher in the following quote: In the formal meetings, most of the parents are [expecting] the teacher to take the initiative and lead, while the parents are waiting for '[the next item on the agenda]'. In contrast, at these social events, we are extremely focused on engaging all the parents. The discussions are not supposed to only relate to school. Instead, we intend to get to know each other in a different arena, so that we get a chance to engage the parents that might have negative relations to the school, and think of school as a distant thing. When they have met us in an informal setting such as this, the threshold on reaching out to us after is lowered. The teachers and/or the school leaders are perceived as less scary. (Interview 17,teacher) Involving pupils in the efforts illustrated above further strengthened broad local engagement. The schools and parents found that when the pupils were engaged in arranging meetings and other events, the parents were increasingly motivated to participate. Parent and teacher respondents stressed that participation in meetings and other activities had increased due to the involvement of pupils.
Generally, there is less engagement among fathers compared to mothers in schools (see, for example, Baeck, 2010;Nordahl, 2000). In local projects aiming to create a linear partnership with the parents, there was a specific focus on involving fathers as a group. With intentions to motivate more fathers to participate in parent meetings and collaborative activities, some schools arranged targeted events for this group. The strategy involved making the events more attractive to men by organizing different sports activities or technically oriented activities.

Key organizational steering mechanisms
Regarding the aspect of key organizational steering mechanisms, we identified measures implemented to build coalitions by inviting parents to set the agenda and together define common goals. These measures serve as an example of policy formulation through dialogue, as the main focus was to further engagement by enabling the parents.
One of the measures was to update the parents about the condition of their child's learning development and how to support their homework in order to increase learning. Additionally, the schools increasingly communicated with the parents by making use of digital services such as mobile phone and/or Internet-learning platforms.
In relation to formulating policy through dialogue, some parents had asked the schools for help to assist with homework, as well as more information on how to support pupils at home.
The attention given to systematic feedback to parents on pupils' learning progress concurs with a national programme implemented in Norwegian schools called Assessments for Learning (AFL). The AFL programme obliges schools to have procedures in place in order to secure the pupils' ongoing assessments with specific information about how they are performing, and what they should do to improve their performances. Giving the same information on pupils' performance to the parents, therefore, represents a mutual strengthening of both the AFL and parent-school cooperation programmes.
We found that one of the schools especially had brought these two programmes together, bringing about mutually positive results. The core measure was to develop individual plans for each pupil consisting of performance information, goals and information about how to reach those goals. Thus, pupils and their parents were given precise and constructive assessments, enabling both the pupils and the parents to increase the pupils' learning results. Lastly, in connection to this, we found that certain schools provided courses for parents in reading and mathematics. The intention was two-fold; to build relations with parents and to educate them on how to help their child with schoolwork. These findings are in line with the dimension of building coalitions between the parents and school in order to meet mutually agreed-upon needs.
The mechanisms of parent-school cooperation presented above are in accordance with the serving strategy. The schools are not simply responding to the parents as customers in order to solve their individual problems. Rather, the schools act in accordance with Denhardt and Denhardt's (2015) and Durose's (2011) definitions of public service focusing on creating opportunities for citizenship, which means forging trusting relationships with members of the public and working together to define problems and develop and implement solutions. We have seen that the content of the collaboration policy is made by building coalitions with parents via dialogue. The measures reflect mutually agreedupon needs to engage parents more broadly by lowering the threshold for participation.

Obliged collaboration (cluster B schools)
The pattern of roles and type of work In contrast to the dialogue-based type of collaboration assessed above, a quite different kind of collaboration practice was spelled out in another group of schools. Instead of inviting the parents to participate in defining projects and measures, the important decisions were made at the school level, without including parents in the process. Cluster B schools found it challenging to engage parents. From this perspective, the schools linked their challenge directly to actions aimed at including the parents in school activities. From a top-down angle, the main objective of these projects became to intervene in order to push parents to get involved in their children's education. Accordingly, with respect to the frontline dimension, the relationship between the school and the parents was clearly non-linear. As the schools defined the projects themselves, they indirectly defined what role the parents should have in parent-school collaboration. They perceived the lack of engagement from parents as caused by a certain kind of parent resistance. Therefore, the schools planned to affect the parents' attitudes. An important measure to influence parent behaviour within this cluster was the introduction of collaboration agreements for each pupil. Both the parents and the school were to sign the agreement. According to school administration and head teachers, the reason for bringing in the agreement was to increase parental support with regard to their children's schooling.
Through the introduction of weekly pupil assessments, the parental role became even more precisely defined in some of the projects. In project cluster A we pointed to the AFL program as a way of informing and empowering parents, as well as pupils, in order to strengthen pupil performance. The cluster A schools intended to enable parents to help pupils reach their goals. In the cluster B projects, however, the goals were even more specific and the responsibility to fulfil them was on the parents. For example, one of the schools introduced specific standards for pupils' level of mastery in all subjects. Each pupil's performance was classified as 'weak', 'mediocre', or 'high'. Each week, the parents were to communicate with the pupils about their achievements and degree of goal attainment as part of the agreement 'obligation' to the three partners: the school/teacher, the pupil, and the parents.
These agreements reflect a view of the parents as taking too little responsibility for their children's wellbeing and learning; an attitude that the school could change by implementing regulating mechanisms. Two schools in one municipality even referred to the agreement as a 'contract', implying that the parents and the school were obligated to consent to the goal of improving the pupils' learning. The decision to implement such agreements came from the municipality's administration (the school owner) in efforts to increase parental involvement. However, the contract was difficult to implement as a number of parents at both schools refused to sign the agreement. In the first round of interviews in 2011/2012, head teachers reported that they had tried hard to persuade sceptical parents to sign. Two years later, both head teachers and parents moderated the significance of the agreements as they now considered it voluntary to sign. Thus, in the end, the element of making the 'contract' obligatory was diminished because of opposition from the parents. The parents' view on the agreements was two-fold: on the one hand, the contracts could be effective in order to engage parents who were not involved in their children's school work; on the other, it was claimed that increased responsibility might have negative effects on parents who did not manage to follow up with their children. The agreement would make their relations to the school even more negative, as the following quotation from a parent illustrates: It could have negative effects on parents who in fact are not able to help their child, those who are not able to follow up, and those who have problems tackling their own everyday demands. Why should these parents not feel this contract is yet another demand they will not manage to follow up on? I think this is negative for them. (Interview 16,parent) Lastly, a measure connected to frontline work and parental role signalled that parents' attendance at school meetings was obligatory. As for the contracts described earlier, this strategy had no statutory authority and the parents seemed to perceive it as encouragement to participate. In one of the municipalities, parents of children who were starting school the upcoming year were also informed that they were obliged to attend a literacy course.
More common and moderate ways of demanding parental attendance were sending an invitation to meetings with a request for a written reply, or to include pupil activities in the meetings. If the pupils had to show up, the schools experienced that the parents were likely to show up as well.

Key organizational mechanisms
As the presented measures have shown, the cluster B schools made use of key organizational mechanisms that differed from the dialogue-based cluster A schools. In the cluster B schools, it seemed crucial for schools to take the lead and shape the parentschool cooperation initiatives at the school level. The techniques of obliging parents stand in contrast to the serving strategy of the cluster A schools, where the parents were included in defining and implementing measures. To oblige parents' engagement and participation in compliance with the NPM steering role mechanisms and incentive structures were created to achieve policy objectives developed by the actors at the school or school owner level. Due to a lack of dialogue with the parents, the frontline workers tried to steer by pushing the parents toward greater involvement in school activities. Although the municipalities/schools considered the agreement as binding, they emphasized that there were no legal sanctions facing those parents who did not cooperate. During the project period, the head teachers tended to accept that some parents did not sign the agreement and both partners considered the agreements as a request to voluntarily engage in parent-school cooperation, rather than an obligation. Nevertheless, this kind of top-down incentive building was done to steer parent behaviour. Announcing courses and parents' meetings as obligatory reveals the intention of steering, rather than serving, the parents. Table 3 summarizes the main findings.

Discussion
In line with the assumption that school leaders and teachers not only deliver prescribed instructions, but shape policy outcomes by interpreting policy and allocating scarce resources, we have empirically • Making participation obligatory • Quasi-contracts to regulate the relationship with parents explored frontline workers' strategies on developing improved collaboration between the school and the parents. Locally developed measures that leave decision making to the frontline are assumed to improve performance, and might be a result of a planned policy. Decentralization by giving certain amounts of discretion to the frontline characterizes both NPM and NPS, although the governing models are based on quite different patterns of roles and relations, as well as steering mechanisms, as is outlined in Table 1. The aim of this paper has been to explore how the school level is reproducing NPM and NPS governing principles. We have investigated how the governing principles affect the schools' strategies to improve parent-school collaboration, and also what kind of impact NPM and NPS have had on the actual parent-school collaboration.
The main objective in research on parent-school collaboration characterizes the collaboration as unpredictable, mainly caused by an uncertainty on how to collaborate (Nordahl, 2003). This is very clearly supported in our analysis. All seven project schools focused on the challenges of having inadequate parent-school collaboration, and emphasized their wish to strengthen the contact and engagement with parents. Moreover, the impression of uncertainty around the content of the collaboration is even strengthened by the fact that the strategies varied largely among the schools. Allowing the streetlevel bureaucrats to use their professional training and expertise, as well as a certain amount of discretion, is seen as essential to improve performance at street level (Elmore, 1979(Elmore, /1980Alford, 2009;deLeon & deLeon, 2002;Hill, 2003;Hupe & Hill, 2007;Lipsky, 1980). On the other hand, the local level is exposed to very strong pressure from government to improve the parent-school collaboration. By combining the perspective of governing models and streetlevel bureaucracies, the analysis contributes to new insights on how steering mechanisms are used in practice.
Concerning the question of how elements from the NPM and NPS governing models are reproduced in school strategies, we identified two different types of strategies related to these two overall models. The first strategy (i.e. the cluster A schools), characterized by involving parents in defining, as well as carrying out, collaboration measures, we conceptualize as a serving strategy. Efforts were made in order to decrease uncertainty, and to increase knowledge on how to improve the collaboration practice. Measures focused on enabling both the parents and the teachers/school leaders to engage. In line with the serving model, our data clearly show that within cluster A schools parents were viewed as stakeholdersas valuable local resources to ensure effective performance at the ground level. Furthermore, we found that the frontline worker in the cluster A schools can be characterized as a civic entrepreneur as he/she related to the parents as an equal partner in the collaboration. Additionally, the frontline workers seemed to attend to knowledge about the local community and the individual parent's interests. This is in line with the main dimension in the civic entrepreneur-type of frontline worker in our analytical scheme. The analysis of the cluster A schools is in line with the strategy of serving in that it includes citizen involvement, both in decision making and in the implementation of collaboration measures. Moreover, the analysis showed that an important motivational factor for establishing collaboration was based on the agreement between parents and frontline workers in that collaboration would expectedly provide progress. Related to our analytical framework, parents in cluster A schools are 'co-producers', per se, along with the schools' frontline workers, in the making of new modes of collaboration (Durose, 2011). Our study revealed that input and feedback from the parents were crucial in both improving collaboration measures and their effects with respect to progress in pupils' learning environment.
The analysis also revealed reproduction of the second strategy (i.e. the cluster B schools), which is in line with the steering model. These schools involved the parents in a quite different way. Instead of engaging parents by inviting them to participate through dialogue, the parents were involved to the extent that the schools held them accountable (Bovens, 2007). In the cluster B schools, the different frontline workers introduced techniques to engage parents by imposing school-parent contracts and mandatory meeting points. These techniques were seen as means to embrace the parents broadly. In addition to believing that making the parents accountable to the school would increase engagement among the parents, the frontline workers acknowledged the parents' contribution and engagement as vital to the children's learning processes, in the same way as within the serving strategy. We claim that the frontline workers within the steering strategy act as agents of social control by requiring certain kinds of behaviours from the parents with whom they interact. This is in line with the literature characterizing frontline workers as agents of social control (Durose, 2011;Meyers & Nielsen, 2012). We also found connections between cluster B schools and the well-known behaviour of street-level bureaucrats from the literature (Lipsky, 1980), characterized by routinizing, modifying goals, rationing their services and limiting the clientele. In our case, the frontline workers developed practices such as standardizing and rationalizing and modifying collaboration into certain agreements in order to effectively make parents act in accordance with schools' demands.
Our findings indicate that both NPM and NPS principles are reproduced at school level and impact on parent-school collaboration. Thus, an interesting question is to what extent the identified practices correspond with the research on successful collaboration. The four elements for good collaborative practices identified by Desforges and Abouchaar (2003) -(1) existence of strategies to engage all the parents, (2) awareness of the divergent groups of pupils and parents, (3) identifying and valuing parents' contribution to pupils' learning, and (4) empowering parentsunderline the vital role of the school in establishing good collaborative practice with parents. Obviously, cluster A seems to fit with all four elements for good practice. These schools build trust and network with all parents through establishing meeting arenas at the school. The precondition for parents to be 'co-producers' in pupils' learning environment by viewing them as a resource and giving them a voice in the classroom arena is also fulfilled. Here, research on the importance of the 'classroom' as an important collaboration arena supports our findings (Nordahl, 2003). One of the key organizational traits of good parent-school collaboration is thus to make use of network-based mechanisms, which provide opportunities to build better relations and trust between the school and the parents. To date, parent-school collaboration in Norwegian schools has been characterized by a lack of dialogue and collaboration. Cluster A schools might be an answer to such criticism. The serving strategy is in accordance with research findings stating that collaboration planned along with parents will have good chances of succeeding. Moreover, different strategies towards different groups of parents in order to include all are important in the cluster A school as well.
However, we find that the steering strategy, too, is in line with research findings regarding positive effects on making parents more responsible (Desforges & Abouchaar, 2003). This seems to be more efficient with respect to some groups of parents compared to others. Some parents perceived the way in which they were held accountable negatively because of the top-down style of policy making. Obviously, this hindered broad acceptance among the parents as a group. Moreover, the steering measures were neither differentiated nor adjusted, but simply implemented identically to all parents. Thus, in these schools the vital element for parent-school collaboration revealed in previous research concerning the need to acknowledge pupil and parent diversity was overlooked. This might have weakened the impact of the steering strategy. Accountability measures, however, seemed to influence engagement for some parents who felt they were motivated by pressure. The pressure was soft; there were no formal sanctions attached to collaboration, as these would generally be incompatible with the Norwegian Education Act. Nevertheless, measures implemented within the steering strategy might serve as opportunities to change parental behaviour in their children's schooling. The accountability strategy described above is therefore an example of schools using their institutional power towards parents, although the intention is to improve mutual collaboration. The top-Down point, as stressed by Durose (2011), is that the performance of frontline workers is effectively becoming the public policies they carry out. The trend of holding parents accountable to the school, based on the belief that lack of collaboration is caused by parents' 'wrong' attitudes to the institution, creates a demand for solutions on how attitudes can be changed. The introduction of contracts that oblige parents may be an example of a measure that could contribute to more school involvement from parents, although our data show the results to be somewhat uncertain.
The steering strategy relates to the frontline workers' resource constraints, as well as demands for congruence in their daily tasks. In order to cope with chronically limited time and other scarce resources, they seem to ration the services, as well as rationalizing program objectives (Brodkin, 1997). Thus, we need to recognize frontline workers' search for congruence with existing organizational norms in their strategies for improving school-parent collaboration.

Conclusion
The main intention of this paper was to explore how two overall governing models are reproduced in strategies at the local level when parent-school collaboration is put on the agenda. First, the approach of analysing street-level discretion in parent-school collaboration related to governing models of NPM and NPS revealed a strong connection between the two levels of policy making. The analysis uncovered the appearance of both NPM and NPS, and that streetlevel discretion implies considerable uncertainty around the achievement of policy objectives. The two models lead to different strategies at the frontline when implementing parent-school collaboration measures. Second, the contribution of the analysis is that it highlights the significance of the serving strategy for improving parent-school collaboration. The serving strategy includes elements stressed in previous collaboration research pointing towards good practices. The awareness of divergent groups of parents, the importance of mutuality in collaboration, valuing parents' diverging resources and, not least, empowering parents are all identified as important in this strategy. Thus, the serving strategy seems to deliver policy in a locally appropriate and mutually beneficial way compared to involving parents via the use of a steering strategy. At the same time, the steering strategy of holding parents accountable is also pointed to as an effective way to involve parents, although top-down accountability measures may have the opposite effect on certain groups of parents.
We should be aware that this study comprises schools striving to engage a relatively passive group of parents. The fact that schools make use of different strategies in parent-school collaboration, with varying degrees of parental engagement, underlines the point that there are uncertainties on how to collaborate (see Nordahl, 2003). This study implies that schools' efforts to improve parent-school collaborations signifies different views on the potential of including local communities, as well as diverging views on what motivates parents to engage with schools. Further, different opinions on parents as a target group seem prominent in explaining how frontline workers act and make strategies. The first strategy implies viewing parents as a resource that is willing to collaborate when the school introduces positive incentives. The other strategy views parents as reluctant to prioritize collaboration with the school and follow up on their children's schoolwork. In our study, these foundational, opposing opinions about the parent were decisive factors in the frontline work produced thereafter. Thus, more research is needed to fully explain why different strategies for collaboration are chosen.
Putting demand on frontline workers to interpret policy, acknowledge the local community, and work with the relevant community actors implies a bottom-up strategy towards parents. This stands in contrast to the top-down approach within the steering strategy. The steering strategy, however, may appear to frontline workers to be the only relevant solution when parents do not respond to dialogue initiatives nor initiate dialogue themselves.