The influence of network orchestration and organizational formalization on goal orientation in public service delivery networks: an experimental study

ABSTRACT

Prior research on networks has established that tensions might arise between organizational structures and processes and network goals. This study investigates the role that network managers can play in mitigating these tensions through network orchestration. Based on an experimental design involving students and practitioners from the social domain, the combined effects of network orchestration and organizational formalization on goal orientation were examined. Results show that a negative effect of organizational formalization on network goal orientation is mitigated by network orchestration. Thereby, this study contributes to our understanding of the role of network orchestration in coping with tensions inherent to networks.

KEYWORDS:

• Network orchestration
• organizational formalization
• organizational and network goals
• public service delivery networks
• experimental design

Introduction

Public administration scholars argue that inter-organizational networks are an important response to complex societal issues that cannot be solved by one organization on its own (e.g. Lecy, Mergel, and Schmitz 2014; Popp et al. 2014). Therefore, increased emphasis is placed on the view that public service delivery is organized in horizontally integrated systems based on inter-organizational relationships between interdependent agents (e.g. Cristofoli, Meneguzzo, and Riccucci 2017; Lecy, Mergel, and Schmitz 2014; Osborne 2006). The increased attention for inter-organizational networks in public administration research is accompanied with a rich literature on network management. In this literature, a broad set of terms related to network management can be identified, such as integrative leadership (e.g. Crosby and Bryson 2010; Silva and McGuire 2010), facilitative leadership (Ansell and Gash 2008) and network orchestration (e.g. Bartelings et al. 2017; Paquin and Howard-Grenville 2013; Busquets 2010). A considerable amount of work has been done on what network managers actually do to operate in networks. Scholars use different terms such as strategies, tasks, roles, activities and behaviours of network managers (e.g. Agranoff and McGuire 2001; Ansell and Gash 2008; Bartelings et al. 2017; Cristofoli, Meneguzzo, and Riccucci 2017; Popp et al. 2014; Edelenbos, Van Buuren, and Klijn 2013; Klijn, Steijn, and Edelenbos 2010; Markovic 2017; McGuire 2002). Although the terminology used by scholars is diverse, they agree that network management is a critical or core activity for achieving the goals of public service networks. In achieving network goals, it is key for network managers to strike a balance between the individual goals of involved network partners on the one hand, and the goals of the network on the other hand (Agranoff and McGuire 2001; Ansell and Gash 2008; Carboni et al. 2019; Markovic 2017; Popp et al., 2014). Herein, an important task for the network manager is to translate the general purpose of a network into goals that give direction to network activities, further align network partners and facilitate the integration of activities of network partners (Carboni et al. 2019). As if this is not challenging enough, hierarchical control mechanisms from the network partners’ home organization, such as organizational formalization, are at play that aim to ensure that the behaviour and actions of network actors contribute to the achievement of individual organizational goals (e.g. Lunenburg 2012).

This research builds on literature on network management with the aim of examining the relationship between network orchestration and organizational formalization with organizational goal orientation and network goal orientation as perceived by organizational representatives participating in a public service delivery network. In this study, network orchestration comprises the activities of network managers aimed at influencing horizontal forms of collaboration towards the realization of collective goals (Goedee and Entken 2019). Specifically, this study focuses on operational work, bridging, and stabilizing as dimensions of network orchestration (Bartelings et al. 2017). The combined effects of network orchestration and organizational formalization on goal orientation are investigated using an experimental design. The design uses simulations of a public service delivery network in the Netherlands addressing families with several simultaneous psychosocial and socioeconomic problems, so called ‘multi-problems families’. Quantitative analyses are applied on data from ten simulation sessions with 118 respondents. During the simulations, respondents had the collective task to develop a joint approach. Results revealed a positive effect of network orchestration on network goal orientation. This contributes to our understanding of the effect of network orchestration and offers valuable insights in the role of network management in realizing network effectiveness. In addition, the results of this study showed that a negative effect of organizational formalization on network goal orientation was mitigated by network orchestration. Therefore, empirical evidence is provided for network orchestration as a way to balance tensions between organizational structures and processes and network level goals, which contributes to literature on network management and boundary spanning. In addition, the use of an experimental design complements other studies on network management using small N case studies or survey-based research. For practitioners in public administration, useful insights are generated that help to guide public service delivery networks towards achievement of common network goals.

Theory

The increased attention in public administration literature for networks and other collaborative forms of organization is rooted in a shift from vertically integrated intra-organizational forms of organization towards horizontally integrated inter-organizational forms of organization. Instead of assuming a government that maintains transactional relationships with independent contractors with an emphasis on efficiency in producing public services, in this perspective government maintains relationships with interdependent actors in which service effectiveness and collective outcomes are stressed (e.g. Cristofoli, Meneguzzo, and Riccucci 2017; Lecy et al., 2013; Osborne 2006). Following this, networks are an alternative form of social organization of public functions that complements and, in some instances, even replaces hierarchical or bureaucratic forms of authority (e.g. Lecy et al., 2014; Popp et al., 2014). Subsequently, increased scholarly attention is given to the management of networks. The underlying premise is that management in the context of inter-organizational networks differs from management in the hierarchical context of single organizations (e.g. Agranoff and McGuire 2001; McGuire 2002). Network management is argued to be a core task of public administration (Agranoff and McGuire 2001) and the ‘ultimate independent variable’ (McGuire 2002, 599) for network effectiveness. Indeed, empirical work has shown the positive effect of network management on network outcomes (e.g. Edelenbos, Van Buuren, and Klijn 2013; Klijn, Steijn, and Edelenbos 2010) and identified different contingencies for this relationship (e.g. Cristofoli, Trivellato, and Verzillo 2019; Markovic 2017).

We argue that, in order to realize network effectiveness, network goal orientation amongst involved network partners is a key issue for network managers because network goal orientation facilitates the integration of the actions and behaviour of network partners. This integration amongst network partners in turn can positively affect the realization of network goals. Therefore, network goal orientation could be important in understanding the mechanism between network management and network effectiveness. For the importance of network goal orientation in facilitating integration amongst network partners, we refer to the concept of purpose-oriented networks. This is defined as ‘a network comprised of three or more autonomous actors who participate in a joint effort based on a common purpose’ (Carboni et al. 2019, 210). The general purpose in networks represents an issue that network partners cannot solve alone (Nowell and Kenis 2019). The purpose of the network must get ‘translated into more actionable goals whose achievement can be monitored’ (Carboni et al. 2019, 214), which is necessary to give direction and further align activities of network partners (Carboni et al. 2019). The importance of shared purpose and network goals is also reflected in literature on what network managers do. For example, Agranoff and McGuire (2001) describe that network managers give shape to (perceptions of) the shared purpose of the network (i.e. framing), induce commitment of network actors to the networks’ purpose (i.e. mobilizing) and manage conflicting interests and goals (i.e. synthesizing). Also, driving network actors towards the achievement of the developed vision and network goals (Markovic 2017) and focusing on mutual goals exceeding the organizational level (Bartelings et al. 2017) are important activities for network managers.

It is clear from the literature that network goals are an important focus of network managers and that a joint network goal orientation presumably has a positive effect on network effectiveness. Nevertheless, to the best of our knowledge, the effects of network management on the orientation of network partners towards network goals has not been examined yet. This paper therefore focusses on the effects of network management on network goal orientation as perceived by network partners. In this, we apply the perspective of network orchestration, which is defined as network managers’ activities that aim to influence horizontal forms of collaboration towards the realization of collective goals (Goedee and Entken 2019). Based on systematic participatory observations of the behaviour of network managers, Bartelings et al. (2017) identified the following activities of network orchestrators that are complementary to traditional management: operational work, preparing documents, travelling, networking, bridging, stabilizing and transferring knowledge. However, effects of network orchestration were not investigated in the study by Bartelings et al. (2017). This research fills this gap by illuminating the effects of network orchestration on network goal orientation as perceived by network partners. This study focuses on operational work, bridging and stabilizing as dimensions of network orchestration (Bartelings et al. 2017). Operational work concerns assisting network partners with their work in the context of the collaboration. Bridging involves bringing partners together, searching for common ground, facilitating information sharing between partners and monitoring the collaboration. Stabilizing refers to involving partners in the collaboration, mediating conflicts between partners and stimulating the exchange of information and resources. Since operations are the most critical activities in goal achievement in public networks (McGuire 2002), this study focuses on the level of the network manager and organizational representatives of participating organizations in the network.

Network actors also have a link with their home organization. Therefore, the interdependency of network goals with organizational goals needs to be taken into account (Carboni et al. 2019). For network managers, as well as other network actors, it is key to strike a balance between the interests and goals of network partners on the one hand, and the purpose and goals of the network on the other (e.g. Ansell and Gash 2008; Popp et al., 2014). Following this, organizational goal orientation of network actors is important in addition to network goal orientation. With regard to organizational goal orientation, there is a long-standing line of research that relates to the work of Hage and Aiken (1967) on organizational structure. This includes mechanisms of coordination and control that aim to ensure that the behaviour and actions of employees contribute to the achievement of organizational goals (e.g. Jones 2013; Lunenburg 2012). One of those control mechanisms is organizational formalization. This is defined as the use of written rules, procedures and instructions of organizations to standardize operations (Jones 2013). People operating in networks are thus simultaneously confronted with the rules, procedures, instructions and goals of their own organizations and with those of the network. Thus, if network managers aim to balance network goals and organizational goals, organizational formalization can be considered as a mechanism of control aimed at attaining organizational goals. This interplay between intra- and inter-organizational processes and goals relates to the concept of boundary spanning. Boundary spanners can be defined as individual actors that cross organizational boundaries and connect the organization they represent to its environment (e.g. van Meerkerk and Edelenbos 2018; Tushman and Scanlan 1981; Williams 2012). This also involves connecting the home organization with inter-organizational networks (van Meerkerk and Edelenbos 2014, 2018). Boundary spanners must be able to understand languages, coding schemes and conceptual frameworks of individuals outside their organization in order to successfully gather relevant information externally and translate and disseminate that information internally (Tushman and Scanlan 1981). This applies to networks managers as well as other network actors or so-called informal boundary spanners (van Meerkerk and Edelenbos 2014). However, network managers play an important role in facilitating boundary spanning behaviour of network actors by creating a context in which collaboration between actors is triggered (van Meerkerk and Edelenbos 2018). Conteh and Harding (2021) argue that boundary spanning is increasingly attracting scholarly attention due to the need of actors to move horizontally in inter-organizational contexts and focus on inter-organizational mechanisms, whilst simultaneously acting vertically and taking internal intra-organizational structures and processes into account. This also involves balancing the tension between the position and responsibilities that boundary spanners have in their home organization with the role and responsibilities they have in forms of collaboration (Williams 2012).

Considering the above, this research aims to examine the relationship between network orchestration and organizational formalization as independent variables with organizational goal orientation and network goal orientation as dependent variables. The relationships between the variables are investigated in an experimental design involving ten simulations of a public service delivery network addressing multi-problem families. Hereafter, the hypotheses for the relationships between network orchestration and organizational formalization with organizational goal orientation and network goal orientation will be discussed.

Network orchestration & goal orientation

The network orchestrator influences network goal orientation in several ways. For example, by involving (relevant) partners, by searching for common ground between partners and bridging their opposing views (Bartelings et al. 2017). Also, a network orchestrator facilitates information sharing between involved organizational representatives (Nambisan and Sawhney 2011). Thus, organizational representatives gain complementary or new insights that make clear that they are dependent on representatives of other organizations with regard to their shared purpose. Additionally, the network orchestrator focuses on knowledge mobility between different organizational representatives (Dhanaraj and Parkhe 2006). Therefore, learning and absorptive capacity of organizational representatives is enhanced (Hurmelinna-Laukkanen et al. 2012). This leads to an improved ability of organizational representatives to recognize, process and apply each other’s knowledge and expertise. A network orchestrator is also expected to have a positive effect on network goal orientation by keeping in touch with all organizational representatives involved in the network (Bartelings et al. 2017). This serves to maintain salience of network purpose and network goals to all involved organizational representatives. Finally, as a central actor of the network, the network orchestrator focuses on creating trust and stability between partners within the network (Busquets 2010). Considering the above, the following hypothesis is formulated:

Hypothesis 1a:

There is a positive effect of network orchestration on network goal orientation.

With regard to the relationship between network orchestration and organizational goal orientation, there are different lines of reasoning. On the one hand, the purpose of the network reflects an issue that individual network partners cannot solve alone (Nowell and Kenis 2019). Therefore, foundationally, there is a positive interdependence between network goals and organizational goals. In this sense, organizational goals and network goals are complementary. Network orchestration could strengthen this complementarity as perceived by organizational representatives. For example, by bridging opposing views of involved network partners by the network orchestrator (Bartelings et al. 2017), organizational representatives might let go of potential incorrect assumptions and ideas about contradictions between goals. Also, the facilitation of information sharing between involved network partners by the network orchestrator (Nambisan and Sawhney 2011), could offer insights in the interdependence and complementarity between network goals and organizational goals to organizational representatives. In this way, network orchestration elicits the complementarity between organizational and network goals and further strengthens alignment and integration of goals and activities. This line of reasoning suggests a positive effect of network orchestration on organizational goal orientation.

In contrast, one could argue a negative effect of network orchestration on organizational goal orientation. Although network actors may acknowledge the network goal, it is unlikely that the network goals completely overlap with the organizational goals of all different network actors involved. Moreover, even if achievement of network goals is necessary for achievement of organizational goals, attention of network actors to network goals may divert attention from organizational goals. A high level of network orchestration motivates organizational representatives to allocate time and effort to the collective, network goal. This could decrease the amount of time and effort invested in individual organizational goals, leading to lower levels of organizational goal orientation (Locke and Latham 2002). Furthermore, in networks with low levels of orchestration, the network forms a more loosely connected entity that lacks integration, collective activities or shared goals (Busquets 2010; Nambisan and Sawhney 2011). Therefore, an increased focus of organizational representatives on organizational goals is expected in case of low levels of orchestration.

Even though both lines of reasoning have merit, it is expected that the level of network orchestration is negatively related to organizational goal orientation as perceived by organizational representatives.

Hypothesis 1b:

There is a negative effect of network orchestration on organizational goal orientation.

Organizational formalization & goal orientation

It is expected that the level of organizational formalization is positively related with the level of organizational goal orientation of organizational representatives in public service delivery networks. Organizational formalization is used as a coordination mechanism to increase alignment of actions and behaviour of organizational representatives with organizational goals (Lunenburg 2012). It does so by prescribing the expected behaviour and actions of employees through written rules, procedures and instructions (Mom, Van Den Bosch, and Volberda 2009). Additionally, organizational formalization is a form of hierarchical control that allows anticipation of decisions in various scenarios. In this, organizational formalization has a function to simplify decision-making, increase predictability and reduce uncertainty for organizational representatives (Gulati and Singh 1998). Therefore, organizational formalization provides a basis for the development of routines. Additionally, organizational formalization involves codification of knowledge (Vlaar, Van den Bosch, and Volberda 2007). Codified knowledge provides clear and concrete anchor points for the actions of organizational representatives. Especially in situations where organizational representatives cannot rely on specific (prior) knowledge, they will use general information they already possess (e.g. organizational formalization) as an anchor point (Hoefnagel, Oerlemans, and Goedee 2014). Taken together, it is expected that, in the context of public service networks, representatives of highly formalized organizations are expected to base their decisions and actions on internal rules, procedures and instructions that are formulated to attain their own organizational goals.

Hypothesis 2a:

There is a positive effect of the level of organizational formalization on organizational goal orientation.

In addition, it is expected that high levels of organizational formalization lead to low levels of network goal orientation of organizational representatives. High levels of organizational formalization impede the flexibility of organizational representatives and their adaptation to complex and new situations (Bechky 2006). Organizational formalization is less supportive for decision-making that deviates from standardized situations in ways that cannot be prescribed (Gulati and Singh 1998). Also, organizational formalization can impede experimentation, ad-hoc problem solving (March and Simon 1958) and innovativeness (Jansen, Van Den Bosch, and Volberda 2006). Moreover, high levels of organizational formalization, especially when imposed or communicated by top management, constrain the autonomy of organizational representatives (Mintzberg 1992), which makes them less inclined to broaden their orientation beyond organizational goals. Therefore, it is expected that organizational representatives perceiving high levels of organizational formalization are likely to have problems in translating the general purpose of the network into specific and contextually bounded network goals. It is expected that a certain level of flexibility and autonomy is needed to set network goals that cross organizational boundaries.

Hypothesis 2b:

There is a negative effect of the level of organizational formalization on network goal orientation.

Interaction effect

Taken together, it is expected that network orchestration moderates the effect of organizational formalization on network goal orientation, in such a way that network orchestration will attenuate the negative effect of high levels of organizational formalization on network goal orientation. If the level of organizational formalization is low, it is expected that organizational representatives are more inclined to focus on network goals. Especially if this closely aligns with the intrinsic motivation and logic of organizational representatives. However, if the level of organizational formalization is high, higher levels of network orchestration are expected to be needed to attenuate the negative effect of organizational formalization on network goal orientation. More concretely, if organizational formalization is high, and organizational representatives are bound by strict internal rules and procedures, strong vision and persuasiveness of the orchestrator could make representatives focus more on the network goal at hand. Network orchestrators could use interpersonal skills (O’leary, Choi, and Gerard 2012; Williams 2002) by making the network stable and trusted, so participants feel responsible to do their share in the network. Also, as network orchestrators create coordinated activities, trust and stability within a network increase (Busquets 2010) and as a consequence organizations in such networks might increasingly adhere to network values. Trust and stability are expected to grow over time, as trust, stability and routines are developed in collaboration by ongoing communication and mutual experiences (Ostrom 2000). However, with low levels of network orchestration, it is expected that the orchestrator is unable to create an environment that facilitates organizational representatives to cross the boundaries of their own organization.

Hypothesis 3:

The negative effect of organizational formalization on network goal orientation is attenuated by network orchestration.

The hypotheses formulated above were investigated by means of a simulation of a public service delivery network addressing multi-problem families. This will be explained in the methods section below.

Methods

Research design

The study adopted an experimental design to deductively test the hypotheses. To this end, it applied a simulation. The simulation used in this study was developed by a foundation in the Netherlands (Foundation Optimal Collaboration) in collaboration with Tilburg University. The simulation was conducted by the foundation with practitioners in the field as well as by Tilburg University with student participants. 118 respondents were distributed over ten simulation sessions organized between May 2019 and May 2020. Of these ten sessions, six simulation sessions were conducted with students and four simulation sessions were conducted with practitioners. In total, 80 students and 38 practitioners participated. The students participated as part of an elective university course focusing on service integration and network orchestration addressing multi-problem families. The practitioners involved in the simulations, dealt with multi-problem families in network settings in their daily work. This offered the possibility to control for differences between students and practitioners, which contributes to the reliability and validity of the results. All the simulation sessions followed a fixed protocol and had the same length in time. The simulation was developed in close cooperation with practitioners from the field. In addition, the simulation sessions with students were monitored by practitioners who confirmed the sessions reflected reality. Appendix A contains more information on the simulation protocol and the external validity of the simulation.

Simulation setting

In each simulation, a network involving six organizations from different sectors (i.e. youth care, social assistance, police, public prosecution, municipality and business consultancy) was faced with a case of a multi-problem family in which domestic violence occurs. The general purpose of the network was to secure the family’s wellbeing by developing a collective action plan to help the family involved. Each respondent played the role of a representative of an organization during the simulation sessions and was randomly assigned to one of the involved organizations. By randomly assigning the respondents, differences in personal characteristics and competences were not controlled for. This will be reflected upon in the discussion section. At least two participants were appointed to each organization by the simulation leaders. One participant actively participated in developing the collective action plan with other organizational representatives. The other participant(s) observed the joint meetings in which the collective action plan was formulated and was consulted by their colleagues in the case of decisions. For three simulation sessions with the practitioners, several organizations were represented by one participant because of the number of participating practitioners. Additional information about the simulation setting is reported in Appendix A. The manipulation procedure will be discussed after addressing the conceptualization, operationalization and measurement of variables below.

Conceptualization, operationalization and measurement of variables

After each simulation session, respondents completed a survey (see Table 1 below). In the survey, all items of constructs were measured using a seven-point Likert scale, ranging from ‘totally disagree’ to ‘totally agree’. A Principal Component Analysis (PCA) with oblique rotation and an analysis of the reliability of the scales (Cronbach’s α) were performed.

Table 1. Operationalization and measures.

Constructs and items
Network goal orientation
NG1) ‘There was a clear goal of the cross-sectoral collaboration’
NG2) ‘I reasoned from what would be best for society’
NG3) ‘I pursued goals that add the most value to society (in this case: improvement of the situation of the family as a whole)’
NG4) ‘For my organization, realizing a collective action plan was central’
NG5) ‘Integral collaboration with partner organizations was of importance for my organization’
Organizational goal orientation
OG1) ‘The goals of my organization were clearly formulated’
OG2) ‘I reasoned from what would be best to my own organization’
OG3) ‘I pursued the goals that are most important to my own organization’
OG4) ‘I was willing to deviate from the goals of my own organization’ (negatively formulated)*
Network orchestration
Operational work: OW1) ‘The orchestrator assists employees of my partner organizations with fulfilling their tasks if they were part of the common goal’
OW2) ‘The orchestrator assists partner organizations by performing supportive tasks’
Bridging: B1) ‘The orchestrator brings partners together to collaborate’
B2) ‘The orchestrator takes care of information sharing in the network’
B3) ‘The orchestrator monitors the collaboration between partners’
B4) ‘The orchestrator creates common ground between partners’
B5) ‘The orchestrator facilitates the collaboration’
B6) ‘The orchestrator consults with involved partners’
Stabilizing: S1) ‘The orchestrator makes sure that every partner feels involved in the collaboration’
S2) ‘The orchestrator mediates between conflicting partners’
S3) ‘The orchestrator stimulates the exchange of information and resources between partners’
Organizational formalization
OF1) ‘Within the organization I work for, for every situation there are procedures to follow to deal with that situation’
OF2) ‘In the organization I work for, work is done following prescriptions’
OF3) ‘In the organization I work for, rules occupy a central place’
OF4) ‘In the organization I work for, compliance to rules is strictly monitored’
OF5) ‘In the organization I work for, strict procedures are followed’

*Factor analysis revealed one factor after dropping the item ‘I was willing to deviate from the goals of my organization’.

The dependent variable network goal orientation was defined as the extent to which network partners were oriented towards network goals as perceived by organizational representatives (i.e. simulation participants). Based on two interviews with experts in the field, five items were used to operationalize network goal orientation. PCA on network goal orientation revealed one factor, the Cronbach’s α for the scale was .719.

The dependent variable organizational goal orientation was defined as the extent to which network partners were oriented towards goals of their home organization as perceived by organizational representatives. Based on two interviews with experts in the field, four items were used to operationalize network goal orientation. PCA on organizational goal orientation revealed one factor, Cronbach’s α for the scale was .713 after dropping one item. Exploratory factor analysis confirmed that network and organizational goal orientation were two separate constructs that were negatively correlated (r=−.136).

The independent variable network orchestration was defined as the activities of the network orchestrator (i.e. municipality) aimed at influencing the horizontal form of collaboration (i.e. public service delivery network) towards the realization of the collective goals (i.e. realizing a collective action plan for the multi-problem family) as perceived by representatives of the network partners (Goedee and Entken 2019). For measuring network orchestration, we focused on operational work, bridging and stabilizing as dimensions of network orchestration (Bartelings et al. 2017) based on existing scales (Schoofs 2014; Van den Berg 2014; Van Knippenberg 2017). PCA on network orchestration revealed one factor, the Cronbach’s α for the scale was .897. In addition, the context of the simulation was a public service delivery network constructed with a mandated lead-organization (i.e. municipality). However, for the measurement of network orchestration, we looked at the actions and behaviour of the network orchestrator as perceived by all network partners (see the manipulation section below for further explanation).

Finally, the independent variable organizational formalization was defined as the use of written rules, procedures and instructions to standardize operations (Jones 2013). Five items were used based on an existing scale (Mom, Van Den Bosch, and Volberda 2009). PCA on organizational formalization revealed one factor, the Cronbach’s α for the scale was .920.

The following control variables were used: age, gender (i.e. male or female) and background of the respondents (i.e. student or practitioner).

Manipulation of independent variables

Manipulation of the independent variables was done by adding specific texts to the instructions participants received for the organization that they represented during the simulation. Three different levels of each independent variable were created (i.e. low manipulation, control group, and high manipulation). For the control group, no text was added to the specific instructions for both independent variables. For low orchestration, the instruction for the municipality stated it was unclear what was expected from the municipality as an orchestrator and not much attention was given to this role. For high orchestration, it was stated that much value was attached to the role of the orchestrator and it was important to connect different parties and stabilize the network. For low organizational formalization, the instruction of individual organizations (see Appendix B for an example) stated that the organization allowed deviation from internal rules, procedures and regulations if necessary. For high organizational formalization, internal rules, procedures and regulations from the own organization were emphasized and it was stated that they should be followed. Since the respondents were randomly distributed over the different simulation sessions and organizations within the sessions, the respondents were also randomly distributed over the manipulation conditions. An overview of the manipulation of the independent variables per session is reported in (see Appendix C).

ANOVA as manipulation check

The One-Way ANOVA showed significant differences between manipulated groups at the p < .05 level for both independent variables. Also, the mean scores on perceived network orchestration (low: mean M = 4.941, control: mean M = 5.261 and high: mean M = 5.455) as well as organizational formalization (low: mean M = 4.243, control: mean M = 5.091, and high: mean M = 5.432) confirmed the manipulation succeeded. The successful manipulation might be explained as follows. Given the background of the students and practitioners, we argue the manipulation activated their (latent) knowledge. Here, we argue that students in this study acted more on structural or theoretical knowledge about the interdependence between variables in a system (i.e. explicit reality models), whereas practitioners operated based on intuition and operational experience (i.e. implicit reality models) (Dörner 1996). For organizational formalization, one could argue whether respondents were sufficiently knowledgeable of the internal rules, procedures and instructions, which in real-world situations can become more routinized and could lead to different effects of organizational formalization.

Content analysis as manipulation check

Qualitative data was used in order to corroborate the results of the questionnaire and the manipulation. Two simulation sessions with low and two simulation sessions with high manipulation on network orchestration were recorded and transcribed. Following this, the actual behaviour of the network orchestrator was counted in the transcripts. In this, content analysis was applied as this provides for a method to quantitatively express phenomena in qualitative data such as documents (Prasad 2008). Content analysis was performed by assigning codes derived from content categories to the qualitative data in order to identify phenomena (Prasad 2008). In this research, the dimensions operational work, bridging and stabilizing of network orchestration were used as content categories. The corresponding item numbers as presented in Table 1 above were used to code the presence of the different dimensions in the data. More specifically, every part of text in the transcripts that related to one of the items was coded likewise. Below, an example of a quote of a respondent reflecting on one of the simulation sessions is given. In the quote, the corresponding item numbers (see Table 1 above) are presented in bold text between parentheses.

You see what they [municipality] were doing, in their role as orchestrator, to keep everything together (B1), give everyone the time to discuss something (S1), monitoring the collaboration (B3). I think they have played a huge role in the effectiveness of this case meeting.

Respondent simulation 2

Below an example of a quote of the network orchestrator. Again, the corresponding item numbers (see Table 1 above) are presented in parentheses in bold text.

As a municipality, we attach great importance to collaboration with other parties (S1). First, I want to know whether there still is any information that is not shared yet, but yet can be crucial in setting up an action plan for this family? (S3)

Respondent simulation 5

After coding the transcripts, differences were analysed by comparing the simulation sessions with low and the simulation sessions with high manipulation on network orchestration. More orchestrational behaviour was observed in the two highly manipulated sessions (60 actions related to items of the dimensions operational work, bridging and stabilizing) compared to the two low manipulated sessions (43 actions related to items of the dimensions operational work, bridging and stabilizing). Thus, our manipulation in the instruction for the network orchestrator (i.e. municipality) made all network participants experience orchestrational activities in the intended direction. Bridging activities of the orchestrator were observed 39 times in high orchestration sessions compared to 30 times for low orchestration. In addition, stabilizing activities were observed 19 times in the high orchestration sessions compared to 10 times for low orchestration. The number of observations regarding items related to the dimension operational work of network orchestration were negligible. The content analysis was peer reviewed by the involved researchers.

Quantitative data analysis

A two-way-between-groups ANOVA was used to investigate differences in the level of goal orientation (i.e. network goal orientation and organizational goal orientation) based on the assigned manipulation conditions. The manipulation scores for network orchestration and organizational formalization refer to the manipulation group to which respondents were allocated. Thus the ANOVA was used to test whether the random allocation of respondents to either the low manipulation group, control group, or the high manipulation caused significant differences in their level of goal orientation. In addition, we conducted a seemingly unrelated regression (SUR) because of its suitability for modelling multiple correlated dependent variables. The SUR was performed on the perceived scores on the variables calculated at the level of individual respondents. We compared the results from the ANOVA and the SUR to draw conclusions about the hypotheses. Assumption tests showed no issues expect with respect to the assumed homogeneity of variance for the ANOVA, which was taken into consideration when interpreting the results.

Results

The results of the two-way-between-groups ANOVA on network goal orientation and organizational goal orientation are presented in Table 2 below. The results of the SUR on both network goal orientation and organizational goal orientation are presented in Table 3 below. Model 1 of the SUR contained the control variables, model 2 added the independent variables network orchestration and organizational goal orientation, and model 3 incorporated the interaction effect. The SUR showed a significant negative correlation between the dependent variables network goal orientation and organizational goal orientation in model 1 (r = −.277, p = .01), model 2 (r = −.162, p = .10) and model 3 (r = −.165, p = .10). After presenting Table 2 and Table 3 below, the hypotheses for network goal orientation and organizational goal orientation are discussed based on the ANOVA and the SUR. The descriptive statistics and correlations are reported in appendix D.

Table 2. Two-way-between groups ANOVA on different dependent variables.

	Sum of Squares	DF	Mean Square	F	P	partial η2
Network goal orientation
Intercept	2511.215	1	2511.215	3440.743	0.000	0.001
Network orchestration (M)	5.745	2	2.872	3.936	0.022	0.067
Organizational formalization (M)	9.985	2	4.992	6.840	0.002	0.111
Interaction	7.381	3	2.460	3.371	0.021	0.084
Error	80.283	110	0.730
Organizational goal orientation
Intercept	1596.327	1	1596.327	1210.054	0.000	0.917
Network orchestration (M)	1.107	2	0.553	0.420	0.658	0.008
Organizational formalization (M)	10.190	2	5.095	3.862	0.024	0.066
Interaction	7.610	3	2.537	1.923	0.130	0.050
Error	145.114	110	1.319

Note: (M) involves manipulated variables.

Table 3. Seemingly unrelated regression.

	Network goal orientation						Organizational goal orientation
	Model 1		Model 2		Model 3		Model 1		Model 2		Model 3
Intercept	5.898**	(0.295)	4.300**	(0.644)	4.665**	(1.197)	4.556***	(0.395)	5.007***	(0.977)	6.055**	(1.813)
Age	−0.013	(0.012)	−0.010	(0.010)	−0.010	(0.010)	−0.008	(0.016)	−0.007	(0.016)	−0.008	(0.016)
Participant (Dummy)	−0.576*	(0.256)	−0.298	(0.221)	−0.284	(0.225)	0.481	(0.343)	0.311	(0.336)	0.351	(0.340)
Gender (Dummy)	0.066	(0.160)	−0.008	(0.136)	−0.006	(0.136)	0.044	(0.214)	0.103	(0.206)	0.108	(0.206)
Network orchestration (P)			0.473**	(0.088)	0.399†	(0.224)			−0.263*	(0.133)	−0.477	(0.339)
Organizational formalization (P)			−0.205**	(0.050)	−0.291	(0.242)			0.190*	(0.076)	−0.056	(0.366)
Interaction					0.017	(0.048)					0.050	(0.073)
F-statistics	7.350**		15.400**		12.870**		0.890		2.630*		2.280*
R^2	0.157		0.395		0.396		0.022		0.100		0.104
Correlation dependent variables							−0.277**		−0.162†		−0.165†

† p < 0.10 *p < 0.05 **p < 0.01.

Note: (P) involves perceived variables.

Note: Estimated standard errors are in parentheses.

Network goal orientation

The results showed a significant positive effect of network orchestration on network goal orientation for the ANOVA at the p < .05 level for the three conditions [F(2) = 3.936, p = 0.022, partial η2 = 0.067] and at the p < .01 level based on model 2 of the SUR [β = .473, p = 0.000]. Post-hoc comparisons using Tukey HSD confirmed the direction of effects for the ANOVA. Thus, confirming hypothesis 1a. There was a significant negative effect of organizational formalization on network goal orientation at the p < .01 level for the three conditions based on the ANOVA [F(2) = 6.840, p = 0.002, partial η2 = 0.11] and at the p < .01 level based on model 2 of the SUR [β = −.205, p = 0.000]. Post-hoc comparisons using Tukey HSD confirmed the direction of effects for the ANOVA. Therefore, hypothesis 2b was confirmed. There was a significant interaction effect of network orchestration and organizational formalization on network goal orientation for the ANOVA at the p < .05 level [F(3) = 3.371, p = 0.021, partial η2 = 0.084]. Based on model 3 of the SUR, there was no significant interaction effect. However, the marginal effects (see Appendix E) revealed that the significant negative effect of organizational formalization on network goal orientation became less negative and less significant under high levels of network orchestration. Therefore, hypothesis 3 was confirmed.

Organizational goal orientation

With regard to the effect of network orchestration on organizational goal orientation, the ANOVA showed no significant effect [F(2) = 0.420, p = 0.658] and model 2 of the SUR showed a significant negative effect at the p < .05 level [β = −.263, p = 0.049]. Based on these mixed results, hypothesis 1b was rejected. There was a significant positive effect of organizational formalization on organizational goal orientation at the p < .05 level for the three conditions based on the ANOVA [F(2) = 3.862, p = 0.024, partial η2 = 0.066] and at the p < .05 level based on model 2 of the SUR [β =.190, p = 0.049]. Post-hoc comparisons using Tukey HSD confirmed the direction of effects for the ANOVA. Thereby, hypothesis 2a was confirmed.

Robustness tests confirmed the results, except for hypothesis 2b. Testing session level clustering of standard errors showed no issues with nestedness of the data.

Effect sizes and explanatory power

In interpreting the above results, it should be noted that all hypotheses confirmed by the two-way-between-groups ANOVA involved medium effect sizes (Cohen 1988), ranging from partial η2 = 0.066 to 0.11. Also, the explanatory power of the SUR should be noted. The full models explained 39.6% at the p < 0.01 level of the dependent variable network goal orientation and 10% at the p < 0.05 level of the dependent variable organizational goal orientation. Based on this, the results of the quantitative analyses indicate that other variables should be taken into account in explaining goal orientation. However, given the medium effect sizes and especially the explanatory power of the full model of the SUR for network goal orientation, we argue that network orchestration and organizational formalization are important variables that explain goal orientation. The former will also be addressed in the discussion section.

Differences between students and practitioners

With regard to above results, differences between student and practitioner respondents were checked. The students involved in this study participated in a course about service integration and therefore might be more sensitive to the topic and more likely to demonstrate significant results. Looking at the differences between students and practitioners, not all the effects were significant for each group separately, which can be explained by small sample sizes by splitting groups. However, directions of effects were confirmed and the analyses showed that both subsets of data contributed to the results of this study. For example, looking at the SUR, the positive effect of organizational formalization on organizational goal orientation was significant for practitioners and not for students. On the other hand, the negative effect of organizational formalization on network goal orientation was significant for students and not for practitioners. Additionally, the results show that practitioners’ level of network goal orientation was lower (b = −0.576, p < 0.05) than that of students, meaning that network goals were even more under pressure in a professional environment compared to a classroom setting. However, network orchestration and organizational formalization remain equally important as the effects are not moderated by the setting. Considering the former, this study seems to counter the critique of limited external validity of experimental study designs using student samples (Kam, Wilking, and Zechmeister 2007).

Discussion

The aim of this study was to investigate the combined effects of network orchestration and organizational formalization on organizational and network goal orientation in public service delivery networks addressing multi-problem families in the Netherlands. Network orchestration was defined as network manager’s activities in order to influence a horizontal form of collaboration (i.e. public service delivery network) towards the realization of collective goals (Goedee and Entken 2019). Using an experimental design, data from 118 respondents (students and practitioners from the social domain) were collected and hypotheses were tested. The results of this study indicate a positive effect of network orchestration on network goal orientation. No effect was found of network orchestration on organizational goal orientation. Organizational formalization had a positive effect on organizational goal orientation and a negative effect on network goal orientation. The negative effect of organizational formalization on network goal orientation was attenuated by (high levels of) network orchestration.

We believe this study has three main contributions. First, this research builds on and further refines scholarly work on network orchestration. Bartelings et al. (2017) identified the activities of network orchestrators that are complementary to traditional management. They conducted systematic participatory observations of the behaviour of a small sample of network managers and identified seven activities that comprise network orchestration. However, the effects of network orchestration were not investigated. This research takes a next step by examining the effects of three activities of network orchestration (i.e. operational work, bridging and stabilizing) based on an experimental design involving a sample of 118 respondents. The findings of this study show that operational work, bridging and stabilizing positively influence network goal orientation amongst network participants. In turn, network goals give direction, align network partners and facilitate the integration of activities of network partners (Carboni et al. 2019). Therefore, we argue that the findings of this study indicate that network orchestration functions as an integration mechanism. As network orchestration comprises network managers’ activities, this finding needs to be viewed in the broader literature of network management and its relationship with network effectiveness. Several studies show that network management positively influences network performance or outcomes (e.g. Cristofoli, Trivellato, and Verzillo 2019; Edelenbos, Van Buuren, and Klijn 2013; Klijn, Steijn, and Edelenbos 2010; Markovic 2017). That network management matters is therefore not new. However, in order to reach network goals, the actions and behaviour of network partners need to be integrated. The findings of this study show that and how the behaviour of the network manager (i.e. network orchestration) potentially facilitates this integration by realizing an orientation amongst network partners towards the network goal, which in turn could benefit the realization of network goals. Therefore this research contributes to possible explanations of the mechanisms at play in the relationship between network management on network effectiveness. Future research could further investigate the causal chain between network management and network effectiveness.

Second, the findings of this study shed light on the tension between intra- and inter-organizational processes and goals and the role of network managers’ activities to balance this tension. The results of this study indicate that organizational formalization, as a formal mechanism of control aimed at attaining organizational goals, hinders alignment of network partners towards network goals. However, the results of this study show that, in these situations, network orchestration can facilitate the orientation of network partners towards the network goal at hand. This interaction effect of network orchestration on the relationship between organizational formalization and network goal orientation relates to the literature on boundary spanning. For example, Conteh and Harding (2021) state that network actors need to act horizontally and focus on inter-organizational processes and goals whilst they cannot underestimate the significance of intra-organizational structures and processes. In order to balance this tension, boundary spanners connect the organization they represent to its environment (e.g. van Meerkerk and Edelenbos 2018; Tushman and Scanlan 1981; Williams 2012). This balancing act by boundary spanners is a challenge for network managers as well as other network actors. However, network managers can play an important role in creating a context in which collaboration between actors is triggered (van Meerkerk and Edelenbos 2018). Reflecting on our study, we argue that the activities of network managers (i.e. network orchestration) contribute to creating a context that is facilitative for collaboration and enables network actors to connect intra- and inter-organizational structures, processes and goals. Thereby, network orchestration could facilitate all network actors in balancing intra- and inter-organizational tensions. This might be explained as follows. Organizational formalization provides organizational representatives with the convenience of simplification of decision-making, increased predictability and reduced uncertainty (Gulati and Singh 1998). However, it could be argued whether this simplification, predictability and certainty fits the nature of complex societal issues that are central in public service networks. In these situations, organizational representatives can be faced with the discomforting dilemma that they cannot always rely on organizational formalization on which they routinely act upon. Through the activities of a network orchestrator, however, collective decision-making and mutual adjustment are facilitated. This can reduce uncertainty and provide organizational representatives a way to cope with the unpredictability and complexity of the issue at hand. This might work as follows. As stated, the activities of an orchestrator are, amongst others, involving (the right) partners, searching for common ground, keeping the network goal under attention of organizational representatives (Bartelings et al. 2017), sharing information between organizational representatives (Nambisan and Sawhney 2011) and focus on knowledge mobility between organizational representatives (Dhanaraj and Parkhe 2006). The aforementioned can lead to improved ability of organizational representatives to recognize, process and apply each other’s knowledge and expertise. This can provide trust and stability for organizational representatives within the network (Busquets 2010). However, while replacing the convenience of organizational formalization, at the same time, relying on collective decision-making can raise the question who is responsible for the decisions that are made and actions that are taken.

Third, we believe this study contributes to public administration research by using an experimental design as an innovative method to provide empirical evidence to understand network management. The experimental design made it possible to isolate the effect of network management (i.e. network orchestration) in a controlled setting, which complements prevailing research methods in network management research, such as small N case studies or survey-based research. By using simulations as a type of experimental design, ethical issues related to experimenting in real-world networks addressing multi-problem families are averted. Simultaneously, this study responds to the call for further improvement of the use of experiments in this field of research (Bouwman and Grimmelikhuijsen 2016) and shows that applying simulations offers the possibility to uncover behaviours of complex phenomena that are hard to observe in real-world settings (Joldersma 2000).

This study also has limitations. First, with regard to the external validity, the authors emphasize the simulation is not one-on-one equal with a real-world situation. Although the simulation was developed with and monitored by practitioners from the field, real-world settings remain different from experimental simulated settings. Future research should prove whether the findings of this study hold in real-world contexts. Second, the effect sizes and explanatory power of the statistical analyses in this study showed that network orchestration and organizational formalization are important variables explaining network goal orientation and therefore need to be taken into account in real-life situations. However, there are other explanatory variables that could be included in future research. For example, several contributions in the literature (e.g. O’leary, Choi, and Gerard 2012; Koppenjan and Klijn 2004) have shown the importance of individual characteristics and skills of successful network managers. In the experimental design of this study the role of the network orchestrator was randomly assigned and individual characteristics or skills of the network orchestrators were not controlled for. Third, several aspects of the type of the network in this study should be taken into account when inferring from the results of this study. The simulated networks in this study are formal or mandated networks in empirical reality. This could have influenced the results of this study, since formal mandates are argued to be powerful incentives to collaborate, indicate the need for collaboration and suggest that organizational and network goals can be aligned quite well (Popp et al. 2014). However, the results of this study indicated that a tension existed between organizational and network goals, since organizational goal orientation and network goal orientation were negatively related. Another aspect relating to the type of the network in this study, is that this study focused on a public service delivery network. Therefore, the results of this study might relate more to scholarly work on networks as a form of organization for service integration (e.g. Provan and Milward 1995; Raab, Mannak, and Cambré 2015). However, a large part of the literature on network management also addresses other types of networks, such as long-term project networks. The results found in this study might differ for these types of networks. Future research could provide interesting insights whether the findings hold for different types of networks. Finally, there are limitations with regard to the context of this study. The network in this study specifically focused on service integration for multi-problem families in the Netherlands. The focus on multi-problem families entails that decisions of network partners directly affect the lives family members involved. Decisions of network actors therefore might be more driven by emotional, moral and/or institutional arguments. This might be different for subjects that have less direct and profound impact on the private lives of people. Future research could reveal whether the results of this study also apply to networks addressing other societal issues and/or in other countries.

This study also has practical implications. Organizational representatives participating in networks maintain hierarchical relationships with their home organization and are often mandated by their top management. Top management or executives should consider the design of organizational rules, procedures and instructions as part of meta-governance strategies that regulate social processes by means of control and steering activities (Klijn and Edelenbos 2007). More specifically, very precise mandates or too strict organizational formalization might lead to a lack of responsiveness of organizational representatives in horizontal, inter-organizational contexts (van Meerkerk and Edelenbos 2018). Loosely formulated rules, procedures and instructions could provide room to be flexible and responsive. Here, top managers should consider the space of employees to deviate from organizational formalization (van Meerkerk and Edelenbos 2018).

Lastly, this article shows the need to deploy network management (i.e. network orchestration) to integrate activities of involved network partners and the role of network management in mitigating organizational routines that impede an orientation of organizational representatives towards network goals. Network managers, as well as other network actors, should be aware of the influence of organizational formalization and the role they have in connecting intra- and inter-organizational realities. That could help to locate impediments for collaboration and effectively deploy interventions aimed at easing network processes and reaching network goals. Thereby, network management enables creating a context that is facilitative for collaborative action between organizational representatives and reaching network goals.

Acknowledgments

We would like to thank the three anonymous reviewers and all the scholars at the 35^th European Group for Organizational Studies (EGOS) Colloquium in Edinburgh who participated in the workshop ‘Networks, Goals and Organizational Effectiveness: The Idea of Network Management’ and the three anonymous reviewers for their comments on earlier versions of the article.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data that support the findings of this study are available from the corresponding author, D. Cremers, upon reasonable request.

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/14719037.2023.2181982