Abstract
The research reported here is a structural analysis of the significance of ties to network leaders in securing the essential effort necessary to whole, goal-directed network functioning. Drawing on the work of Chester Barnard, we focus on one of Barnard’s three functions of the executive, securing essential effort and then examine the importance of certain network ties for securing effort in a goal-directed network. We specifically focus on Simmelian or mutual third-party ties to network leaders and the conditions under which those Simmelian ties are of greater significance for securing effort. Our study examines the Southern Alberta Child and Youth Network (SACYHN), a multisector publicly funded network that worked to facilitate interorganizational connections to improve child and youth health and well-being. Data were collected via an organizational questionnaire and elite interviews and were analyzed using Multiple Regression Quadratic Assignment Procedure (MRQAP). Implications are discussed for network management and leadership, for both theory and practice, focusing especially on the role of ties to network leaders in facilitating connections among member organizations working in different domains.
A core assumption in practice and in much of the public management literature over the past two decades has been that addressing “wicked” problems requires a multiorganizational strategy (Kettl, 2008; O’Toole, 1997, 2015). As a consequence, interorganizational networks have played an increasingly important role in the delivery of public and publicly funded services. Publicly funded, goal-oriented networks are the building blocks of many government programs in the 21st century. Provan, Fish, and Sydow (2007) characterized these types of goal-oriented interorganizational networks as “whole networks” and defined them as “a group of three or more organizations connected in ways that facilitate achievement of a common goal” (p. 482).
Early research offered evidence that whole networks have positive effects (see, for example, Lehman, Postrado, Roth, McNary, & Goldman, 1994; Provan & Milward, 1995), but it has also been demonstrated that collaboration among organizations is not easy and often results in its own set of problems (Huxham & Vangen, 2005). Though the study of whole networks is becoming more prevalent (Isett, Mergel, LeRoux, Mischen, & Rethemeyer, 2011; O’Toole, 2015), there is still a great deal we do not know about how to achieve the collaborative advantage assumed from the formal, goal-directed network approach. How leaders can effectively align multiple players across messy boundaries of action has been an enduring question in research and practice for the past decade (Kettl, 2008).
To achieve the collective action inherent in making a whole network approach work, securing effort from various organizations, often with different organizational goals and technologies, is essential. A distinguishing characteristic of the “whole network” is the pursuit of the common goal; unlike “serendipitous” networks, goal-directed networks are intended to build coordination around a specific common purpose (Kilduff & Tsai, 2003). The common purpose around which the autonomous organizations are deliberately organizing themselves is one that is difficult for any one organization to achieve on its own, such as serious mental illness (Provan & Milward, 1995), unemployment and workforce development (Herranz, 2008), emergency response (Moynihan, 2009), and social change (Saz-Carranza & Ospina, 2011). Thus, the collective action inherent in goal-directed whole networks is reliant on member organizations contributing effort toward the network goal and not just their own organizational goal.
Securing effort in a network setting is complex, however, because of the voluntary nature of most of network work. There are very few direct incentives for an autonomous organization to contribute to a voluntary network and few direct penalties if it chooses not to. Even if an organization wishes to contribute to action, actually doing so may not be so simple since any contribution to network action by a network member is often in addition to organizational demands. In addition, the incentive to move forward with network action may not always be ongoing, even if network members are committed to the network, because the directions of the collective may not always be where the members are most committed.
This article is, therefore, an examination of the securing of effort that is essential to achieving whole network goals and the role that ties to network leaders may play in securing effort. Using the categorization of leaders proposed by Bryson, Crosby, and Stone (2006), we categorize network leaders for the aim of this article as those formal and informal champions who acted as individuals but who were affiliated with a network member organization. Focusing specifically on securing essential network effort, we examine whether and when ties to network champions make a difference in securing effort from organizational members in a public goal-directed network. Thus, our research is guided by two research questions. First, are ties to network champions important in securing effort from network members? And second, under what conditions are their importance greatest?
To address these research questions, we examine the case of the Southern Alberta Child and Youth Health Network (SACYHN). SACYHN was a large, publicly funded goal-directed network with a mission to improve child and youth health and well-being. Understanding how a large, formal, goal-directed network like SACYHN functioned is important given the contemporary role of similar networks in a variety of areas including health, human services, and disaster management. We specifically analyze the underlying structural patterns and the quality of the dyadic and triadic relationships that comprise the whole network to propose that certain types of relationship structures, such as Simmelian ties to network champions, are important to securing effort in voluntary, heterogeneous goal-directed networks. A whole network is an aggregate of the dyadic connections among member organizations. Examining the dyadic and triadic structure underlying the network and how ties to network champions fit into that structure is important to understanding the tools available to achieve collaborative advantage in a multiorganizational whole network context.
Securing Essential Effort in Goal-Directed Networks
Leadership and management of goal-directed whole networks are important areas of study, as collaboration is not easy and collaborative inertia is a major challenge to overcome (Huxham & Vangen, 2005). Overcoming the frustrations of collaboration among organizations, though, may not be completely different from overcoming challenges to cooperation in organizations. After all, organizations are entities of cooperative action and the role of management is to overcome the challenges to cooperation (Barnard, 1938). Thus, we believe that a consideration of the early ideas of Barnard at the intraorganizational level has considerable value for understanding interorganizational network functioning more thoroughly.
One major reason we know comparatively little about managing a network versus managing an organization is because the mechanisms for managing in organizational settings conflict with what defines a network, as for instance hierarchy and the idea of networks as horizontal structures. Though we have learned that networks do have some elements of hierarchy or other formal control elements found in organizations (McGuire & Agranoff, 2011; Moynihan, 2009), networks as a unique form of organization are governed and managed in a different way (Provan & Kenis, 2008). For example, Agranoff (2006) argues that in networks the principles of “soft guidance,” such as trust, common purpose, mutual dependency, and leadership, function as replacements for command and control. The idea that these soft guidance principles function as replacements for command and control suggest that the aims of the mechanisms are the same. Though the same mechanisms for the management of an organization are not necessarily found in a network, such as hierarchy or formal means of accountability, these are only tools to achieve an end, which is overcoming challenges to collective action. Overcoming these challenges is a fundamental issue in networks, just as in organizations.
In this article, we will examine closely Barnard’s (1938) executive function of securing effort, which we assess in terms of network member relationship quality and strength. Barnard (1938) defined the function of securing the essential efforts as the work necessary to first bring persons into cooperative relationships with the organization and second, to elicit services from that person. Barnard’s definition of securing effort as a primary management task is consistent with the activities proposed in the two dominant network management frameworks. Of the four proposed network manager activities by Agranoff and McGuire (2001), three of these activities relate to securing essential effort as defined by Barnard. Activating involves identifying network participants and tapping the resources those participants can bring to the network, mobilizing is about securing commitment, and synthesizing is working to make favorable conditions to increase and improve the productivity of interactions among network participants. Similarly, game management according to Kickert, Klijn, and Koppenjan (1997) involves network activation, arranging interaction, brokerage, facilitating interaction, mediation, and arbitration.
While securing essential effort in a network setting entails many components, we examine one aspect that is principal in a formal, goal-directed network; namely, the quality and strength of ties among organizational participants. A network is comprised not of organizations working cooperatively with another entity, but with organizations working cooperatively with each other, creating a larger entity—the whole network. Therefore, one of the primary components of securing essential effort in a network entails promoting high-quality relationships based on network activities among organizational participants in the network, and then ensuring that organizations are contributing to those relationships.
High-quality relationships are defined for this purpose as strong trust-based ties. Even if two organizations have a relationship, one organization may in fact be frustrated by the lack of effort on the other organization’s part. Therefore, the first step of securing essential effort by first bringing persons into cooperative relationships means those relationships must indeed be seen as cooperative, both sides must trust that the other organization values the relationship. Even if the organizations trust one another, though, to ensure the second step of securing effort by eliciting services means the organizations should be working together in a meaningful collaborative way. An agreed upon way of measuring collaboration has not been established in the public management literature (Thomson, Perry, & Miller, 2008), but multiplexity is one proposed way of measuring the extent of collaboration of relationships (Provan & Milward, 2001). A multiplex relationship is one that is based on multiple activity ties and the more network activities a relationship is based on, the greater the commitment and contribution of both organizations to the relationship and thus, the greater the effort given to the network.
A whole network will likely be comprised of high-quality ties that originate from serendipitous connections at times, but from targeted and arranged connections at other times (Paquin & Howard-Grenville, 2013). Targeted and arranged connections are the result of orchestration by those acting to coordinate activities and promote value for the network as a whole. As Bryson et al. (2006) note in their review of the literature, multiple leadership roles are involved in the orchestration of successful collaborations, both formal and informal leadership positions. They categorized the key leadership roles into two categories: sponsors and champions. Sponsors are not necessarily closely involved in the collaborative work, whereas champions are those closely involved in the day-to-day work of the collaboration and work to keep the collaboration going and achieve its goals. We examine those network leaders involved in the day-to-day work, namely champions, and the structural role they play in network orchestration, specifically in securing essential effort in the case of SACYHN by (a) whether ties to network champions are indeed essential in facilitating cooperative, high-quality relationships between network members, and (b) when these ties to champions may be especially important.
Similarity, Simmelian Ties and Norms of Cooperation
Securing effort in a network is often more complicated than in an organizational setting because the mechanisms for ensuring accountability in organizations are generally not in evidence in interorganizational networks based on voluntary (though possibly mandated) and not contractual participation. This does not mean there are not mechanisms for ensuring accountability and essential effort, but rather, that the mechanisms possible in a network setting are different from those in organizations. Our research examines three structural components of networks that we hypothesize are critical for building essential effort among and across network members. These are norm similarity, domain similarity, the concept of Simmelian ties to network champions, and the interaction between these components.
Homophily has long been an important construct in the study of networks (McPherson, Smith-Lovin, & Cook, 2001). Organizations that are similar are more likely to work together (Monge & Contractor, 2003), thereby sustaining relationship quality and enhancing overall network outcomes. However, it is unclear whether this is because of the ease of collaboration, shared norms, the increased likelihood of crossing paths (Paquin & Howard-Grenville, 2013), or because participants are concerned about upholding their reputation. Though the exact reasons for why similarity impacts relationships among people/organizations may not be fully understood, similarity due to shared norms has been found to impact overall network outcomes. Moynihan (2009) found in his examination of the Incident Command System (ICS) in emergency management that the success of ICS was greater in the cases where the organizations involved in the emergency shared common norms, such as with fire fighters fighting fires, leading Moynihan to propose the importance of common norms even with a formal governance form is in place. When organizations are similar to one another, the norms are in place to ease the difficulties of collaboration. Therefore, consistent with these ideas from past research on networks, we propose the following baseline hypothesis regarding the relationship between norm similarity and the likelihood of organizational participants contributing effort to the network, at least in regard to entering strong collaborative relationships with each other and sustaining those relationships at a high-quality level.
When organizations are working together to achieve a collective goal, the network goals may not always be clearly beneficial to their own organizations. In fact, often times network goals and individual organizational goals will conflict (Provan & Milward, 2001). Therefore, the relationships that organizations maintain with other network members may instead be based on their dependency on one another and have little to do with network efforts. A cooperative strategy as a means to gain power when organizations are interdependent is the basis of Resource Dependency Theory (Pfeffer & Salancik, 1978). Interdependency, especially due to resource dependency, has been the focus of much research on the formation of network ties (Brass, Galaskiewicz, Greve, & Tsai, 2004; Ebers, 1997; Oliver, 1990). As recognized as early as Thompson (1967), interdependence is closely linked to organizational domains. A great deal of early research on interorganizational relations, especially the research examining the public sector or social sector organizations, has shown that the greater the domain overlap, the more likely organizations were to establish relationships (Oliver, 1990). Therefore, again consistent with the past research on networks, we propose the following additional baseline hypothesis regarding the relationship between interdependence due to domain similarity and the likelihood of organizational participants contributing effort to the network, at least in regard to entering strong collaborative relationships with each other and sustaining those relationships at a high-quality level.
Even though norm and domain similarity likely affect which organizations are collaborating with which other organizations, we propose that network champions may also be essential in achieving collaboration, especially in a formal, goal-directed network where organizations are not necessarily coming together serendipitously. Building on the Simmelian tie argument (Krackhardt, 1999), we argue that mutual ties to network champions can facilitate connections among organizations and assure the organizations are contributing effort to the network. The Simmelian tie argument, based on Simmel’s classic work, makes the case that members of a triad (or larger structure) are more constrained than members of dyads. The reasons for this greater restraint are due to the less individuality, less bargaining power, and less hardened positions with the addition of a third member to a group. Members of a dyad have greater power to act independently, but with the addition of a third member, each member of the group must play by the group norms to continue to participate (Krackhardt, 1999). The addition of the third party is much like the observations regarding cohesive structures (Coleman, 1988) and “cooperation becomes a shared value in densely connected structures, and individuals with common third-party ties are naturally inclined to devote time and effort to knowledge-sharing interactions with other members of their clique” (Tortoriello & Krackhardt, 2010).
In that case, any triadic ties could be important for securing effort in a network; however, we focus specifically on network champions as the third party. Boundary spanning is no easy feat and processes are often necessary for translating across boundaries and achieving integration of different perspectives (Carlile, 2002). Network champions, who are committed to the goals of the network, are in position to work to integrate the different perspectives of organizations with each other and with the network goals. Paquin and Howard-Grenville (2013) found network members concerned with the whole network value to be important in shaping a network. According to Paquin and Howard-Grenville, “They can influence the network by developing common goals, spurring actor interest and engagement, and/or defining norms of action” (p. 1625). Therefore, Simmelian ties which include network champions create the substructures where cooperative norms can be used to influence network members, along the lines of Coleman’s cohesion argument and as Paquin and Howard-Grenville argued.
The concept of the Simmelian tie is about more than just third-party ties, as the idea is that a Simmelian tie represents an even stronger relationship than a strong dyadic tie because of the additional closure aspect (Krackhardt, 1999). This dimension of strength is consistent with Granovetter’s (1973) strength of weak ties argument since Granovetter’s notion is that strong ties equate closure; weak ties will not necessarily lead to the triadic closure associated with Simmel and Coleman’s arguments. Using ties to influence network members to cooperate with one another, therefore, is only likely if network champions have strong, high-quality ties with the individual member organizations. With strong ties in place to a pair of organizations, champions can work to influence those organizations to collaborate through various leadership activities. For instance, they can attempt to influence organizations by working to resolve conflict between organizations, persuading them by drawing on network social capital, or by finding the resources to support their collaboration. Thus, we hypothesize that the greater the strength of the Simmelian tie with network champions, based on multiplexity, the greater the likelihood that a pair of organizations can be influenced to contribute effort and be held accountable for doing so.
Though we hypothesize that the ties a pair of organizations have to network champions will be a factor in securing effort, these ties are likely to be of greater importance under certain conditions. Organizations working together in a network are interdependent, thus why would they otherwise be working together under the umbrella of a goal-directed network; but the extent of the interdependency may vary. In fact rather than domain consensus, Van de Ven (1976) proposed that intermediate levels of domain similarity is what will likely lead to the establishment of relationships. Therefore, it is important to consider the nature of the interdependency of organizations and when collaboration will be more instrumental to the purposes of the organization and when it will be more for the purposes of the network work and thus, possibly a more difficult relationship to build.
Where collaboration is less likely to happen serendipitously, either because of the lack of domain similarity between organizations or the lack of norm similarity, we argue that ties to network champions will be important in assuring organizations are putting forth network effort. Under those conditions where organizations are unlikely to pursue a collaborative relationship with another organization for its own purposes or efforts to do so may be complicated due to the nature of the interdependency, we argue ties to network champions will be most important. Obstfeld (2005) contends that overlooked in Simmel’s argument about the introduction of a third party are the various strategies of the third party noted by Simmel. Burt (2000) emphasized the “tertius gaudens” concept to explain his structural hole theory, but Obstfeld claims that this manipulating orientation of tertius gaudens is only one of the strategic orientations embedded in Simmel’s argument. Obstfeld resurrects another orientation which he labels as “tertius iungens” and which is the third who joins and connects either disconnected or adversarial individuals. Along the lines of this “tertius iungens” orientation, we argue that network champions as a third party will be most essential in facilitating cooperation among two organizations when domain similarity is not present as a driving force behind the relationships. In those cases where organizations are dissimilar, a network champion may need to act as the third who connects, which leads to the following final hypotheses.
Research Methods
Research Setting
This study is an examination of the case of SACYHN. SACYHN was founded in 2001 to facilitate more decentralized services for children and youth and to address the problem of fragmentation in the delivery of health services for children. The mission of SACYHN has been to use the collective resources and expertise of participant organizations to advance high-quality, coordinated programs and services for children, youth, and families.
At its inception, network leaders decided to define health in the broadest sense; not as health care, but as health and well-being. Thus, to address child and youth health and wellness, an intersectoral perspective was needed, building respect and collaboration across organizations in multiple child-serving sectors. These included both public and nonprofit organizations in physical health, mental health, education, social services, and justice. In addition to the cross-sectoral focus, importance was placed on coordinating services across geographical regions. The founding of the network was announced as part of the funding of a children’s hospital in Calgary, with the network seen as reinforcing a mandate of the children’s hospital to offer specialized health services to children and youth throughout the Southern Alberta region.
The purpose of SACYHN was not to provide services or create a new service system, but rather, to strengthen the existing connections between service organizations and to create new linkages. The work of the network was to look for opportunities to facilitate linkages and connections through which best practices could flow as well as ideas on how to better coordinate services. Working groups were formed to undertake specific SACYHN projects/strategies and any organization or individual could participate in the working group. Some of the working groups included the Child & Youth Advisory Council (CAYAC), Interregional Children’s Mental Health, Obesity, Family Youth Participation, and Education. Certain initiatives and workgroups implemented strategies to improve services, like the mental health workgroup; whereas other workgroups like Education worked to create opportunities for linkages between providers through professional development opportunities or linkages between providers and clients as with CAYAC.
SACYHN’s governance model resembled the Network Administrative Organization (NAO) model proposed by Provan and Kenis (2008). A steering committee consisting of a subset of network members acted as a board of directors and was responsible for the setting of policy and planning decisions. The actual operations of SACYHN were managed and coordinated by the SACYHN staff, which consisted of a full-time director and several full-time staff members. Though the SACYHN staff were officially employed by one government agency, the Calgary Health Authority, this arrangement was an in-kind resource contribution and the main responsibilities of the staff was to be administrators of the network and not the employing agency. One third of SACYHN’s funding came from the financial resources committed by participating organizations, while the remaining two thirds funding was from the Calgary Health Authority.
Since Southern Alberta is a large region, the SACYHN network was divided into four smaller regions consisting of three rural and one urban (Calgary) region. These regions are geographically distant, covering the entire lower half of a province that spans over 250,000 square miles. The network members in the four regions were formally organized into subnetworks, whose members were to represent the needs of that region and to serve primarily in an advisory capacity to the SACYHN Steering Committee. The official tasks of the regional network representatives and the Steering Committee were to engage partners in the four regions, extend the impact of SACYHN initiatives, and contribute to a seamless system of care.
Between the Steering Committee, the regional committees, and the working groups the network consists of various players: parents and youth, ministries, organizations, regional authorities, First Nations (indigenous populations), and universities. These players represent all public sectors serving children and youth (health, education, children’s services, and justice), to address children’s health defined in the broadest sense. All of these agencies and individuals voluntarily participate in the network with the goal of focusing on optimizing the health and well-being of children and youth. Thus, SACYHN is an example of leveraged government (Kettl, 2008). Most of the member organizations are traditional public organizations working to serve the children and youth; however, to better address child and youth health and well-being, the need for a multiorganizational strategy was recognized by network members as a better way to approach this larger complex problem.
Data Collection
The collection of data on SACYHN occurred between September 2008 and March 2009. Due to announcements of a possible reorganization of the entire province’s health system, an effort was made to collect the data before any system-wide changes were implemented. A complete reorganization of the health system began in January 2009, which placed SACYHN in limbo until its formal disintegration by summer of 2010. During the 8 years of its existence, SACYHN implemented several successful initiatives and perceptions of impact, both of network members and external stakeholders, were highly positive (Provan and Lemaire, 2012). However, the intent of the system-wide changes to the health system were to centralize the previously decentralized system, which meant that regional health authorities no longer had the authority to support the network. Thus, the disintegration of SACYHN is not an indicator of failure, at least in regards to the success of its efforts, but rather of an environmental shift that presented a discordant context for a network approach.
The data on SACYHN before its dissolution were collected using an organizational questionnaire and through a series of focused interviews. The data collection effort was bounded by the formal structure of SACYHN; any organization with representation on the Steering Committee, a regional subnetwork, or a working group was asked to respond to the questionnaire. The total number of organizations initially contacted was 53. The total number of respondents surveyed was 137 since multiple individuals were asked to respond to the questionnaire for those organizations that were very large. The actual organizational response rate was 88% (42/48 – 5 organizations did not have a respondent identified or no longer existed), while the individual response rate was 76%. As the unit of analysis for this study is at the organizational level, the individual responses to the questionnaire on behalf of organizations were aggregated for each organization. Many of the organizations were very large organizations and so acquiring multiple responses for these large organizations was considered to be important as no one individual would be aware of all of the organizations’ network ties. The aggregation process varied by the type of data, details of which are provided in the discussion below.
The questionnaire was one adapted from network research by Provan and colleagues (cf. Provan & Milward, 1995). There were three main components of the questionnaire: organizational demographics, questions regarding organizational ties (i.e., network relationships), and perspectives regarding the impact of SACYHN. Measures for the analysis of this particular investigation are drawn only from the first two main components of the questionnaire; network impact was not included in the analysis of the importance of ties to network leaders for securing essential effort.
The first main section of the questionnaire involved specific questions about the organization, such as its general staff and budget size and the proportion of its resources that were devoted specifically to children and youth services. In addition, respondents were asked to break down the percent of their resources that were spent on specific children and youth areas (i.e., mental health, physical health, education, recreational services, etc.) totaling to 100%.
The second main section consisted of the network relationship questions. To determine the whole network structure, respondents were provided with a matrix listing all 53 organizational members of SACYHN plus one “organization” added in to represent the NAO staff. Respondents were asked to identify which of six types of activity links (if any) their organization had with the other 53 organizations over the past year. More specifically, respondents were asked to
Please go through the list and indicate which ones your organization has been involved with for the provision of services, programs, or activities related to child and youth clients over the past 12 months for each of the types of relationships listed. Please only consider those relationships that have value to your organization and are more than incidental and infrequent (i.e., not just emails, large meetings, occasional phone calls, etc.).
The types of activity links were derived from those that were deemed by network staff, with input from the Steering Committee, as the most important activities to accomplishing the work of SACYHN: strategic planning, shared resources, service delivery, education, research/evaluation, and information sharing. Definitions were provided for each type of activity on the questionnaire. If there was more than one respondent for an organization, relationships with other organizations were recorded as long as one respondent indicated a relationship. For instance, if Respondents 1 and 2 represented Organization A, but Respondent 1 indicated a tie with Organization B and Respondent 2 did not, it would still be recorded that Organization A indicated it had a tie with Organization B. The rationale for using this approach is that not all individuals would have been aware of the working relationships their organization had with all other organizations. The various respondents represented different program areas, management levels, or network involvement and thus, may have maintained relationships with different organizations.
In addition, in the last column of the matrix, for each existing relationship respondents were also asked to rate the overall quality of the relationship. Quality of the relationship was defined as confidence, which is a key dimension of trust (Luhmann, 1988; Mayer, Davis, & Schoorman, 1995; Nooteboom, Berger, & Noorderhaven, 1997). More specifically, respondents were asked,
Please base the rating of the quality of the relationship on how confident you are that the organization will do what they say they will do in its dealings with your organization, based on your expectations, and not just focus on the needs of their own organization.
A 5-point Likert scale was provided for rating relationship quality with scores ranging from 1 = poor relationship to 5 = excellent relationship, with the midpoint of 3 = good relationship. If there were multiple respondents for an organization, the individual responses for relationship quality were averaged across the organizational respondents as an average would provide a better indication of the overall perception of the quality of the relationship as perceived by different respondents.
In addition to the network matrices, there were four additional network relationship questions. Respondents were first asked about the organizational relationships they believed were most critical, which five organizations they believed were most influential, five they believed to be most similar to their own organization, and five they most admired for their work related to children and youth.
In the last stage of data collection, elite interviews were conducted with SACYHN staff and key individuals in the system, to try to ascertain the key role of leadership in the network. The list of individuals who were contacted for an interview was developed using strategic sampling. That is, individuals representing different sectors, different organizational levels, and different levels of involvement in the network (i.e., core/periphery) were identified, to get a representative sample of opinions across the full network. The interviews were conducted by three different individuals, so a list of questions was used as a guide. The goal of the interviews was to gain a better contextual understanding of the network, its operation, and the interviewee organization’s role in the network, so the interview protocol was used only as a guide. Some of the interviews involved multiple interviewees resulting in a total of 25 individuals who were interviewed during 16 interviews, which lasted between 30 to 75 min.
Measures
The two dependent variables derive from the two components of securing essential effort as defined by Barnard (1938), as the work necessary to first bring persons into cooperative relationships with the organization and second, to elicit services from that person. As discussed earlier, a network is based on cooperative relationships among members, rather than with a separate entity as in the case of organizations. Bringing organizations into cooperative relationships relates to the quality of the relationship between organizations, and whether organizations are contributing to the relationship relates to the extent of the collaboration between network members. Therefore, the two variables used to operationalize securing effort are the quality of the network relationships and the extent of collaboration.
Data measuring both of these variables come from the relationship matrix in the organizational questionnaire. Though organizations probably choose to collaborate with those organizations with which they have a good working relationship, this may not always be the case. Especially in the context of a network comprised mostly of public agencies, coercive pressures are likely to be present (Frumkin & Galaskiewicz, 2004). Organizations may have little choice but to work with certain other organizations, though one organization may in fact be frustrated by the lack of effort on the other organization’s part. Still, there is likely to be a high correlation between the extent of collaboration between organizations and the quality of the relationship, figuring out which comes first has been a dilemma in the network literature for many decades. The point here is not to attempt to address this chicken and egg dilemma, but rather, to examine whether ties to leaders are important for both facets of securing effort.
The first dependent variable, the quality of the working relationship, was measured by the item in the relationship matrix of the organizational questionnaire asking organizational respondents to rate the quality of the relationship between their organization and the organizations with which their organization had a relationship. As relationship quality was defined in the questionnaire as “how confident you are that the organization will do what they say they will do in its dealings with your organization, based on your expectations, and not just focus on the needs of their own organization,” then this dimension of relationship quality aligns well with the idea of eliciting effort from the organizations, in assessing how confident the organization is that their partner will follow through on their commitment to the relationship. Relationship quality was rated by organizational respondents on a 5-point scale ranging from 1 (poor relationship) to 5 (excellent relationship). The quality of the dyadic relationship was then confirmed. The two relationship quality values reported by both organizations in a dyad were averaged and this averaged value was used as the measure of the relationship quality of the dyad. Thus, if Organization A rated the quality of the relationship with Organization B as a 4, but Organization B rated the quality of the relationship with A as a 3, then the relationship quality was confirmed as a 3.5. As the purpose of this analysis is not in examining the differences in the perceptions of the quality of the relationship, using the average value is a more conservative approach. 1 This process increases the reliability much like confirmation does with the reporting of relationships, as it helps in offsetting the variation in how respondents rate items.
The second dependent variable relates to the effort the organizations put into their relationships with other network members. The extent of collaboration between organizations was measured by the multiplexity of the tie between dyads. How to measure collaboration is an unresolved issue in the public management literature (Thomson et al., 2008). We are measuring the extent of collaboration through the multiplexity of the relationships, which is a measure of how robust a tie is (Provan and Lemaire, 2012). We use multiplexity to operationalize the extent of collaboration among network members because the greater the multiplexity of a relationship, the more activities a relationship is based on and thus, the greater the commitment of both organizations to the relationship. Specifically, the multiplexity of the relationships is measured by the total number of confirmed activities (information sharing, strategic planning, etc.) between each dyad, with possible values ranging from 0 (no relationship) to 6 (relationship present based on all 6 types of activities assessed in the questionnaire). As the relationships were confirmed using the minimization process, the extent of multiplexity for each dyad is the total number of confirmed activities by each partner in the dyad. Table 1 provides the descriptive statistics for all of the variables.
Descriptive Statistics.
Note. Number of observations = 1,482, total number of dyads between the 39 organizations (42 organizational respondents but SACYHN staff and the two regional leader organizations were removed from the dyadic variables).
The independent variables included in the various models consisted of six measures, two for similarity, two for network champions, and two interaction terms. The first measure of similarity, “norm similarity,” was respondents’ perception of similarity based on similar norms as indicated in the organizational questionnaire. The item in the questionnaire asked respondents to nominate up to five organizations with which they believed had “norms, values, and ways of working that are most similar to yours.” This original variable was then transformed using the minimization process to indicate dyads of organizations that perceived themselves to be similar. A 1 indicates that both organizations of the dyad nominated the other organization as similar to their organization and 0 for all other cells.
The second similarity variable, domain similarity, was based on organizational attributes, rather than perceptions as with norm similarity, and is a measure of similarity due to regional location and service sector. Domain similarity was a concept often found discussed in early organizational (Thompson, 1967) and interorganizational research (Levine & White, 1961). Often, domain is categorized using slight variations of three categories: services, clients, and sector. In the context of the SACYHN network, sector is based on the services provided 2 and the division of clients by regions due to the health regions structure that existed before the restructuring of the health system. Therefore, our measure of domain similarity is based on service sector and region.
The actual variable, “domain similarity,” is a similarity matrix that was calculated using both service sector and region. Service sector was constructed based on the item in the questionnaire asking respondents to indicate the percentage of resources their organization spent on activities in regard to children and youth. The organization was then included in the service sector where it spends majority of its resources. There were five service sectors: health, justice, social services, education, and community services. A same service sector variable was compiled at the dyadic level; for each dyad, if both organizations were considered in the same service sector this was coded as 1, and coded as 0 if they were not in the same service sector. A variable was then constructed in the same way for regional location. Which region an organization was associated with was derived from its membership on the regional committees, and secondary data were used to place organizations not serving on regional committees. There were five regions, the four regions forming the basis of the regional committee structure and one region for those organizations based in the provincial capital located outside of the geographical boundary of the network, in Northern Alberta. For each dyad, if the two organizations were located in the same region, the variable was coded as 1 and 0 if they were not located in the same region. “Domain similarity” was then created by correlating the two matrices of same service sector and same region using the “Correlate columns across datasets” function in UCINET VI (Borgatti, Everett, & Freeman, 2002). This function produces a similarity matrix where the individual scores, ranging from −1 to 1, indicate how similar the organizations in a dyad are to one another while also considering their similarity to other organizations. 3 This variable, “domain similarity” therefore measures the extent to which organizations are similar to one another by simultaneously considering service sector and regional location.
Simmelian ties to network leaders were also measured through two different variables, representing two groups of network champions. The first group was the SACYHN staff, who were discussed as very important to the overall network in every interview. 4 In the words of one interviewee, “I would say it [the network] is dependent on them [SACYHN staff], their knowledge, their leadership, and their positions.”
The variable “SACYHN staff ties” is based on the particular dyadic ties between network members and SACYHN staff, which included regional network coordinators. These ties were not included elsewhere in the analysis, but the ties organizations had to the SACYHN staff was collected in the same way as the ties among organizations. The actual values for the variable are the minimum number of shared activities as the basis of the ties that both organizations of the dyad had with SACYHN staff. This way of operationalizing the Simmelian tie as valued, that is, the level of multiplexity of the tie, incorporates the strength of the tie dimension and allows for the measure to account for variation of the strength of Simmelian ties to network champions across the many different pairs in the network. 5 The calculation process was the same minimization process used for confirming the network ties between all other network members. So first, the multiplexity of each dyadic relationship between network members and SACYHN staff was confirmed, that is confirmed at the minimum number of activities indicated by the network member as well as by SACYHN staff. This first step captures the confirmed multiplexity between each network member and SACYHN staff. Then, to transform this into Simmelian SACYHN staff ties, for each pair of network members, their mutual tie to SACYHN staff was based on the minimum confirmed number of relationships either of the pair had with SACYHN staff. For instance, if organization A had a confirmed relationship with SACYHN staff based on 5 shared activities, but organization B had a confirmed relationship to SACYHN staff that was based on only 3 shared activities, then the value of the Simmelian tie to SACYHN staff for that particular dyad was set at 3.
The other network champion variable, “regional champion ties” was the same as ties to SACYHN staff, except that rather than the ties organizations had to the SACYHN staff node, it is based on the ties network members had to two key regional organizations. These regional champions were selected because the leadership role they played in the network overall and in their regions was specifically noted in the interviews, both in reference to the organizations and to specific individuals representing those organizations. One of these regional champions was the organization that was the driver behind the network and so its role in securing effort had been discussed in all of the interviews with SACYHN staff and with all the regional representatives, either serving on regional committees or the Steering Committee. The other regional champion was specifically discussed in two of the SACYHN staff interviews and in all of the interviews with regional committee members, except in the interview with the representatives of that organization. In the interviews with the SACYHN staff and with the regional committee members, emphasis was given to the importance of that regional champion, especially in making the comparison between the leadership present in that region and lacking in other regions:
It’s knowing who to start with and knowing the culture, and knowing it works different in different regions. In one of the regions we have nothing, and it has everything to do with who you talk to and whether they rise to the challenges. In [one particular region] it has only worked because of the leadership at the far end. In one of the other regions, we have no buy-in and no interest. There is a subtlety in the leadership and it’s not just at the frontline, it’s also the senior support supporting them. (SACYHN Staff)
The two organizations perceived as regional champions were also two organizations repeatedly playing a key role in the structure of the network, either due to a central position in the overall network or to an important brokerage position. 6 Therefore, between the role these organizations played structurally and the importance placed on them by interviewees, these two organizations were clearly important network champions.
The process used for creating the Simmelian tie to regional leaders variable, “Regional champion ties” was the same as “SACYHN staff ties” except that relationships to the two regional champions were eventually merged. The maximum value of the confirmed ties dyads had to each of the regional champion organizations was taken, indicating the highest number of shared activities connecting the dyad with either of the regional champions. We combined the mutual ties to the two regional champions into one matrix because very few dyads would have a tie to both regional champions. Since these champions played important roles in the regions, most of the mutual ties were regional based. Our interest is in capturing the importance of these regional roles, rather than the role of each champion specifically. 7
Finally, to test Hypotheses 4a and 4b, whether ties to network champions are a greater factor under conditions of dissimilarity, the variable “domain similarity” was truncated to include only the values for dissimilar dyads. This variable “domain dissimilarity” captures the extent of dissimilarity between dyads that are not similar and treats all other dyads as equal to “0.” Interaction terms were then constructed to test the interaction between the level of dissimilarity and ties to network champions, to examine whether the extent of dissimilarity conditions the effect of Simmelian ties to network champions. Using the “Make Interaction Term for Regression” function in UCINet 6 (Borgatti et al., 2002) interaction terms between To test whether Simmelian ties to network champions are more important when organizations are dissimilar, dummy variables were created to compare the effect of ties to network champions when organizations are similar versus dissimilar. In these models instead of the domain similarity variable, the three dummy variables extracted from “domain similarity” were included. In addition, interaction terms were constructed to test the interaction between the level of similarity and ties to network champions, to examine whether the extent of similarity conditions the effect of Simmelian ties to network champions. Using the “Make Interaction Term for Regression” function in UCINet 6 (Borgatti et al., 2002), interaction terms between “domain dissimilarity” and each of the network champion variables were created. The standard variables were mean-centered in the process of calculating the interactions to prevent high correlation between the standard variables and the interaction variables.
Analysis
To test the hypotheses, Multiple Regression Quadratic Assignment Procedure (MRQAP) was used. Using standard statistical tools can be problematic when analyzing network data because the data often does not fit the criteria of the basic assumptions behind standard regression. Specifically, network data often does not conform to the assumption of independence among the observations which can lead to unreliable standard errors. To address this problem, Quadratic Assignment Procedure (QAP) was created and uses a two-step process to increase the reliability of the standard errors (Krackhardt, 1987). QAP produces correlations among matrices of data, but the same process underlying QAP was used to develop MRQAP. MRQAP is a two-step regression process, in which the first step is a standard regression model. The second step then involves permutations of the rows and columns of a matrix and estimation of the standard errors, creating a distribution of standard errors. The standard errors from the first step are then compared to the distribution of standard errors constructed from the second step to determine if the original standard errors are significantly different (Krackhardt, 1988). For the MRQAP, we used the Double Dekkar Partition Semi-Partialing method in UCINET (Borgatti et al., 2002) This method performs a regression on the residuals and independent variables and then on all the variables and residuals, thus a double regression to partial out the effect of collinearity amongst the independent variables. We used this method because it has been found to be robust under a variety of conditions either because of autocorrelation, spuriousness, or skewness (Dekker, Krackhardt, & Snijders, 2007).
MRQAP is similar to OLS, as it is appropriate for the analysis of continuous variables; however, all variables for MRQAP must be in dyadic form and square matrices are specifically required by the procedure we used. This structural requirement of the data is appropriate though because our research questions and hypotheses pertain to the dyadic level of analysis. The dependent variable is the relationship between two nodes and all independent variables have meaning in regard to the dyads. A total of four MRQAP models were performed as two different dependent variables were developed for this analysis. The first two models have relationship quality as the dependent variable and the second two models, multiplexity. Models 1 and 3 test Hypotheses 1a to 3b whereas Models 2 and 4 test the interaction Hypotheses (4a and 4b).
Before running the MRQAPs, correlation matrices of the variables were constructed using QAP. As indicated in Table 2, the two dependent variables, relationship quality and multiplexity, were highly correlated with one another, at 0.870. Table 2 also shows that several of the independent variables were positively correlated with each other and significantly similar, which is to be expected given the conceptual similarity of many of the variables. The Pearson correlation values are not so high however to prevent proceeding with the analysis.
Variable Correlations.
p ≤ .05. **p ≤ .01. ***p ≤ .001.
Results
The results of the analysis of relationship quality are shown in Table 3. All of the variables in the first model were significant, indicating that similarity and Simmelian ties to network champions are both significant factors in explaining the quality of relationships between network members. Specifically, supporting Hypothesis 1a and 2a, the more similar members of a dyad are to one another, either because of norms or domain similarity, the higher the quality of the relationships. Supporting Hypothesis 3a, the stronger the Simmelian ties to network champions, either to the SACYHN staff or to the regional champion organizations, also the higher the quality of the relationships. Though all three hypotheses are supported, the similarity variables are significant at the p < .001 level while the ties to network champion variables are significant at the p < .01 level.
MRQAP Models Predicting Relationship Quality.
Note. N = 1,482 (total number of dyads). MRQAP = multiple regression quadratic assignment procedure.
p ≤ .05. **p ≤ .01. ***p ≤ .001.
Model 1 provides support for a significant relationship between Simmelian ties to network champions and the quality of network members’ relationships with each other. As this article focuses, though, on examining under which conditions ties to network champions may be more important, Model 2 includes terms that interact domain dissimilarity and ties to network champions. The results of Model 2 are similar to Model 1, except now domain similarity is no longer significant and ties to SACYHN staff is significant at the p < .001 level. Of the two interaction terms, only ties to SACYHN staff is significant (p < .01); ties to regional champions for dissimilar domain dyads is not significant.
The results of the models with multiplexity as the dependent variable are provided in Table 4. These results are similar to the results for relationship quality, which is not surprising given the high correlation between the two dependent variables. The one difference between the two sets of results is that ties to SACYHN staff is significant at the p < .001 level in Model 3 whereas it was significant at the p < .01 level in Model 1.
MRQAP Models Predicting Multiplexity.
Note. N = 1,482 (total number of dyads). MRQAP = multiple regression quadratic assignment procedure.
p ≤ .05. **p ≤ .01. ***p ≤ .001.
All together, the four models indicate that the hypotheses were mostly supported. Hypotheses 1a to 3b were supported in that similarity and Simmelian ties to network champions do have a significant effect on the quality and the multiplexity of relationships. Hypotheses 4a and 4b were also supported, but only for one network champion group, SACYHN staff. When network members work in different domains, Simmelian ties to SACYHN staff is a significant predictor of higher quality of relationships and multiplex ties.
In summary, similarity is a strong predictor of both relationship quality and multiplexity, as was expected from network theory. Simmelian ties is a significant factor in general as well. Though the magnitude was smaller than for similarity, Simmelian ties to SACYHN staff and to regional champions were also significant predictors of relationship quality and multiplexity. Most important, though, is the support for our argument that Simmelian ties are especially important under conditions of domain dissimilarity. We did not find support for this argument in regard to regional champions, which we believe may partially be due to the narrower role that regional champions play in the overall network. In the interviews where regional leadership was discussed, the importance placed on that leadership was more general. For instance, as one regional committee member explained in referring to the role of the regional champion,
There are dedicated people prepared to participate in SACYHN and carry on the initiatives and they have all seen how it has made our work easier and better for our families. So they are starting to influence people in another department and so we are starting to move it [SACYHN initiative] out.
On the other hand, we did find support for our argument in regard to the network staff. As the following quote by a Steering Committee member indicates, the role the SACYHN staff, specifically the Executive Director went beyond influencing others to carry out the work of the network:
From my vantage point I think [Executive Director] is very responsive if she feels that there are those tensions coming into the meetings, or I’ve worked with her and there have been a couple of problems and she’s gotten right to that discussion. You know she hasn’t wasted any time in opening it up and dealing with it. I would say from my perspective they [SACYHN staff] are very good at doing that problem solving.
Because network members that were dissimilar were less likely to have high-quality and multiplex relationships with one another, the finding that strong Simmelian ties to network staff is correlated with higher relationship quality and multiplexity has important implications for securing essential effort. This finding offers some evidence that ties to network champions may have an important role to play in facilitating cooperative relationships among network members that are less interdependent, but that role also may depend on the type of network champion.
Discussion and Implications for Network Management
This study is an attempt to contribute to the research on network functioning by focusing on the importance of ties to network champions in securing the effort from network members necessary to achieve collective outcomes. In examining the dyadic and triadic structure underlying the whole network of SACYHN, we were able to investigate if and when Simmelian ties to network champions make a difference in securing essential effort from members of a dyad. We found that network members that are similar to one another, because of norm similarity or domain similarity, are more likely to have higher quality relationships and to collaborate with one another. We also found that strong Simmelian ties to network champions, either the SACYHN management staff or regional champions, were also associated with higher quality and greater collaboration relationships between network members. In addition, though Simmelian ties to regional champions was a factor in securing effort only in general, Simmelian ties to the network management staff was also significant when network members were working in different domains.
We have proposed that one of the primary aims of network leadership and management is to secure effort from network members. The findings reported here provide evidence to suggest that ties to network champions may make a difference in securing the effort necessary to achieve interorganizational collaboration and thus, further research is warranted to examine the relationship between ties to network leaders and securing effort. More research is especially needed to examine the conditions under which ties to network leaders matter the most, since in this study evidence for the importance to ties for only one group of network champions under conditions of dissimilarity was supported. Our understanding of securing effort can then be progressed by also recognizing when securing that effort will be more difficult, and thus may require nudging by network leaders.
Our findings overall do have important implications for both theory and practice. First, by focusing on the Barnardian function of securing essential effort, what emerges is the suggested role network champions play in getting organizations into cooperative relationships and assuring they contribute the effort necessary to sustain those relationships. By examining one of the functions Barnard proposed as necessary for achieving collective action, we were able to target the importance of ties to network champions for securing effort. Isolating the structural importance of Simmelian ties to network champions was made possible by concentrating not on the tasks that network leaders perform but why they perform the tasks—the aim of securing the effort necessary for collective action. By focusing the lens around this basic function, future research can compare our results to other network contexts in a way that is more generalizable than if we compare the numerous and various tasks that may be involved in securing effort across different networks.
In addition, the support we found for Simmelian tie theory and the role of the network champions as third-party influence is consistent with previous research and theorizing. Our findings are similar to the findings by Gulati (1995) on the role of indirect ties as a referral mechanism for bringing firms together in alliances. This influence of a third party on dyadic relationships is also as theorized by Granovetter (1973) that strong ties will lead to triadic closures and is consistent with the closure argument of Coleman (1988), that closed networks will provide a means to enforce norms of behavior. We extend that research regarding the role of third-party actors to network leaders specifically. Thus, strong Simmelian ties with network leaders provide the relationship structure for persuading network members that are otherwise unlikely to work closely together when left to their own means to realize the value of working together, as the following quote from a SACYHN Steering Committee member reveals:
[Executive Director] is really good . . . And so she’ll usually say we need you here today because we’re going to talk about such and such. So she’ll also use me that way. She knows when to play on certain people to be a catalyst to move people forward as well. It is two ways. She’ll come back to us and say I really would appreciate it if you could be there today because I don’t think people are going to get this and this is what we need to have happen, right? Or you need to ask those questions. So it’s not so much . . . yes we use it [the network] too, but I think with [Executive Director] there she’ll also know where to pick off . . . where she needs certain people to move something forward too. She knows who her players are and she knows where the strengths are. You wouldn’t have that otherwise.
In addition, our findings have implications for understanding when network leaders may have a greater role to play in achieving collective action. The significance of Simmelian ties to network managers in general and under conditions of dissimilarity could be interpreted as further support for the findings by Lemaire, Provan, Mercken, and Leischow (2017) on the positive effect of centralization on service delivery network effectiveness. Ties to a central actor, such as network managers may be important to securing effort in general, no matter the extent of similarity of organizations. However, the importance of these ties may not be as essential when organizations are similar enough, either because it is easier for them to collaborate in the first place because of the domains they work in or because there are similar norms holding the organizations accountable to one other. When network members are dissimilar, though, then certain network leaders may be essential in helping to elevate the extent to which organizations collaborate with one another. As described in one of the SACYHN interviews, it takes time to develop the collaborative relationships that are necessary to achieve the greater network goals, especially when there is less familiarity:
You need to have the right people that are like minded. A lot of people I worked with, we don’t worry about territory, we’re more concerned about outcomes. Because we already know each other we don’t have hidden agendas, we’re just trying to get to the end point here. But if you don’t know somebody, it takes that time of just figuring it out, and it becomes a little more territorial and less about what you are trying to achieve.
Under these conditions of less familiarity is where network leadership has a greater role to play, in facilitating the relationships that may not be as easily or as quickly to form. As described in another SACYHN interview with a regional network coordinator,
People will often approach me [Regional Network Coordinator] as the point person for the network and where do they go for this problem and I will then basically start with [SACYHN staff] and help connect the right people together. Even within the region, the same as well. The bureaucracy will get confusing and it’s not always clear who has responsibility for that particular area and so who should they talk to. So the regional table, as well as myself, provides the vehicle for people from the community to find who they can work with on a particular problem. I get more requests over time with how do I start with X and I help them figure out how to move that forward.
Strong Simmelian ties to network champions provide a means through which network members can be persuaded that working together is valuable to the overall outcome. This persuasion may be necessary under conditions of uncertainty, as in the quote above, or under conditions where one party is less motivated to work with the other. For instance in the case of SACYHN, persuading the Health sector to work with the other service sectors was perceived as an area where network leadership had been key:
If you don’t have other systems saying to Health, in a really strong way, “you can’t do this; you still need to work with us, we still need to be at your table and you need to be at our table.” And they need help to be able to say that, they need ammunition to be able to do that and in some ways, that is the role of the secretariat. (SACYHN staff)
Our argument here is not to suggest that all network members should be connected to one another to secure essential effort; relationships among all network members may not be necessary or feasible. Rather, the role of network leaders may be to facilitate certain necessary relationships and ones which are less likely to happen serendipitously, specifically when organizations are more independent of one another. Network leaders have to divide their efforts between facilitating serendipitous interactions and targeting the development of certain ties, which Paquin and Howard-Grenville (2013) refer to as “blind dates” and “arranged marriages,” respectively. Our argument is rather, that where network orchestration may be necessary (i.e., arranged marriages), more nudging by network leaders may be essential if those network members are less likely to recognize the interdependency of their actions.
This research, therefore, also has practical implications as it can help network leaders in attaining the effort necessary to achieve collaborative advantage. Understanding that the role of network leader in getting cooperative relationships among organizations depends upon the type of interdependence between organizations can help network leaders know where in the network they need to focus their efforts. Also, the idea that this role is the role of network managers is not necessarily true. Though network managers may surely have a leadership role to play in securing effort among members that are less interdependent, as they did in the case of SACYHN, other leadership may be as effective in influencing similar agencies to work together. Taking some of the burden off the network management staff, network members in place to be leaders could fulfill this leadership role as effectively as network managers, allowing network managers to focus on the relationships that require more “arranging.”
This analysis is, of course, not without its limitations. The direction of causality is a clear limitation. Longitudinal research on the case would help determine causality, specifically if network leaders are making a difference in the extent of collaboration among organizations and the quality of the relationships, or if the relationships found in this network are only associational and represent what naturally occurs over time in working within a network. Also, though the unit of analysis was dyadic, these dyads are embedded in one case. Our results were compared with randomly simulated networks with a similar structure, but the aim of this work is analytic generalization and not statistical generalization (Yin, 1989). Our goal was to examine whether our results would conform to the application of Simmelian tie theory to understanding the importance of networks leaders in a goal-directed network. Our results offer only enough empirical evidence to suggest that Simmelian ties to network leaders made a difference in securing essential effort in the particular case of SACYHN, but finding evidence to support our argument indicates a worthy avenue for future research. Future work should examine and test whether network leaders do, over time, impact the extent of collaboration and quality of relationships between network members and whether their importance is indeed greater when interdependency of network members is taken into account.
Future work should also examine how different context from that of our case may make a difference of when network leadership has a greater impact. In the particular context of SACYHN, the more similar network members were, the more likely they were to have high-quality multiplex relationships. In other contexts, though, similar organizations may compete with one another. If organizations are too similar then competition may become an issue (Oliver, 1990). Therefore, in other network contexts, it may be in those cases of high similarity where network leadership becomes more essential. Alternatively, it may not be a matter of domain similarity, as was the focus in this study, but rather the power dynamic behind the interdependency that may render network leadership more essential. Casciaro and Piskorski (2005), in refining Resource Dependency Theory, distinguished between interdependence that is based on mutual dependence and power imbalance. They found that interdependency based on mutual dependence was a lubricant for mergers and acquisitions whereas power imbalance was an impediment to the formation of mergers and acquisitions. Thus, in some network contexts, as well as that of SACYHN, it may be that ties to network leaders are important in overcoming the tensions created by power imbalance.
Despite these limitations, this work offers a start to better understand why and when network leaders have an important role to play in ensuring organizations in a formal, goal-directed network are contributing effort to the network. With the often stated importance of goal-directed networks for addressing wicked problems, the case presented here is an example of the leveraged government approach imperative to the public administrative agenda (O’Toole, 2015). We have presented an examination of this case based on Barnard’s function of securing effort and Simmel’s third-party tie argument, concepts that can be applied to other goal-directed networks. In addition, we linked these concepts to a structural analysis of ties to network leaders, an analysis that can be performed similarly to examine goal-directed networks in different contexts. Therefore, we believe our work offers theoretical and practical insights on the value of network leaders and where the effort of these leaders may be most essential in securing the effort necessary to attain the benefits of a network approach in general, beyond the application to our case of SACYHN. Aligning multiple players across many boundaries of action means aligning many different organizations with varied levels and forms of interdependencies. Securing the effort it takes for these multiple players to work together just may be the essential role of network leaders in making a multiorganizational approach to contemporary public management work.
Footnotes
Acknowledgements
We would like to thank Brint Milward, Janice Popp, Anne Casebeer, Carol Adair, and all of the Southern Alberta Child and Youth Network (SACYHN) coordinators for their help with the formulation of the research and the collection of the data. We would also like to thank Adam Eckerd for the invaluable advice he provided as well as the hepful comments and suggestions of four anonymous reviewers for their useful and detailed comments and suggestions on previous drafts.
Authors’ Note
Keith Provan passed away after contributing to the draft of this manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Southern Alberta Child and Youth Health Network provided partial support for research expenses.