Collaborative Governance and the Challenges of Network-Based Research

Abstract

We review the literature examining collaborative governance processes from a network perspective and evaluate the extent to which it tackles important conceptual and methodological challenges. In particular, we assess whether scholars clearly identify the boundaries of the network, define nodes and the nature of ties, and examine how they deal with missing data, account for tie strength, take tie multiplexity into account, and study networks over time. We discuss the implications of our findings for the collaborative governance literature and ways to address the shortcomings in existing research.

Keywords

collaborative governance network analysis nodes ties network boundaries

Introduction

Research on collaborative governance in public policy, administration, and management has experienced a sustained growth in recent years, as scholars have sought to disentangle the complex array of relationships among myriad variables that affect the propensity of policy actors to engage in collaborative behavior, as well as the outputs and outcomes of collaborative governance (Ansell & Gash, 2008; Emerson et al., 2012). One of the common traits of the foundational work in this area is repeated reference to the importance of networks in promoting and sustaining collaboration among actors who do not always have the same goals or interests when participating in public arenas (Amsler, 2016; Lecy et al., 2014; Vangen et al., 2015). In general, the literature has espoused the idea that collaborative governance involves structures of interactions among actors (i.e., networks) that span sectoral, hierarchical, and geographic boundaries (Scott & Thomas, 2017).

Scholarly interest in interactions among policy actors has spurred diverse applications of network-based approaches for understanding collaborative governance. However, we remain relatively ignorant about exactly how scholars examine networks when they study collaborative governance. Against this background, we review the literature on collaborative governance that explicitly focuses on and measures networks. Doing so can enrich discussions on the strengths and pitfalls of research undertaken to explain how patterns of interactions among policy actors shape collaborative governance outcomes.

In this article, we aim to document how collaborative governance scholars who examine networks formally (i.e., observe and measure one or more properties of the network) approach conceptual and methodological decisions that have important implications in network-based research. Specifically, we seek to answer the following three questions:

How do researchers define the boundaries of the networks they study (including which actors participate in them), and how do they deal with the common problem of missing data in network studies?

How do researchers measure the ties that connect network participants, and to what extent do studies account for tie strength and multiplexity (i.e., multiple types of relationships)?

To what extent do researchers examine the evolution of collaborative governance systems?

Answering these questions is important first because the literature in collaborative governance and networks has experienced a rapid development in the last two decades, yet it remains unclear whether scholars publishing in this area abide by “best practices” commonly shared among social network analysts. Second, the issue merits attention because conceptual and methodological choices have important implications for the interpretation of results of network-based research on collaborative governance and beyond (Bergenholtz & Waldstrøm, 2011). Third, such an exercise can help us highlight ways in which the field of collaborative governance studies may move forward (Isett et al., 2011; Lecy et al., 2014). Our review is designed to encourage researchers to think carefully about how to improve research design in collaborative governance studies that invokes the importance of networks as a conduit for collaborative behavior.

The next three sections explain in more detail why these three questions are central to network-related research, and why asking and answering them is important for the study of collaborative governance. We then describe the methods we use for our review and present and discuss our results. We conclude with a summary of key findings and highlight important directions for future research on collaborative governance and networks.

Who Participates in a Collaborative Governance Process?

To understand collaborative governance processes, a first critical task for researchers is to clearly identify the set of relevant actors. From the perspective of networks, researchers need to precisely define the so-called network boundaries (Wasserman & Faust, 1994). Failing to clearly establish these boundaries can have noxious effects on findings. For example, if important actors who are part of the network are left out of the analysis, the internal validity of findings that purportedly assess how the whole network operates will undoubtedly be threatened. The external validity of findings will also be hampered if network boundaries are not clearly defined, as generalizations of findings to other cases will be necessarily off-target.

Of course, defining the boundaries of the network is not always easy. In some cases, the set of actors that form the network is a relatively bounded population, as when scholars are able to rely on publicly available exhaustive rosters of participants of collaborative governance systems (Emerson & Nabatchi, 2015; Scott & Thomas, 2017) or policy forums (Fischer et al., 2017; Fischer & Leifeld, 2015). But, in many cases, such simplicity can be elusive, and researchers will need to spend time carefully considering how network boundaries are defined, which may depend on both theoretical and case-related criteria.

Often, defining who exactly is part of a network is an empirical challenge that can be addressed in different ways. Laumann et al. (1989) describe two possible approaches. The first one is the realist approach (see Wasserman & Faust, 1994 for more discussion on this approach), which simply consists of extracting the information about the network boundaries from some subset of study participants who can volunteer that information for the researcher. For example, a researcher interested in the study of collaborative governance on transportation policy in a metropolitan area could initially identify a number of actors that unequivocally belong to the network (the head of the local transportation department, representatives of the local public transit provider, etc.). Then, using a “snowball” technique, the researcher could iterate between inquiring about other relevant actors and recruiting those actors as study participants until no new actors are nominated, indicating that the study's participants comprise the network. Importantly, this approach requires that study participants collectively consider themselves to be members of a “social entity” (Laumann et al., 1989, p. 65), which defines the boundaries of the network of interest to the researcher.

A second approach is the so-called nominalist approach, in which nodes are identified as part of the network given the analytical purposes of the researcher. For instance, continuing with the previous example of collaborative governance in local transportation, the researcher could decide to define the boundaries of the network by incorporating to the study only the actors that are mentioned in a specific local transportation plan or document. Regardless of what approach is used, determining the boundaries of the network should be a basic first step for researchers studying networks in the context of collaborative governance.

A closely related issue is the precise definition of what type of entity qualifies as an actor in a collaborative governance system, or a node in the network. Actors can be individuals, organizations, parts of an organization, or a mix of all these types of entities (Huxham et al., 2000). Defining the participants in a collaborative governance system is a precondition for understanding collaboration among actors (Emerson et al., 2012, p. 11). From a network perspective, clearly defined nodes enables assessment of the structural properties of the network, which in turn affects the scope of insight on how collaborative behavior unfolds (Scholz et al., 2008).

Finally, related to the substantive issue of network boundary definition is the topic of missing data and its potentially negative effects on the analysis of collaborative governance systems. As we explained above, lack of attention to network boundaries can invalidate findings. But even when network boundaries are properly defined, the absence of data on nontrivial numbers of actors and ties can bias results (Kossinets, 2006; Smith & Moody, 2013). Particularly when researchers use surveys to collect data to gauge how interactions occur, missing data can be a rather critical issue, as high enough levels of nonresponse can lead to very unreliable estimates of network-level statistics (Burt, 1983, 1987). For example, centrality measures can be valuable to determine activity and power levels in networks, as well as influence and control overflows of information—depending on the type of centrality measure that the researcher uses. However, network centrality measures can be very unstable in the presence of missing data (Costenbader & Valente, 2003). Thus, scholars interested in the study of collaborative governance systems from a network perspective must make extra efforts to minimize the amount of missing data and/or mitigate its effects through specialized analytical techniques.

What Interactions Structure Collaborative Governance Processes?

Collaboration is not a unidimensional concept and can be indicated by different types of interactions. Examples abound. Bardach’s (1998) classic description of collaboration among governmental agencies, for instance, includes seven different types of collaborative behavior that could be analyzed as distinct ties in a network (exchange of information, joint litigation efforts, etc.). Similarly, Ansell and Gash (2008, p. 558) define five steps of the collaborative process (dialogue, trust, commitment, shared understanding, outcomes), each of which can by itself indicate the presence of collaborative behavior. Others observe that actors in a network may be connected by different types of ties that may indicate collaboration, including the joint participation in policy venues or forums (Berardo & Lubell, 2016; Shrestha et al., 2014). A final example is Emerson et al. (2012, p. 6), for whom collaboration involves principled engagement, shared motivation, and capacity for joint action (in addition, all three have subcategories). Each of these three dimensions can require different types of interactions, all of which can be thought of as indicators of collaboration. For instance, principled engagement requires face-to-face deliberation to jointly define problems, and scholars have attempted to carefully measure it (e.g., Ulibarri & Scott, 2017). Yet, such deliberations are rarely simple or represented by single types of network interactions. Social interactions can exist on a variety of planes, and each realization of a social interaction aggregates to form a discrete relationship. Furthermore, some of these multiplex ties influence each other, as when venue co-participation leads to more intense exchanges of information (Fischer & Sciarini, 2016; Leifeld & Schneider, 2012).

To complicate matters further, relations can have different intensities or strengths. Information exchange, collaboration, and cooperation are sometimes distinguished as representing different degrees of collaboration intensity (Margerum, 2008). Furthermore, Ansell and Gash (2008, p. 546) mention that collaboration requires strong communication or multilateral deliberation, thus emphasizing the importance of strong connections among actors. When possible, scholars need to understand how these complex relationships unfold and co-evolve, as they are collectively important to better understand the inner working of collaborative governance processes.

It is also important to note that multiplexity can depend on how nodes in a network are defined, as discussed in the previous section. For example, while two individuals might share a relation of friendship, their respective organizations may additionally be linked by ties of project collaboration. While a multimode network (i.e., with multiple types of nodes representing individuals and organizations) would not be by definition a multiplex network, aggregating individuals within organizations as unique entities (a common practice among collaborative governance network scholars) would result in a network featuring multiplexity (i.e., with one inter-organizational tie defined by friendship and another by project collaboration). Again, mutual dependencies and influence between these two types of relations are likely to exist (for a more elaborate discussion of interpersonal vs. institutional/structural ties, see Marsh, 1998).

How Do Collaborative Governance Processes Evolve Over Time?

Emerson et al. (2012) claim that collaborative governance processes and actions therein are nested within a general “system context that produces opportunities and constraints and influences the dynamics of the collaboration at the outset and over time” (Emerson et al., 2012, p. 5). They argue that some of the opportunities and constraints from political, legal, socioeconomic, or environmental influences are the emergence of leadership, incentives for individual actors, and uncertainty, which can affect the way the collaborative governance regime performs in the long run, particularly as new possibilities and challenges arise (Emerson et al., 2012, p. 9). Other scholarship also highlights the fact that networks evolve as exogenous forces operate on them. For example, transitions of the legal framework from a monopoly to a liberalized economic sector (Fischer et al., 2012), an external environmental shock (Fischer, 2015), or the fact that an issue gains traction on the political agenda (Ingold et al., 2017) can all affect how networks evolve.

Networks also evolve over time due to actors’ strategic behavior, independently of the evolving context. Actors cultivate relationships to increase the likelihood of achieving their goals, constrained by formal authority and existing relations, costs and uncertainties, as well as limitations imposed by cognitive capacities (Scholz et al., 2008; Snijders et al., 2010). They tend to connect to others that are already powerful and central in the network (Stokman & Zeggelink, 1996), or they reciprocate ties, close structural holes, or create transitive ties based on the types of risks they face (Berardo & Scholz, 2010).

In addition, the importance and the purpose of collaboration in networks might change as the collaborative process itself unfolds (Huxham et al., 2000, p. 345). For example, Johnston et al. (2011, p. 704) claim that collaboration early in the process is what matters for understanding outputs that are produced later. Early collaboration fosters trust and reciprocity, upon which further collaboration can be built in a sort of positive feedback process that increases the likelihood that actors find negotiated solutions to joint problems (Berardo, 2010). In sum, analyzing a collaborative governance system over time is important given that actors’ interactions are not stable over time, and that these dynamic changes might be crucial to assess outputs of collaborative governance, as well as reactions of the collaborative governance process to changing context conditions.

Review Method

To identify articles focusing on collaborative governance and networks, and analyze their content to answer our set of three research questions, we conducted a search in Scopus based on the following key terms in English: “Collaborative Governance” or “Collaboration” and “Governance,” plus “Network” or “Networks.” These terms had to appear in the title, abstract, or keywords accompanying the articles. We identified all articles published between January 2000 and May 2019 in journals listed in the “Social Sciences & Humanities” category.¹ The search produced a total of 472 articles, of which we discarded 133 as they either were clearly unconnected to the topic of collaborative governance or were not empirical studies (i.e., we excluded reviews, conceptual papers, and perspectives). Consequently, the universe of articles we analyzed is formed by the remaining 339 articles.

It is interesting to note that not all of these articles are strictly about collaborative governance. In fact, a majority of them (199) are what we consider to be “loosely” about collaborative governance, while a minority (140) are “strictly” about collaborative governance (Figure 1).

Figure 1.

Types of articles analyzed.

Articles that are “strictly” about collaborative governance explicitly mention the term “collaborative governance” somewhere in the text and tend to adopt conceptual perspectives on governance consistent with Ansell and Gash (2008). Articles that are “loosely” about collaborative governance do not use the term explicitly, yet are concerned with collaboration and governance and pay attention to the role of networks to analyze the former.^{2
3}

Figures 2 and 3 plot the number of articles on collaborative governance and networks published since the year 2000. The first finding worth mentioning is that for the first 8 years of the panel (2000-2007), the number of published articles was very low, with less than 10 per year. This starts changing in 2008, when an upward trend becomes clear, peaking, in 2015, with 50 articles. Although the number of articles declined 2 years in a row (2016-2017), the overall trend indicates an increasing volume of research over time.⁴

Figure 2.

Number of articles published yearly since 2000.

Figure 3.

Number of articles published yearly since 2000 (disaggregated by type of article).

Figure 3 simply disaggregates the results of Figure 2 to distinguish between articles that are strictly about collaborative governance versus those that are loosely about collaborative governance.

We observe two noteworthy findings. First, in general, there have been more articles that are loosely about collaborative governance than those that are strictly about collaborative governance. Of the 19 years plotted in Figure 3, only in 3 do articles that are strictly about collaborative governance exceed those that are only loosely about it. Second, despite the general dominance of articles loosely about collaborative governance, there have been more articles published that are strictly about collaborative governance in the last several years (2017-2018, as well as the first 5 months of 2019, which is not plotted). It is not possible with such few data points to argue that this is a definite upward trend, but, at least, this finding hints at the fact that scholars who study collaboration, governance, and networks may be increasingly more likely to explicitly embrace the collaborative governance terminology as they do so. We believe this bodes well for this field of study, as a profusion of new scholarship will undoubtedly result in both theoretical and methodological advancements in the study of collaborative governance from a network perspective.

Before analyzing how articles assess the network-related variables of interest, it is important to note that a majority of the articles on collaborative governance and networks that we identified do not empirically examine networks, but rather treat them “metaphorically” (Dowding, 1995). In other words, most articles describe networks as proxies for collaboration, or, in some cases, as structures of interactions that facilitate the flow of resources that are needed to build and sustain collaborative relationships. Yet, they do not actually measure networks and their components (i.e., nodes and interactions between them). This precludes us from answering the questions we posed at the beginning of the article. Thus, we focus our analysis on the articles that do, in fact, observe and measure networks and their components. Out of the 339 articles we examined, only 73 measure a network or networks explicitly, as shown in Figure 4.

Figure 4.

Distribution of articles based on whether they measure networks or not (disaggregated by type of article).

Out of these 73 articles that explicitly measure networks, we identified 44 as being strictly about collaborative governance and 29 as being loosely about collaborative governance. The next step in our analysis was to code these 73 articles to answer the three research questions we included in the introduction of this article.⁵ First, we determined whether the network boundaries and the nodes of the network were clearly defined, and whether the authors discussed missing data and tried to remediate the problems associated with it (e.g., by performing missing data imputation, running simulations to check for robustness of results, etc.). With respect to the definition of network boundaries, we gauged whether the article clearly explained and justified the criteria for membership in the network studied. With respect to the definition of nodes, we assessed whether the article clearly defined the type of entity that counted as a node in the network. Articles that correctly identified nodes contained statements such as “the nodes in the network are organizations/individuals who are active in the topic of . . .” or “members of these network are the individuals who are interested in . . .”

Next, to address our second research question, we evaluated whether the articles (a) clearly defined the ties in the network, (b) accounted for tie strength if appropriate, and (c) analyzed multiplex relationships. With respect to tie definition, we coded whether the article defined relationships among actors (e.g., the transference of information, money, expertise, etc.). With respect to tie strength, we coded articles to gauge whether the ties analyzed were dichotomous (i.e., present/absent) or valued (i.e., of varying intensity) in nature. Among those articles that examined networks with valued ties, we then observed whether the authors took the strength of tie into account for the analysis or elected to dichotomize ties to simplify the analyses. Regarding multiplexity, we coded whether the article explicitly examined more than one type of relationship or not. Finally, to assess whether an article examined the evolution of the network, we simply coded whether the authors observed and analyzed the network at more than one point in time.

Results

Defining Network Boundaries and Nodes, and Dealing With Missing Data

Figure 5 presents results describing whether articles clearly define the boundaries of the networks they study, and the nodes within the networks.

Figure 5.

Distribution of articles based on whether they define network boundaries and nodes or not (disaggregated by type of article).

First, we notice that most studies (>70%), indeed, define the boundaries of the network by clearly describing the criteria for membership. Examples include Bodin and Nohrstedt (2016), who study the network of individuals engaged in the disaster response organization during the acute phase of a Swedish wildfire; Fischer et al.’s (2012) analysis of networks of actors in Swiss telecommunication policy; and Faul (2016), who identifies the actors that participate in a network of global education policy. Whether values exceeding 70% are high or low is, to a large extent, a matter of opinion. The optimistic reader would claim that these are high scores, because more than seven out of 10 articles on collaborative governance that deal with networks are successfully dealing with the challenge of properly defining the boundaries of the network. In contrast, the pessimist would regard this value as low, as properly defining the boundaries of the network is critical to lend internal and external validity to studies that engage in formal network analytic techniques (Wasserman & Faust, 1994). Improperly defining the boundaries of a network can lead to biased estimates of network-generating processes (e.g., partner selection) and can have significant implications for the interpretation of research findings. That almost 30% of the relevant literature exhibits this problem indicates the need for greater recognition of the importance for researchers to think carefully about how networks are delimited at the outset of their studies.

Regarding the definition of nodes, a very large majority of the articles carefully define the nodes that are part of the network under analysis. Most studies defined nodes as organizations (Kapucu & Garayev, 2013), although others defined them as individuals operating as members of organizations (e.g., Berardo, 2013; Song et al., 2019), or individuals without organizational affiliation (e.g., Matti & Sandström, 2011; Person et al., 2017). Of the subset of 44 articles that are strictly about collaborative governance, only two did not provide enough information for us to understand what exactly qualified as a node in the examined networks.

Figure 6 contains our findings about how articles deal with missing data.

Figure 6.

Distribution of articles based on whether they identify missing data problems and implement fixes (disaggregated by type of article).

Within the broader group of articles we reviewed, 15 did not have missing data problems, or they were negligible to the point of not posing significant threats to the internal or external validity of the studies. Thus, for the two variables reported in Figure 6, we analyzed the 58 articles that have a larger, and potentially more threatening, percentage of missing data. We found that in about 59% of the articles, the authors either mentioned that there is missing data in their studies or explicitly acknowledged that missing data could pose problems in their research design and potentially affect their findings. Interestingly, the difference between articles that acknowledge these issues and those that do not is more pronounced among authors of articles that are strictly about collaborative governance. Thus, it should come as no surprise that the percentage of studies that implement fixes to missing data problems is also greater among studies that are strictly about collaborative governance. Whereas about 40% of articles that are strictly about collaborative governance implement measures to address missing data problems (e.g., by limiting analysis to the network of actors that responded to a survey; Bodin et al., 2017), only about 16% of articles that are loosely about collaborative governance do so.

Defining Ties and Accounting for Tie Strength and Multiplexity

Figure 7 presents our findings on whether researchers properly define the ties that connect the nodes in the networks, account for tie strength, and study multiplex relationships that are often present in collaborative networks.

Figure 7.

Distribution of articles based on whether they define ties, account for tie strength, and analyze multiplex relationships (disaggregated by type of article).

As was the case with the definition of nodes, panel A in the Figure 7 shows that a large percentage of articles (∼93%) precisely described ties (i.e., they provided enough information about what “flows” through the ties in the network). Five articles did not include enough information, and all of them were strictly about collaborative governance. Ties included information exchange and provision (Horning et al., 2016; Provan et al., 2011), dyadic collaboration (Bergsten et al., 2019; Fischer et al., 2012), and project support (Gallemore & Jespersen, 2016), to name a few.

While almost all studies properly define ties, a minority of the articles we reviewed (approximately 30%) accounted for tie strength, as seen in panel B of Figure 7. Most studies, in fact, performed analyses that depended on dichotomizing ties, turning naturally valued ties into binary ones. For example, inter-firm collaboration (Della Corte & Aria, 2014), collaboration between municipalities (Kininmonth et al., 2015), or collaboration between organizations (Fischer et al., 2012), which, per definition, can be of various intensity, were measured and analyzed as dichotomous ties. This approach was all the more prevalent in studies that are strictly about collaborative governance, a finding we will discuss in more detail later. Dichotomizing ties in social network analysis (e.g., applying a threshold to a continuous variable to transform it into values of 0 and 1) is a common strategy, in part, because more readily accessible analytical tools have been developed to analyze nonvalued, or binary, matrices.

Finally, panel C in Figure 7 shows that almost half of the articles that measured networks analyzed multiplex relationships that took place among the members of the network. Collecting data on multiplex relationships is generally more difficult to do, especially in studies that use surveys or interviews, which can tax the attention of respondents when they are asked to provide detailed information about different types of social interactions. Yet, the careful examination of multiplex relationships can be critical in assessing the potential for collaborative governance; after all, collaborative relationships can be cemented on informal interactions to exchange valuable resources (Lee et al., 2012; Weare et al., 2014) but also formal exchanges that take place in multiple forums where conflict can be managed (Berardo & Lubell, 2016; Fischer & Leifeld, 2015). Among the studies included in this review, researchers examined multiplex relationships in a variety of ways. For instance, Faul (2016) observes both formal and informal partnership ties, whereas Kapucu and Garayev (2013) examine different types of advice exchanged by actors in emergency situations (Kapucu & Garayev, 2013). Lee et al. (2012) and Weible (2011) pay attention to both relationships of competition and cooperation among stakeholders, whereas Calliari et al. (2019) describe the complex interrelationship among actors who exchange information, coordinate their behavior, and collaborate with each other.

Evolution of Networks

Figure 8 shows that there is a small number of studies that examine the evolution of collaborative governance processes by analyzing dynamic changes in networks (∼15% of the 73 articles that measured networks).

Figure 8.

Distribution of articles based on whether they analyzed networks longitudinally or not (disaggregated by type of article).

The study of collaborative governance processes is enriched when researchers can understand how collaboration evolves over time. For example, Kanwar et al. (2016) analyzed how the governance network around the Kaipara Harbour in New Zealand became larger and more integrated because of ecological challenges, cultural integration efforts, and the change of legislation. Fischer et al. (2012) assessed the evolution of collaboration ties in Swiss telecommunication policy across two phases of decision-making. Berardo (2013) studied perceptions of procedural fairness and the establishment of relationships among organizations in policy networks in five U.S. estuaries. Finally, Hughes and Pincetl (2014) studied networks of actor co-participation in venues before and after the introduction of Integrated Water Resource Management plans in California.

A Multiple Correspondence Analysis of Articles Measuring Networks

To explore whether studies tend to cluster based on the types of network challenges they tackle, we perform a multiple correspondence analysis (MCA), which is a version of factor analysis designed for categorical variables (Greenacre & Blasius, 2006; Le Roux & Rouanet, 2010). MCA allows us to detect patterns of differences and similarities between articles in a statistically inductive manner, identifying variables that best describe the dimensions defining the similarities between the articles (see Lieberherr et al., 2019; Trein et al., 2019 for recent applications of this method to literature review articles). MCA relies on a symmetric matrix of all two-way cross-tabulations between the categorical variables included in the analysis. From this matrix, MCA then identifies variable categories that cluster together into a dimension (Trein et al., 2019). We use the FactoMiner (Lê et al., 2008) and factoextra (Kassambara & Mundt, 2017) packages in R to perform the analysis.⁶

The first two dimensions of the MCA cover almost 40% of the overall variance (24.5% and 14.7%, respectively⁷). Figure 9 shows the seven variables included in the MCA and how they align with the two dimensions (see figures in the supplemental appendix for the contributions of the single variables categories to each dimension). We did not include the variable capturing whether articles properly defined nodes because 71 out of the 73 cases did a good job in defining nodes. We also report results for all articles, not disaggregated by whether they are strictly or loosely about collaborative governance.⁸

Figure 9.

Contribution of individual variables to the main two dimensions of the MCA performed in articles that measure networks.

The first dimension (horizontal axis) identified by the MCA is strongly related to whether the article correctly defines network boundaries, identifies and addresses missing data problems, and analyzes multiplex relationships. The first three of those variables (definition of boundaries and identification and fixing of missing data problems) hint that this dimension is defined by what we could call a concern with methodological “quality” about collaborative governance and networks. Furthermore, on the top right corner of the figure, the variable “ties defined” also appears to contribute to this first dimension (although this variable also contributes strongly to the second dimension). Properly defining ties, properly defining and discussing network boundaries, and discussing and addressing missing data issues are issues that every collaborative governance paper on networks should deal with. The second dimension identified by the MCA is structured by variables that seem to indicate a “trade-off” between different aspects of networks. The two variables most strongly associated with this dimension are those that define whether the article includes a longitudinal analysis, and whether tie strength was accounted for, and how. Neither are issues directly related to the quality of the article, in the sense that a network might be rightly defined by simple dichotomous ties (no tie strength) and a static analysis (no network dynamics). This dimension indicates a trade-off in the sense that longitudinal analyses might require in many cases the dichotomization of ties, or that an analysis of longitudinal networks might complicate the correct analysis of weighted ties; variable categories “longitudinal analysis_1” (yes) and “tie strength analysis_3” (dichotomization of valued ties) do, indeed, lie at opposing ends of this dimension (see supplemental appendix, Figure A2).

Discussion

Most of the studies we reviewed that treated networks in a nonmetaphorical way properly defined network boundaries and clearly identified nodes and ties that connect them. We see this as a very positive result considering that, for example, not defining the boundaries of networks may have the most negative impact on the internal and external validity of results. Yet, our results show that despite the fact that a majority of studies do a good job at defining boundaries, nodes, and ties, there is still room for improvement.

Perhaps the simplest, most cost-effective way for researchers to achieve more precision in these tasks is to seek support from nonacademic sources that can provide complementary information to the sources the researcher has already identified. For instance, it may be the case that a researcher using a “realist” approach to network boundary identification (Laumann et al. (1989) may fail to identify the proper contours of the network if the network is fragmented, with multiple components that are disconnected from each other. In such cases, for example, a snowball technique used in a survey to identify actors who are members of the network but were not identified at the outset of the study may result in missing actors who are, indeed, part of the network, but are not located in the components populated by the initial respondents to the survey. In these types of situations, a researcher may benefit immensely from assembling a study advisory committee formed by a handful of individuals who are highly knowledgeable about the study area, the problem of interest for the researcher, and the group of actors and/or stakeholders that may be active in addressing the problem. Such a group of knowledgeable people can help the researcher be more exhaustive in identifying nodes that may have otherwise been overlooked. Regardless of what approach a researcher uses to better approximate the boundaries of the network, we believe that resources spent at the beginning of the research process to address this issue is likely to be cost-effective simply because the failure to account for all relevant network members affects all subsequent stages of the research process and may ultimately diminish the validity of results.

Our results also show other areas in which research on networks and collaborative governance can improve considerably. For instance, missing data is a common problem that is not often faced head on. Sure enough, acknowledging missing data problems is by itself valuable, but researchers should, when possible, take a more militant approach to reduce the effect of missing data on the validity of findings. This is, by no means, easy to do. Given that most studies on collaborative governance rely on survey data, the underlying problem that a certain (and probably not random) percentage of survey participants simply decide not to participate is hard to fix. There are, however, different types of strategies to deal with this issue, some of which were, in fact, used in studies included in our review. These include the reconsideration of network boundaries (Bodin et al., 2017; Kronenberg et al., 2016), the transformation of directed network data to undirected network data (Balsiger & Ingold, 2016), or the application of extrapolation methods to infer patterns of missing data (Claro et al., 2003).

In addition to these approaches, others have recently explored the interchangeability of nonintrusive methods to reduce missing data problems. Specifically, Yi and Scholz (2016) collect data on network interactions based on media reports, hyperlinks available on individual policy actors’ websites, and on public policy partnership documentation, and find that all three methods produce similar results in terms of descriptive network-level statistics and dominant local structures in exponential random graph models. The two nonintrusive methods may constitute viable approaches for data collection for researchers with limited resources that can be quickly depleted in survey-based research.

It is also important to keep in mind that, in the last few years, there have been rapid methodological developments in the areas of text analysis that hold promise to circumvent the problems of missing data that are pervasive in survey-based research. For instance, scholars have leveraged the benefits of mining textual sources to gather information about level of agreement and interactions among stakeholders in collaborative partnerships (Berardo et al., 2014; Scott et al., 2020) as well as the effect that collective-action problems have on the design of formal institutional arrangements that force actors into formal relationships of collaboration with each other (Olivier, 2019). Collaborative governance scholarship will benefit from aggressively pursuing these new approaches to data collection, particularly given the trove of invaluable data that is available online to examine collaborative processes.

Our results also show that a large percentage of studies using networks in a nonmetaphorical way simplify their modeling strategies by dichotomizing ties that are valued in nature. There are potential problems with this type of choice, as dichotomizing ties might distort the latent structure of the network (Aicher et al., 2014; Thomas & Blitzstein, 2011), which, in turn, will likely lead to biased results. Historically, these choices made sense as methodological tools to describe and model behavior in valued networks were not readily available for a majority of scholars. Fortunately, new methodological advances in the last few years allow researchers to examine valued ties without resorting to the practice of dichotomization. These advances include the design of statistical models for valued networks (Krivitsky, 2012) that can be used to explore collaborative behavior in organizational networks (Scott, 2016) and new approaches to network partitioning for valued data that overcome some of the limitation of older techniques (Nordlund, 2016). Of course, dichotomization can be a defensible research strategy under certain conditions, as when different positive values indicating tie strength are heavily skewed to the right. But at the very least, the dichotomization of ties should be accompanied by a sound theoretical justification for the choice of dichotomization thresholds and (when possible) robustness tests with other potentially meaningful thresholds.

Regarding the study of multiplex ties, which can be a distinctive feature of collaborative governance studies, we found that approximately 50% of the articles, indeed, explored the multiplex nature of connections in networks. We see this as a positive development, as it demonstrates that collaborative governance studies are increasingly adding empirical evidence to the long-held claim that collaborative network relationships are multidimensional in nature (Allen et al., 2007; Håakansson & Johanson, 1988; Shrestha et al., 2014). As was the case for the missing data problem, new methodologies have been developed that can facilitate the measurement of multiplex networks. For example, Hayes and Scott (2018) study a policy network composed by 221 organizational nodes, and collect information on ties using multiple approaches (surveys, Twitter interactions, and hyperlinks). They find that, while correlated, the objective measures (online-based behavior) produce a different pattern of interactions than the subjective, memory-based survey approach. Their approach can be easily (and, as importantly, cheaply) imitated to produce more nuanced explanations of how different types of ties may affect the collaborative practices of actors.

While multiplex networks can offer a richer depiction of the social processes that influence collaborative governance, it is important to acknowledge that scholars may have good reasons to limit analysis to single forms of relationships among actors. For example, researchers with extensive familiarity of governance systems may conclude that a particular form of interaction has disproportionate influence in shaping policy processes and outcomes, and, thus, deserves more attention. Likewise, a study’s research questions may require a narrow focus on a certain form of social interaction. However, even in these cases, it is important that authors explain why a uniplex approach is appropriate, given growing recognition of how multiplex networks can enrich understanding of collaborative governance (Hayes & Scott, 2018; Robins et al., 2011; Shrestha & Feiock, 2009).

Finally, we found a minority of studies in collaborative governance that assess how networks evolved over time. Contrary to cross-sectional studies, longitudinal network studies can analyze not only the evolution of network structures, but also their co-evolution with collaborative practices (Berardo & Scholz, 2010). If it is true that system context matters in the study of collaborative processes, and that multiple variables help explain how a collaborative governance regime unfolds (Emerson et al., 2012), then a more significant portion of collaborative governance scholarship will need to devote considerable resources in the future to measuring and analyzing the evolution of networks.

Clearly, there are justifiable reasons why researchers rarely study networks over time. Chief among them is the time- and resource-intensive task of gathering multiple waves of data. Fortunately, there are ways of addressing this challenge. One simple approach that survey-based studies can adopt, for instance, is to ask actors about their current as well as their past relations (Ingold & Fischer, 2014). Although convenient, this approach suffers from the obvious shortcoming of depending on the memory of respondents to identify past ties, which may be affected by their current set of interactions (i.e., a collaborative tie in the present may bias respondents to name a collaborative tie in the past when none existed). But some of the approaches we highlighted above when discussing missing data and the multiplexity of networks can be tailored to the longitudinal study of networks. For example, the media reports used by Yi and Scholz (2016), and the online exchanges examined by Hayes and Scott (2018), could be mined to easily reconstruct a different snapshot of collaborative networks. In any event, the advent of new techniques to uncover networks from textual sources may aid researchers in further studying collaborative governance as the dynamic process that it is.

Conclusion

The fundamental aim of this article is to encourage researchers to think carefully about how to improve research design in studies that invoke the importance of networks for a better understanding of collaborative governance. Actual research on collaborative governance often—implicitly or explicitly—deals with networks of actors that operate dynamically across sectoral, hierarchical, and geographic boundaries.

This article evaluates how collaborative governance scholars who examine networks address some of the basic conceptual and methodological challenges that arise in network-based research. While our review focuses on studies that formally analyze networks, it is important to note that the majority of the literature, instead, treats networks in a metaphorical way. The metaphorical use of the network concept has spurred the development of new theoretical ideas in this vein of literature and will continue to do so. In tandem with our review, as well as reviews of qualitative research on networks and collaborative governance (e.g., Ashworth et al., 2019), the evaluation of collaborative governance scholarship that conceptualizes networks metaphorically could contribute significantly to the broader literature.

Our review focuses on research that applies formal network analytical techniques, which present unique challenges as well as advantages. For instance, formal network analysis can help create a body of literature that facilitates comparison across different collaborative governance processes, in different countries, and over time. Furthermore, this approach facilitates the application of theoretical ideas from other domains of network studies to the analysis of collaborative governance systems. Of the studies that explicitly observe and measure networks, we found that most of them properly defined network boundaries, and clearly identified nodes and the ties that connect them. However, we also found considerable potential for scholarship on networks to better address missing data, properly account for tie strength in relationships, evaluate multiplexity of network ties, and study the evolution of networks over time. These findings somewhat parallel results from Lecy et al.’s (2014) review of network research in public administration scholarship that identifies a need for greater conceptual clarity and broader usage of methods capable of capturing specific network activities.

Researchers obviously need to simplify some aspects of their research designs, particularly when collecting data on complex social interactions becomes an expensive, time-intensive task. Within a given research project, it will, thus, always be a major challenge to address the different considerations we have discussed. Yet, researchers must think carefully about the implications of their methodological choices, especially those that relate to the bounding of networks and the measurement of interactions. We hope that our propositions on how to address and overcome the different challenges will spur innovative ideas on how to perform research on networks in collaborative governance.

We would like to close by noting optimistically that a nontrivial number of the articles we reviewed were extremely nuanced from both a theoretical and methodological standpoint. More than a good share of this work was clear in describing simplifying assumptions and the inner work of some of the sophisticated social network analysis techniques used in the analyses. If future research continues on the path that some of this recent scholarship has embarked on, the study of collaborative governance through a network perspective will continue to strengthen, attracting a new crop of social scientists striving to answer theoretically interesting questions through the use of sound methodological practices.

Supplemental Material

sj-pdf-1-arp-10.1177_0275074020927792 - Supplemental material for Collaborative Governance and the Challenges of Network-Based Research

Supplemental material, sj-pdf-1-arp-10.1177_0275074020927792 for Collaborative Governance and the Challenges of Network-Based Research by Ramiro Berardo, Manuel Fischer and Matthew Hamilton in The American Review of Public Administration

Footnotes

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) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Ramiro Berardo

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

Ramiro Berardo is an associate professor of Environmental and Natural Resources Policy. Most of his work uses a social network analysis approach to examine the structure and function of collaborative processes to tackle environmental problems.

Manuel Fischer is a researcher in Policy Analysis and Environmental Governance. His work analyzes governance arrangements, decision-making processes and political networks, with a focus on water and environmental issues.

Matthew Hamilton is an assistant professor in the School of Environment and Natural Resources at The Ohio State University. He conducts research on how social interaction affects environmental policy processes, especially in hazard-prone governance systems.

References

Aicher

Jacobs

A. Z.

Clauset

(2014). Learning latent block structure in weighted networks. Journal of Complex Networks, 3(2), 221–248.

Allen

James

A. D.

Gamlen

(2007). Formal versus informal knowledge networks in R&D: A case study using social network analysis. R&D Management, 37(3), 179–196.

Amsler

L. B.

(2016). Collaborative governance: Integrating management, politics, and law. Public Administration Review, 76(5), 700–711. https://doi.org/10.1111/puar.12605

Ansell

Gash

(2008). Collaborative governance in theory and practice. Journal of Public Administration Research and Theory, 18(4), 543–571.

Ashworth

R. E.

McDermott

A. M.

Currie

(2019). Theorizing from qualitative research in public administration: Plurality through a combination of rigor and richness. Journal of Public Administration Research and Theory, 29(2), 318–333. https://doi.org/10.1093/jopart/muy057

Balsiger

Ingold

(2016). In the eye of the beholder: Network location and sustainability perception in flood prevention. Environmental Policy and Governance, 26(4), 242–256. https://doi.org/10.1002/eet.1715

Bardach

(1998). Getting agencies to work together: The practice and theory of managerial craftsmanship. Brookings Institution Press.

Berardo, R. (2010). Sustaining joint ventures: The role of resource exchange and the strength of inter-organizational relationships. In R.C. Feiock & J.T. Scholz (Eds.) Self-organizing federalism: Collaborative mechanisms to mitigate institutional collective action (pp. 204–228). Cambridge University Press.

Berardo, R. (2013). The coevolution of perceptions of procedural fairness and link formation in self-organizing policy networks. The Journal of Politics, 75(3), 686–700.

10.

Berardo, R., Heikkila, T., & Gerlak, A. K. (2014). Interorganizational engagement in collaborative environmental management: Evidence from the South Florida Ecosystem Restoration Task Force. Journal of Public Administration Research and Theory, 24(3), 697–719.

11.

Berardo, R., & Lubell, M. (2016). Understanding what shapes a polycentric governance system. Public Administration Review, 76(5), 738–751.

12.

Berardo, R., & Scholz, J. T. (2010). Self-organizing policy networks: Risk, partner selection, and cooperation in estuaries. American Journal of Political Science, 54(3), 632–649.

13.

Bergenholtz

Waldstrøm

(2011). Inter-organizational network studies—A literature review. Industry and Innovation, 18(6), 539–562.

14.

Bergsten

Jiren

T. S.

Leventon

Dorresteijn

Schultner

Fischer

(2019). Identifying governance gaps among interlinked sustainability challenges. Environmental Science & Policy, 91, 27–38.

15.

Bodin

Ö.

Nohrstedt

(2016). Formation and performance of collaborative disaster management networks: Evidence from a Swedish wildfire response. Global Environmental Change, 41, 183–194.

16.

Bodin

Ö.

Sandström

Crona

(2017). Collaborative networks for effective ecosystem-based management: A set of working hypotheses. Policy Studies Journal, 45(2), 289–314.

17.

Burt

R. S.

(1983). Studying status/role-sets using mass surveys. In Burt

R. S.

Minor

(Eds.), Applied network analysis (pp. 100–118). SAGE.

18.

Burt

R. S.

(1987). A note on missing network data in the general social survey. Social Networks, 9(1), 63–73.

19.

Calliari

Michetti

Farnia

Ramieri

(2019). A network approach for moving from planning to implementation in climate change adaptation: Evidence from southern Mexico. Environmental Science & Policy, 93, 146–157.

20.

Claro

D. P.

Hagelaar

G. J.

Kemp

R. G.

Omta

S. W. F.

(2003). The impact of downstream network subgroups on collaboration and performance: A survey of buyer-supplier relationships in the Dutch flower sector. International Food and Agribusiness Management Review, 6(2), 87–98.

21.

Costenbader

Valente

T. W.

(2003). The stability of centrality measures when networks are sampled. Social Networks, 25(4), 283–307.

22.

Della Corte

Aria

(2014). Why strategic networks often fail: Some empirical evidence from the area of Naples. Tourism Management, 45, 3–15.

23.

Dowding, K. (1995). Model or metaphor? A critical review of the policy network approach. Political Studies, 43(1), 136–158.

24.

Emerson, K., & Nabatchi, T. (2015). Collaborative governance regimes. Georgetown University Press.

25.

Emerson

Nabatchi

Balogh

(2012). An integrative framework for collaborative governance. Journal of Public Administration Research and Theory, 22(1), 1–29. https://doi.org/10.1093/jopart/mur011

26.

Faul

M. V.

(2016). Networks and power: Why networks are hierarchical not flat and what can be done about it. Global Policy, 7(2), 185–197.

27.

Fischer

(2015). Collaboration patterns, external shocks and uncertainty: Swiss nuclear energy politics before and after Fukushima. Energy Policy, 86, 520–528.

28.

Fischer

Angst

Maag

(2017). Co-participation in the Swiss water forum network. International Journal of Water Resources Development, 35(3), 446–464.

29.

Fischer

Ingold

Sciarini

Varone

(2012). Impacts of market liberalization on regulatory network: A longitudinal analysis of the Swiss telecommunications sector. Policy Studies Journal, 40(3), 435–457. https://doi.org/10.1111/j.1541-0072.2012.00460.x

30.

Fischer

Leifeld

(2015). Policy forums: Why do they exist and what are they used for? Policy Sciences, 48, 363–382. https://doi.org/10.1007/s11077-015-9224-y

31.

Fischer

Sciarini

(2016). Collaborative tie formation in policy networks: A cross-sector perspective. The Journal of Politics, 78, 63–74.

32.

Gallemore

Jespersen

(2016). Transnational markets for sustainable development governance: The case of REDD+. World Development, 86, 79–94.

33.

Greenacre

Blasius

(2006). Multiple correspondence analysis and related methods. Chapman and Hall/CRC Press.

34.

Håakansson

Johanson

(1988). Formal and informal cooperation strategies in international industrial networks. In Contractor

F. J.

Lorange

(Eds.), Cooperative strategies in international business (pp. 369–379). Lexington Books.

35.

Hayes

A. L.

Scott

T. A.

(2018). Multiplex network analysis for complex governance systems using surveys and online behavior. Policy Studies Journal, 46(2), 327–353. https://doi.org/10.1111/psj.12210

36.

Horning

Bauer

B. O.

Cohen

S. J.

(2016). Missing bridges: Social network (dis)connectivity in water governance. Utilities Policy, 43, 59–70.

37.

Hughes

Pincetl

(2014). Evaluating collaborative institutions in context: The case of regional water management in southern California. Environment and Planning C: Government and Policy, 32(1), 20–38. https://doi.org/10.1068/c1210

38.

Huxham

Vangen

Huxham

Eden

(2000). The challenge of collaborative governance. Public Management: An International Journal of Research and Theory, 2(3), 337–358. https://doi.org/10.1080/14719030000000021

39.

Ingold

Fischer

(2014). Drivers of collaboration to mitigate climate change: An illustration of Swiss climate policy over 15 years. Global Environmental Change, 24, 88–98.

40.

Ingold

Fischer

Cairney

(2017). Drivers for policy agreement in nascent subsystems: An application of the advocacy coalition framework to fracking policy in Switzerland and the UK. Policy Studies Journal, 45(3), 442–463.

41.

Isett

K. R.

Mergel

I. A.

LeRoux

Mischen

P. A.

Rethemeyer

R. K.

(2011). Networks in public administration scholarship: Understanding where we are and where we need to go. Journal of Public Administration Research and Theory, 21(Suppl. 1), i157–i173.

42.

Johnston

E. W.

Hicks

Nan

Auer

J. C.

(2011). Managing the inclusion process in collaborative governance. Journal of Public Administration Research and Theory, 21(4), 699–721. https://doi.org/10.1093/jopart/muq045

43.

Josse, J., & Husson, F. (2016). missMDA: A package for handling missing values in multivariate data analysis. Journal of Statistical Software, 70(1), 1–31.

44.

Kanwar

Koliba

Greenhalgh

Bowden

W. B.

(2016). An institutional analysis of the Kaipara Harbour governance network in New Zealand. Society & Natural Resources, 29(11), 1359–1374.

45.

Kapucu

Garayev

(2013). Designing, managing, and sustaining functionally collaborative emergency management networks. The American Review of Public Administration, 43(3), 312–330.

46.

Kassambara

Mundt

(2017). factoextra: Extract and visualize the results of multivariate data analyses. https://cran.r-project.org/web/packages/factoextra/index.html

47.

Kininmonth

Bergsten

Bodin

Ö.

(2015). Closing the collaborative gap: Aligning social and ecological connectivity for better management of interconnected wetlands. Ambio, 44(1), 138–148. https://doi.org/10.1007/s13280-014-0605-9

48.

Kossinets

(2006). Effects of missing data in social networks. Social Networks, 28(3), 247–268. https://doi.org/10.1016/j.socnet.2005.07.002

49.

Krivitsky

P. N.

(2012). Exponential-family random graph models for valued networks. Electronic Journal of Statistics, 6, 1100–1128.

50.

Kronenberg

Pietrzyk-Kaszyńska

Zbieg

Żak

(2016). Wasting collaboration potential: A study in urban green space governance in a post-transition country. Environmental Science & Policy, 62, 69–78.

51.

Laumann

E. O.

Marsden

P. V.

Prensky

(1989). The boundary specification problem in network analysis. In Freeman

L. C.

White

D. R.

Romney

A. K.

(Eds.), Research methods in social network analysis (pp. 61–87). George Mason University Press.

52.

Lê

Josse

Husson

(2008). FactoMineR: An R package for multivariate analysis. Journal of Statistical Software, 25(1), 1–18.

53.

Lecy

J. D.

Mergel

I. A.

Schmitz

H. P.

(2014). Networks in public administration: Current scholarship in review. Public Management Review, 16(5), 643–665.

54.

Lee

I. W.

Feiock

R. C.

(2012). Interorganizational collaboration networks in economic development policy: An exponential random graph model analysis. Policy Studies Journal, 40(3), 547–573.

55.

Leifeld

Schneider

(2012). Information exchange in policy networks. American Journal of Political Science, 56(3), 731–744.

56.

Le Roux

Rouanet

(2010). Multiple correspondence analysis. SAGE.

57.

Lieberherr

Fischer

Tschannen

(2019). Taking stock of institutional resource regime research: A meta-analysis. Environmental Science & Policy, 97, 81–89.

58.

Margerum

R. D.

(2008). A typology of collaboration efforts in environmental management. Environmental Management, 41(4), 487–500.

59.

Marsh

(1998). Comparing policy networks. Open University Press.

60.

Matti

Sandström

(2011). The rationale determining advocacy coalitions: Examining coordination networks and corresponding beliefs. Policy Studies Journal, 39(3), 385–410. https://doi.org/10.1111/j.1541-0072.2011.00414.x

61.

Nordlund

(2016). A deviational approach to blockmodeling of valued networks. Social Networks, 44, 160–178.

62.

Olivier

(2019). How do institutions address collective-action problems? Bridging and bonding in institutional design. Political Research Quarterly, 72(1), 162–176.

63.

Person

M. T.

Delea

M. G.

Garn

J. V.

Alexander

Abaire

Freeman

M. C.

(2017). Assessing the influence of social capital on water point sustainability in rural Ethiopia. Journal of Water, Sanitation and Hygiene for Development, 7(4), 611–622.

64.

Provan

K. G.

Beagles

J. E.

Leischow

S. J.

(2011). Network formation, governance, and evolution in public health: The North American Quitline Consortium case. Health Care Management Review, 36(4), 315–326. https://doi.org/10.1097/HMR.0b013e31820e1124

65.

Robins

Bates

Pattison

(2011). Network governance and environmental management: Conflict and cooperation. Public Administration, 89(4), 1293–1313. https://doi.org/10.1111/j.1467-9299.2010.01884.x

66.

Scholz

J. T.

Berardo

Kile

(2008). Do networks solve collective action problems? Credibility, search, and collaboration. Journal of Politics, 70(2), 393–406.

67.

Scott

T. A.

(2016). Analyzing policy networks using valued exponential random graph models: Do government-sponsored collaborative groups enhance organizational networks? Policy Studies Journal, 44(2), 215–244. https://doi.org/10.1111/psj.12118

68.

Scott

T. A.

Thomas

C. W.

(2017). Winners and losers in the ecology of games: Network position, connectivity, and the benefits of collaborative governance regimes. Journal of Public Administration Research and Theory, 27(4), 647–660.

69.

Scott

T. A.

Ulibarri

Scott

R. P.

(2020). Stakeholder involvement in collaborative regulatory processes: Using automated coding to track attendance and actions. Regulation & Governance, 14, 219–237.

70.

Shrestha

M. K.

Berardo

Feiock

(2014). Solving institutional collective action problems in multiplex networks. Complexity, Governance & Networks, 1(1), 49–60.

71.

Shrestha

M. K.

Feiock

R. C.

(2009). Governing US metropolitan areas: Self-organizing and multiplex service networks. American Politics Research, 37(5), 801–823.

72.

Smith

J. A.

Moody

(2013). Structural effects of network sampling coverage I: Nodes missing at random. Social Networks, 35(4), 652–668.

73.

Snijders

T. A.

Van de Bunt

G. G.

Steglich

C. E.

(2010). Introduction to stochastic actor-based models for network dynamics. Social Networks, 32(1), 44–60.

74.

Song

A. M.

Temby

Kim

Cisneros

A. S.

Hickey

G. M.

(2019). Measuring, mapping and quantifying the effects of trust and informal communication on transboundary collaboration in the Great Lakes fisheries policy network. Global Environmental Change, 54, 6–18.

75.

Stokman

F. N.

Zeggelink

E. P.

(1996). Is politics power or policy oriented? A comparative analysis of dynamic access models in policy networks. Journal of Mathematical Sociology, 21(1–2), 77–111.

76.

Thomas

A. C.

Blitzstein

J. K.

(2011). Valued ties tell fewer lies: Why not to dichotomize network edges with thresholds. ArXiv Preprint ArXiv:1101.0788. https://arxiv.org/abs/1101.0788

77.

Trein

Meyer

Maggetti

(2019). The integration and coordination of public policies: A systematic comparative review. Journal of Comparative Policy Analysis: Research and Practice, 21(4), 332–349.

78.

Ulibarri

Scott

T. A.

(2017). Linking network structure to collaborative governance. Journal of Public Administration Research and Theory, 27, 163–181. https://doi.org/10.1093/jopart/muw041

79.

Vangen

Hayes

J. P.

Cornforth

(2015). Governing cross-sector, inter-organizational collaborations. Public Management Review, 17(9), 1237–1260.

80.

Wasserman

Faust

(1994). Social network analysis: Methods and applications (Vol. 8). Cambridge University Press.

81.

Weare

Lichterman

Esparza

(2014). Collaboration and culture: Organizational culture and the dynamics of collaborative policy networks. Policy Studies Journal, 42(4), 590–619.

82.

Weible

C. M.

(2011). Political-administrative relations in collaborative environmental management. International Journal of Public Administration, 34(7), 424–435. https://doi.org/10.1080/01900692.2011.569916

83.

Scholz

J. T.

(2016). Policy networks in complex governance subsystems: Observing and comparing hyperlink, media, and partnership networks. Policy Studies Journal, 44(3), 248–279.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB