Examining the Role of Trust and Informal Communication on Mutual Learning in Government

Abstract

Although public agencies must mutually coordinate climate policy and other complex environmental issues, the extent and relative importance of informal networks and different dimensions of trust to the process remains underresearched. Addressing this, we conducted surveys and interviews with civil servants from numerous agencies and three levels of government working on climate change–related policy in the state of New York. We examined the effect of two network properties on mutual learning on climate change–related issues: the extent to which interagency communication takes places through formal and informal channels, and the distribution of two dimensions of trust (“fair play” and “relational comfort”) across the network. Our analysis revealed that formal communication among staff at different agencies was utilized more often than informal and that interagency relationships were more characterized by a feeling of “fair play” than by “relational comfort,” yet informal communication and Relational Comfort were the most important in facilitating interagency collaboration.

Keywords

environmental change environmental management social capital public administration collaborative networks innovation adaptation interagency knowledge

Introduction

Recent research on climate change–related policy (both adaptation and mitigation) acknowledges the highly complex, crosscutting, and transdisciplinary nature of the challenge, requiring a substantial amount of interagency coordination within and across governments in order to foster more resilient societies (Bryner & Duffy, 2012; Burch, 2010; Craik, Studer, & VanNijnatten, 2013). At the same time, it is well known that the bureaucratic structure of public agencies imposes serious institutional constraints on their capacity to cultivate a collaborative environment that enhances learning and coordination among agencies (see Allen & Gunderson, 2011; Barnett & Finnemore, 1999; Crozier 1964; Holling & Meffe, 1996). This, in turn, limits the capacity of agencies to mobilize and translate evidence-based knowledge across agency boundaries and, therefore, to collaboratively develop integrated assessments of, and policy responses to, dynamic environmental challenges (Lalor & Hickey, 2014). The public bureaucracy, seeking to avoid the pitfalls of organizational “silos,” regularly supports the development of nonhierarchical mechanisms to manage these issue-specific networks. Perhaps most visible for U.S. climate change policy in recent years has been President Obama’s appointment of successive “climate czars” who, despite lacking formal authority, have been rich in symbolic significance. Many other complementary and alternative organizational arrangements of varying degrees of formality are also used, including issue-specific interagency task forces, working groups, and coordinating bureaus.

In the field of public administration, scholarly responses to such developments have been part of the New Public Management approach, which calls for “post-bureaucratic” ways for government organizations to function (see Aucoin, 1995; Heckshcer, 1994), with a particular emphasis on the important role of public management networks (see Isett, Mergel, LeRoux, Mischen, & Rethemeyer, 2011, for the state of the art). While the public management network literature is quite diverse, among the most pressing (and insufficiently investigated) challenges remains the need to determine what network features enable policy coordination—features that Agranoff and McGuire (2001, p. 301) call “groupware” (see also, McGuire & Agranoff, 2011). Similarly, contemporary literature on climate change–related governance acknowledges an important role for networks in increasing government capacity and facilitating necessary mutual learning and information exchange (Healy, VanNijnatten, & López-Vallejo, 2014, chap. 5; Selin & VanDeveer, 2009; VanNijnatten & Craik, 2013).

Accompanying this research priority in public network scholarship is an acknowledgment, albeit underresearched, that the management of complex issues requires a considerable degree of trust among network participants and also a robust informal dimension to their relations (Edelenbos & Klijn, 2007). For example, previous research suggests that dimensions of what is often conceptualized as “social capital” (such as interorganizational trust and informal communication networks) can create flexibility in how organizations work together, particularly in complex decision-making environments where co-learning and innovation is required to respond to diverse and changing circumstances (Agranoff, 2006; Berardo, 2009; Coleman, 1988; Klenk, Hickey, & MacLellan, 2010; Liebeskind, Oliver, Zucker, & Brewer, 1996). Yet in their recent review of network scholarship in public administration, Isett et al. (2011, p. i165) observed that, for informal relations, “there has been very little advancement of our understanding of this pervasive mechanism of governance.” While trust is a heavily researched subject in other fields, knowledge of its significance in the collaborative process of managing environment and sustainability-related issues remains understudied. Stern and Coleman (2015, p. 118) point out that “trust theory remains underexplored in this context when compared to other fields,” and consider this a problem due to trust’s importance in these governance processes. Some important studies on trust have focused on the antecedents of interorganizational trust in a public administration context (Lambright, Mischen, & Laramee, 2010; cf. Delgado-Márquez, Hurtado-Torres, & Aragón-Correa, 2012; Ferrin, Dirks, & Shah, 2006). However, there is still a lack of evidence about the effects of these relational traits on the capacity of agencies to mutually learn and adjust, despite the fact that governments are already encouraging the formation of nonhierarchical arrangements for climate change–related policy formulation and implementation. Furthermore, there is now a recognized need for the development of “theories of sustainability management,” with concepts and observations specific to this transdisciplinary field of policy and management (Starik & Kanashiro, 2013, p. 7).

The objective of this article is therefore to better understand how interagency trust and communication patterns in the policy networks of civil servants working in climate change–related issues influences their mutual learning and adjustment. Given that little existing research has sought to operationalize and measure these network dimensions in the context of public environmental management and governance, our research was necessarily exploratory. Focusing on New York State (NYS), we conducted surveys and interviews with civil servants from across government to measure the influence and interactions of these variables. In what follows we briefly discuss the literature on collaborative public management networks and mutual learning in order to frame our analysis and highlight our contribution to public administration and environmental management theory. We then provide a brief overview of the interjurisdictional dynamics of climate change governance in New York in order to help clarify the complex organizational challenge faced by its public agencies. This is followed by a description of our research design and methods, including the assumptions and limitations affecting our study. We then present our results and discuss their broader implications for research, policy, and practice.

Collaborative Networks and Mutual Learning

Collaborative public management networks, which represent a conceptual subset of policy networks (Börzel, 1998), are the instantiation of “the process of facilitating and operating in multiorganization arrangements to solve problems that cannot be solved, or solved easily, by single organizations” (Agranoff & McGuire, 2003, p. 4). Such networks have been identified in diverse social issue areas such as child abuse in low-income neighborhoods (Mulroy & Shay, 1998), urban redevelopment (Agranoff & McGuire, 2003), watershed management (Imperial, 2005), land-use and transportation planning (Henry, Lubell, & McCoy, 2011), among others, and have been examined in a variety of ways in terms of their properties, antecedents, and outcomes.

Issue-specific collaborative networks are thought to facilitate learning by enabling civil servants implicated in the management of the issue, yet with diverse tasks and expertise, to share information and formulate strategies. John Agranoff (2006, pp. 59-60) uses the term, “mutual learning and adjustment,” and argues that the “most distinctive collaborative activity” of the networks he examined was their ability to manage knowledge as a basis for coordinated action.¹ Yet as the study of collaborative networks has lagged practice, so have studies of what factors in the collaborative process lead to successful mutual learning and adjustment. This has resulted in calls for researchers to provide evidence that contributes to a fuller understanding of the pathways enabling collaboration and policy innovation (see Agranoff & McGuire, 2001; McGuire & Agranoff, 2011; Thomson & Perry, 2006). Using empirical evidence collected from civil servants working in NYS, we examine two of these pathways: (1) the distribution of interagency trust within the network and (2) the extent to which interagency communication among network members are conducted formally and informally.

Trust and Collaborative Public Management Networks

The existence of trust and norms of reciprocity among network participants has been identified as enabling the success of collaborative networks by numerous authors (Edelenbos & Klijn, 2007; O’Toole, 1995; Provan & Kenis, 2008). Trust reduces the process costs of collaborating (Agranoff, 2007; Chiles & McMackin, 1996), enables the sharing of information and coordination of tasks (Cummings & Bromiley, 1996), is a condition for the acceptance of network leadership (Bardach, 2001), leads network participants to more favorably assess network outcomes (Klijn, Steijn, & Edelenbos, 2010), and affects the structure of the network (Henry et al., 2011). However, the development of empirical measures of interagency trust that can enable generalizable observations about its effects on mutual learning and adjustment within public management networks has lagged. And while trust has been heavily researched in other organizational contexts, the generalizability of the findings to public environment and sustainability-related governance is questionable.²

Noting this problem, Stern and Coleman (2015, p. 117) have sought to “reorganize trust theory in a robust and practical way for collaborative natural resource management.” In doing so, they distinguished between two forms of trust particularly relevant to interagency climate change–related policy development and implementation, namely, “affinitive” trust and “procedural” trust. The former, affinitive trust, is “based primarily on the emotions and associated judgements resulting from either cognitive or subconscious assessments of the qualities of the potential trustee”; the latter, procedural trust, is “trust in procedures or other systems that decrease vulnerability or the potential trustor, enabling action in the absence of other forms of trust” (Stern & Coleman, 2015, p. 122). Stern and Coleman thus distinguish between assessments about behavior as a basis for trust and assessments of the broader context decreasing vulnerability and potentially enabling mutual learning and adjustment even when behavioral assessments do not contribute to this. We will examine the relevance of this distinction and relative roles of the different forms of trust in our analysis.

Formality and Informality in Collaborative Management Networks

Another related empirical problem stems from the recognition that all collaborative networks are communicative networks since their defining traits are the interactions among network members. This communication can have varying levels of formality, from executive-level task force meetings to encounters at a local coffee shop. Formal networks are “multi-actor arrangements explicitly constituted by public managers to produce and deliver public services” (Isett et al., 2011, p. i162). Studies on formal networks have examined their effectiveness in terms of a variety of outcomes, as well as the organizational dynamics that lead to their formation (see Isett & Provan, 2005; Provan, Beagles, Mercken, & Leischow, 2013; Provan & Milward, 1995; Rethemeyer & Hatmaker, 2008; Shrestha, 2013). Some of these have shown that the development of a formal network is conducive to well-coordinated outcomes, especially when the networks are organized for the purposes of coordinating the delivery of services (see, Brass, Galaskiewicz, Greve, & Tsai, 2004). Informal ties are generally understood here as less important network dimensions in terms of the delivery of services, yet are also precursors for relationships that eventually become formalized as participants become more committed to working together (see Imperial, 2005; Provan & Milward, 2001). Few studies directly relevant to climate change–related public management have underscored the relationship between the informality of communicative networks and the existence of trust among network members. In one well-known study of cooperative relationships among business firms, Ring and Van de Ven (1994) proposed that trust enables informal understanding and commitments to replace more specific reciprocity. Yet the importance of informal communication within collaborative public management networks remains underresearched, as does its relationship to trust, and research is needed to untangle their effects on mutual learning and adaptation (Isett et al., 2011).

Research Questions

In this exploratory case study, we first ask: (1) To what extent do civil servants operating in different public agencies communicate with one another within and across jurisdictions and (2) How are measures of trust and communication distributed across the network? Given a distribution of trust (two dimensions) and informal/formal communication, we then examine the independent effect these exert on the degree of interagency mutual learning and adjustment reported by civil servants in the network.

More specifically, we ask: (3) Does informal communication have a direct effect on mutual learning and adjustment, or is its influence channeled through formal network dimensions? and, recognizing Stern and Coleman’s (2015) distinction between affinitive trust and procedural trust, (4) What are the relative contributions of different measures of trust?

Case Study: New York

NYS was selected because it represents an excellent case study of the challenges associated with the multilevel governance of complex and dynamic environmental challenges. It is one of the most populous and economically diversified states in the United States, with a population of 19.7 million and gross state product of more than $1.3 trillion (2014). NYS is also among the most urbanized U.S. states with its population living in urban rather than rural areas by a factor of greater than 10 to 1, while the majority of the land in the northern part of the state is protected from development. More recently, the impacts of Hurricanes and Tropical Storms (e.g., Irene in 2011 and Sandy in 2012) have encouraged policy makers to prioritize strategic climate change adaptation responses. For the purposes of our study, we identified four categories of government that play a central decision-making role in climate change adaptation governance: (1) New York State (NYS) government, (2) New York City (NYC) government, (3) federal government, and (4) local government (other than NYC). These categories reflect the three jurisdictions relevant to climate change–related governance and also the higher degree of influence and autonomy of NYC compared with the state’s other municipalities.

Table 1 lists the sectors (with associated components) relevant to climate change–related governance in New York, as identified by both NYS and NYC reports. The state’s home rule laws entail that virtually all policy measures involving land use must be coordinated with the municipalities having jurisdiction over that area. This includes, but is not limited to, energy production, water procurement and disposal, coastal zoning and flood prevention, forest management, and food production. NYC, in particular, engages in strategic planning for climate change adaptation using multi-stakeholder task forces and has produced three thematic reports on the topic identifying the city’s risks and outlining policy options (NYC Panel on Climate Change, 2010, 2013, 2015).

Table 1.

Participation in Climate Change Policy Tasks Across Jurisdictions.

Sector	Component	Jurisdiction
		Federal	NYS	Local
		Federal	NYS	NYC	Other local
Energy	Production	×	×	×	×
	Transmission and distribution	×	×	×	×
Transportation	Roadways	×	×	×	×
	Transit	×	×	×	×
Water resources	Quantity	×	×	×	×
	Distribution of supply		×	×	×
	Quality	×	×	×	×
Waste management	Wastewater	×	×	×	×
	Solid waste	×	×	×	×
Telecommunications	Power supplies	×	×	×	×
	Equipment	×	×	×	×
Coastal zones	Zoning and building codes	×	×	×	×
	Flood prevention	×	×	×	×
	Marshes and aquatic ecosystems	×	×	×	×
Ecosystems	Forests and grasslands	×	×	×	×
	Noncoastal aquatic ecosystems	×	×	×
Agriculture	Food and agricultural production	×	×	×	×
	Food safety	×	×	×	×
Public health	Infectious disease	×	×	×	×
	Emergency response	×	×	×	×

Note. NYC = New York City; NYS = New York State. Adapted from lists in reports by the NYC Panel on Climate Change (2010) and Rosenzweig et al. (2011) for the NYS Energy Research and Development Authority.

At the same time, NYS has assumed the responsibility of coordinating policy measures with the municipalities in pursuance of statewide objectives such as greenhouse gas abatement and climate change adaptation. The NYS government has more control over the state’s energy portfolio than the municipalities and negotiates directly with the federal Environmental Protection Agency and with other states over energy and environmental policy. Moreover, the NYS government has substantial land holdings in the form of state parks, and roads, and also has jurisdiction over the management of the fisheries along its coast.

Figure 1 is a simplified representation of the tasks of each jurisdiction which highlights the fact that, while all of the jurisdictions exercise a degree of autonomy in the management of their land holdings (and thus are all implicated in climate change adaptation measures in these areas), they also serve unique functions. The NYS government engages in statewide strategic planning affecting the municipalities whose behavior it seeks to influence. With respect to climate change, the NYS government has convened several climate change task forces that have produced reports on the extent and consequences of climate change for the state that contain policy recommendations for state and municipal agencies to implement.³ NYC arguably blurs the line between state and municipal given the extent to which, like the state government, it also engages in long-term strategic planning for climate change mitigation and adaptation. The federal government is implicated in nearly all of the sector components listed in Table 1. However, apart from its management of the natural resources under its direct control (which is comparatively minimal in NYS as a percentage of total land), the federal government’s role rests most significantly in its regulation of water and air quality, food safety, waste disposal, as well as its participation in managing natural disasters.

Figure 1.

Unique tasks in New York climate change policy.

This substantial jurisdictional diversity underscores the fact that interagency collaboration, knowledge sharing, and co-learning, both within and among jurisdictions, is a necessary precondition for effective and coherent intersectoral climate change policy integration.

Materials and Methods

Data

Qualitative and quantitative data were obtained from primary sources through a survey of, and face-to-face interviews with, public agency staff working to address climate change issues in NYS. All data were collected between December 2011 and February 2012. The qualitative data consists of key informant interviews that were conducted with 20 senior civil servants working in management and policy analysis positions related to sustainability and climate change abatement/adaptation in the NYS government (14), NYC government (4), and other local governments (2). Following a semistructured design, we sought descriptive insights into the role of collaboration, interagency trust, and formal and informal communication networks on climate change–related policy and implementation outcomes in government. Each interview was anonymous and took approximately 1 hour to complete.

The quantitative data were gathered using an online survey instrument. We identified potential participants using a snowball sampling strategy, and from publicly available online resources such as agency reports related to climate change and agency websites. As with the qualitative data, we specifically targeted participants involved in policy making and implementation related to sustainability and climate change. However, here we sought respondents from a larger population of agencies and in a wider array of positions within these agencies. Overall we received 103 completed responses, including staff members of the NYS government (59), the NYC government (21), county and other municipal governments (8), federal employees stationed in New York whose work related to climate change policy in that region (14), and a regional intergovernmental agency (1). Our sample exhibited diverse representation of gender (male: 53%; female: 47%), government agency focus (main responses: environmental regulation, 29%; parks and recreation, 16%; transportation infrastructure, 16%; planning and administration, 14%; sustainability and climate change, 12%); work role (main responses: management, 19%; natural science/research, 16%; environmental assessment, 14%; policy analysis, 10%; executive, 8%), and years working at current agency (<5 years, 60%; >5 years, 40%).

Following these demographic questions, participants were instructed to select the government agency at which they were employed at the time, and select from a list of agencies with which they communicate with in their professional role. The survey then prompted participants to answer questions related to interagency trust, formal and informal communication, and influence, with each of the agencies selected.

Measures

Although several measures of organizational trust have been developed within the field of organizational management, no reliable measure of interagency trust exists for the public sector (Arnott, 2007; Seppänen, Blomqvist, & Sundqvist, 2007). We thus adapted an existing measure of private sector interorganizational trust, developed by Nooteboom, Berger, and Noorderhaven (1997), consisting of six questions designed to operationalize two dimensions of organizational trust (see also Temby, Rastogi, Sandall, Cooksey, & Hickey, 2015, in the context of transboundary Pacific salmon governance). The first trust dimension, labeled “institutionalization” by Nooteboom et al., refers to trust based on the existence of norms that decrease the likelihood that one party will take advantage of the other, and is utilized because it lowers transaction costs among parties. The second trust dimension, “habitualization,” refers to “attachments . . . in the form of friendship or kinship bonds,” and “depends on time and context, on habit formation, and on the positive development of a relation” (Nooteboom et al., 1997, p. 314).

Because the context of application for these items was the public sector as opposed to the private sector (including a necessary rewording of the questions to suite the public sector context), we employed principal component analysis with promax rotation (to permit correlated components) to verify the alignment of the six trust-related items on the survey (from the Nooteboom et al., 1997, conceptualization) with the originally intended trust dimensions. Two components were subsequently retained for interpretation, both with eigenvalues exceeding 1.0 (for details on methods, see Temby et al., 2015). As a result of this procedure (which suggested a modified alignment of the survey questions with underlying trust dimensions), we labeled our two trust measures as follows: (1) Fair Play, which combined three trust-related items reflecting relationship expectations, that is, that a collaborator would be fair and unbiased in their dealings and (2) Relational Comfort, which combined three trust-related items reflecting longer term investment in the relationship, that is, that the relationship with a collaborator had gone on long enough for it to become comfortable, well understood, and equally reliant on informal and formal outcomes (see Table 2). These two components were moderately positively correlated (r = .36). The questions that represented these two dimensions were based on a 5-point Likert-type scale (see Table A1 of the appendix). Importantly, these two dimensions represent an operationalization of the two previously described trust dimensions identified by Stern and Coleman (2015; i.e., Fair Play represents “procedural trust” and Relational Comfort represents “affinitive trust”).

Table 2.

Pattern and Correlation Matrices From the Principal Components Analysis of the Six Trust Items.

Pattern matrix
Trust item content	Trust component
Trust item content	Fair Play	Relational Comfort
INSTA: Expected not to make demands that seriously damage interests of other	.77
HABC: Never felt misled in contacts with this agency	.75
INSTB: Expected not pursue own interests at all costs	.69
HABA: Working with agency so long, procedures have become self-evident		.95
HABB: Working with agency so long, understand each other well and quickly		.66
HABINST: Informal agreements have same significance as formal contracts		.63
Component correlation matrix
Trust component	1	2
Fair Play (expectations of relationship)	.60^a
Relational Comfort (longer term investment in relationship)	.36	.65

Cronbach’s alpha reliability in diagonal cells.

We also created measures to obtain insights into the formality and informality of the interagency collaborative networks. These networks were operationalized as communicative networks and measured by asking respondents which agencies they communicated with via informal and/or formal channels. Informal channels here include chance conversations, spontaneous meeting, personal e-mails and phone calls, and coffee or drinks outside of work. Formal channels include committee meetings, memos, and official written or verbal communication. If a respondent reported communicating with an agency through informal or formal channels they were asked a follow-up question about the frequency of that communication (occasional or regular). Answers to these questions represented values for Informal Communication Frequency and Formal Communication Frequency (see Table A2 of the appendix).

Furthermore, we sought to measure the effects of trust and formal and informal communication on mutual learning and adjustment within collaborative networks, and created variables representing our operationalization of this concept. As we were interested in the communicative avenues though which learning takes place, and the effects of this collaboration, we distinguished between two types of mutual learning and adjustment: that which occurred as a result of formal communication and that which resulted from informal communication. For each of these two measures, and consistent with Agranoff’s definition of mutual learning and adjustment (quoted above), respondents were asked to indicate how often communication (formal or informal) had led them to make professional choices and decisions they would not have otherwise made. The answers to these questions represented values for Informal Communication Impact and Formal Communication Impact on decision making.

Four questions were included in the survey to elicit information on the extent to which respondents saw collaboration and scientific information as important in their professional role, whether the information available to them from other agencies met their needs, and their satisfaction with available scientific information. Measures based on these questions were included to control for the influence of preexisting dispositions about these topics on feelings of trust or the tendency to discuss work-related matters across agency boundaries. A second principal components analysis was conducted to condense the four professional role-related items (Collab, SciCrit, InfoAdeq, and SciSat). Two components were examined for further interpretation, both with eigenvalues exceeding 1.0. The first component reflected a composite combining the InfoAdeq and SciSat items (pattern coefficients of .87 and .83, respectively; Cronbach’s α = .66). This component was labelled InfoQual, reflecting the extent to which a participant felt that the quality of information made available by other agencies was adequate for performing their role. However, the second component, combining the Collab and SciCrit items, was found not to have an acceptable level of reliability (Cronbach’s α = .30) and was not retained. The two items were therefore used as single-item measures in subsequent analyses. Table A3 of the appendix lists the survey questions comprising these measures.

For all measures where components were involved, participants were given scores on a component by averaging the nonmissing scores of the constituent items within that composite.

Data Analysis

Network diagrams were created to present the extent to which civil servants working on climate change–related issues in New York communicate with staff in other agencies within and across jurisdictions, and also to depict the patterns of trust among these agencies (see also, Temby et al., 2015, in the context of transboundary Pacific salmon governance). Scores for informal and formal communication between agencies (standardized on a 0 to 1 scale) were calculated separately for each case and averaged across the agencies within a specific jurisdiction to indicate communicative intensity based on the agencies that respondents reported communicating with. Interagency trust was analyzed using mean response scores on the six trust variables for each case and standardizing these on a −1 to 1 scale to represent the average scores across agencies of each of the four jurisdictions (Temby et al., 2015).

Two distinct hierarchical regression models were constructed to examine the relationships between various predictor variables from the survey and the

impact of informal channel communications with specific target agencies on the participant’s decision making

impact of formal channel communications with specific target agencies on the participant’s decision making.

Two-way interactions, designed to rule out the moderating effect of trust on various aspects of interagency communication, were also tested in the hierarchical regression models. As per recommendations by Cohen, Cohen, West, and Aiken (2003), each interaction term was created by first centering (i.e., subtracting the predictor mean from each participant’s score) each main effect predictor involved in the interaction. Then the resulting differences were multiplied together to produce the interaction predictor. This process ensured that the interaction terms were not multicollinear with their main effect constituents. Seven interactions terms were created for incorporation into specific hierarchical regression models:

Fair Play Trust by Relational Comfort Trust (FP * RC)

Fair Play Trust by Frequency of Informal Communication (FP * FI)

Fair Play Trust by Impact of Informal Communication (FP * II)

Fair Play Trust by Frequency of Formal Communication (FP * FF)

Relational Comfort Trust by Frequency of Informal Communication (RC * FI)

Relational Comfort Trust by Impact of Informal Communication (RC * II)

Relational Comfort Trust by Frequency of Formal Communication (RC * FF)

Since the survey required each participant to rate multiple target agencies, there was effectively a repeated measures aspect to the survey design. If included in a regression analysis, the 103 survey participants with valid responses would have required 102 dummy-coded predictors to represent. This was untenable as an analytical strategy. However, for repeated measures designs, Pedhazur (1977; see also, Gibbons & Sherwood, 1985) described an alternative analytical strategy for encoding predictors with a large number of categories for regression analysis, namely criterion scaling. We applied this strategy to effectively identify each individual participant in a single predictor, by using each participant’s mean score on the dependent variable as the predictor value for all target agencies she or he rated. As shown by Pedhazur (1977), the only modification to the regression analysis strategy that this procedure entails is recalculating the F test for the predictor, using the original degrees of freedom (in this case, the degrees of freedom for participants = 102) instead of the single degree of freedom that the criterion scaled predictor would yield and reducing the residual degrees of freedom accordingly. For model testing purposes, a more stringent significance criterion of p < .01 was adopted, given the large number of residual degrees of freedom available for all tests and to protect against error inflation due to the conduct of multiple significance tests.

For purposes of ease of incorporation into analyses and interpretation, both participant and target agencies were aggregated into four groups: (1) Federal Government, (2) NYS agencies, (3) NYC agencies, and (4) Non-NYC Local. These aggregated Participant and Target agency categories were then dummy-coded for use as regression predictors, in each case using Non-NYC Local as the reference category. This process yielded three dummy-coded predictors for Participant agencies and three for Target agencies.

Predictor sets were defined and entered into each hierarchical regression model as shown in Table 3, which also identifies the dependent variable for each analysis. The hierarchical ordering of predictor sets was determined using the following general logic: most general, group-level, and least relationship-specific predictors entered early to most specifically targeted and most relationship-specific predictors entered late. The specific order entry logic for each predictor set is also shown in Table 3. Interactions were always entered into a model after all their constituent main effects had been entered and accounted for. Model 2 error (Cohen et al., 2003), based on the residual unexplained variance after all predictors had entered the model (1 − R²), was used for all F tests for assessing predictor set contributions and, where appropriate, the contributions of predictors within a predictor set. This ensured that the error term was not contaminated by any potential systematic effects. The degrees of freedom adjustment for the use of a criterion-scaled predictor was also implemented.

Table 3.

Summary of the Hierarchical Regression Predictor Sets and the Order in Which They Entered Each Regression Model.

Predictor sets in order entered	Logic for ordering of predictor set	Impact of informal communication	Impact of formal communication
Participant agencies (three dummy-coded variables)	Codes the most general way of classifying survey participants by type of agency they work for, irrespective of target agency they relate to	1	1
Professional role expectations (Collab and SciCrit, and the Information Quality component)	Reflects general professional role expectations held by the participant as a member of the agency group they belong to, irrespective of target agency they relate to	2	2
Criterion-scaled participants predictor	Codes individual participants to control for individual differences in rating relationships with individual agencies	3	3
Target agencies (three dummy-coded variables)	Codes the specific agency that is a target for trust development and communications for an individual participant	4	4
Fair Play Trust component (FP)	Assesses the level of Fair Play Trust associated with a specific target agency, effectively as a precursor to building an inter-agency relationship	5	5
Frequency of Informal Communication (FI)	Given Fair Play Trust has been accounted for, assesses how frequently the individual participant informally communicates with a specific target agency	6	6
Relational Comfort Trust component (RC)	Assesses the level of Relational Comfort Trust associated with a specific target agency, effectively built up over a period of time	7	7
Trust component interactions I (FP * RC, FP * FI, and RC * FI)	Two-way interactions between trust components and between trust components and frequency of informal communication, entered after the relevant main effects have been accounted for	8	8
Impact of Informal Communication (II)	Assesses the impact of informal communication on individual participant’s own choices and decision making, after trust has been established and frequency of informal communication has been accounted for	DV	9
Trust component interactions II (FP * II and RC * II)	Two-way interactions between trust components and impact of informal communication, entered after the relevant main effects have been accounted for	—	10
Frequency of Formal Communication (FF)	Given trust levels and informal communication frequency and impact have been accounted for, assesses how frequently the individual participant formally communicates with a specific target agency	—	11
Trust component interactions III (FP * FF and RC * FF)	Two-way interactions between trust components and frequency of formal communication, entered after the relevant main effects have been accounted for	—	12
Impact of Formal Communication (IF)	Assesses the impact of formal communication on individual participant’s own choices and decision making, after trust has been established, after informal communications have been accounted for, and frequency of formal communication has been accounted for	—	DV

Note. DV = dependent variable.

For each analysis, the hierarchical model summary table and predictor-specific coefficients are reported. For predictor sets with more than one variable, individual predictor contributions to the model were examined only if the predictor set as a whole was found to make a significant contribution. Following Field’s (2013) suggestion, for each individual predictor, both unstandardized and standardized regression coefficients as well as the part correlation are reported; the square of the part correlation is also reported (identified as the squared semipartial correlation, sr²—see Cohen et al., 2003) and reflects the proportion of variance that the predictor uniquely explains in the dependent variable over and above all previously entered predictors.

Last, qualitative interview data were transcribed and analyzed using content analysis techniques (Babbie, 2012). We examined the patterns and relationships yielded by the social network and hierarchical regression analysis and mined the qualitative data for discursive explanations of these phenomena. The aim in analyzing complimentary data in this way was to develop clarifications and richer explanations of our findings than would be possible using survey data alone (Bailey, 2007).

Limitations and Assumptions

Some important limitations and assumptions of our study need to be acknowledged. First, our research focused on climate change–related policy issues in a broad sense and went beyond specific government activities and initiatives related to formal climate change adaptation or mitigation policies. As noted above, the civil servants we surveyed self-identified their particular role in climate change–related governance, representing a wide variety of agencies and, in many instances, dealing with climate change–related issues only occasionally or indirectly through their role-specific tasks (e.g., forest management). This decentralized and cross-cutting nature of climate change policy challenges is not only a central motivation of the problematique we seek to address, but it also limits our concern with the specific policy initiatives with which our sample participated.

Second, our single case study approach lacks what Gerring and McDermott (2007) call “spatial variation” across case studies, and thus limits the generalizability of our findings to theory. Third, our data is cross-sectional and thus lacks temporal variation. Recognizing the limitations associated with cross-sectional and purposive sample data, we undertook several measures to strengthen the reliability and internal validity of our results, including defining the measures of our study before initiating data collection, pretesting our survey instrument to reduce the potential for bias, and consulting with experts at different stages of the study to validate and triangulate the results (Miles & Huberman, 1984; Uiterkamp & Vlek, 2007).

Results and Discussion

Mapping Formal and Informal Communication and Trust

The results of our descriptive network analysis are presented in Figures 2 and 3. Associated with all of the relationships depicted is a percent communicating (PC) measure, which is the percent of respondents from the source jurisdiction indicating that they communicated with the target jurisdiction. As a substantial majority of respondents were from NYS and NYC agencies, the federal government and non-NYC local governments are included only as target jurisdictions. The results indicated that NYS and federal government were the most common points of contact for the sample of NYS government civil servants working on climate change–related issues, as ≥95% of respondents reported communicating with the former and 83% with the latter, irrespective of source agency jurisdiction. Specific municipalities (NYC among them) were less contacted by the state government and between each other. Only 58% of respondents in the state government reported communication with employees in the NYC government, 73% with all other local government agencies, and 50% of respondents from NYC reported being in contact with other municipalities.

Figure 2.

Distribution of formal and informal communication among agencies.

Figure 3.

Distribution of trust among jurisdictions.

Figures 2 and 3 also present the results of the analysis of the distribution of communicative intensity and trust among jurisdictions comprising the network. In each, the thickness of the arrows represents the relative size of the corresponding number. Figure 2 shows that formal communicative intensity (FCI; solid arrows) was higher than informal communicative intensity (ICI; dashed arrows) among respondents reporting communication with the target jurisdiction in every inter- and intrajurisdictional relationship.

Figure 3 presents data in a similar way, and features brackets around the arrows representing a negative value between a source and target jurisdiction on one of the two trust scales. As with the values for communicative intensity in Figure 2, all trust values refer only to the cases that reported contact with the target agency. Figure 3 indicates that all values for Relational Comfort (RC; solid arrows) were negative except for those referring to interagency relationships within jurisdictions. Especially notable were the high negative levels of RC from NYS to NYC (−.12), NYC to NYS (−.14), NYC to other local governments (−.15), and NYC to the federal government (−.20). The latter value was striking given that many of the federal officials that NYC employees maintained contact with were geographically located within NYC. Yet this value was repeated in the score that NYC employees gave the federal government for the other category of trust, Fair Play (FP; dashed arrows). With this notable exception, there did not appear to be a meaningful pattern in the distribution of FP among agencies.

The results suggested that informal channels were utilized less often than formal channels among civil servants in regular communication, and that a relative absence of familiarity and comfort in interactions (i.e., Relational Comfort) was more common among professionals in agencies in different jurisdictions. The finding about informal communication was curious, as some of our interviewees testified to the value of the informal, in particular of undertaking projects without upper-level support. It is plausible that part of the explanation lies in the availability of opportunities for informal communication. Several interviewees indicated that cost-saving travel restrictions in place for NYS employees since the 2008 financial crisis had adversely affected their ability to develop and maintain relationships with colleagues in other parts of the state.

The Relational Comfort findings for the NYS–NYC agency relationship were consistent with the fact that these government agencies had developed a degree of relative autonomy in their climate change policy-making programs. The two jurisdictions developed relatively distinct policy discourses for NYC’s land-use approach to climate change adaptation, with minimal involvement from the other. Indeed, interview participants indicated that the turf battles in New York were more common between these jurisdictions than within them, and that NYC in particular guarded its autonomy.

Trust, Formal/Informal Communication, and Communicative Impact

Table 4 reports the hierarchical regression analysis for predicting the impact of informal communication on decision making among collaborative network participants. The summary of the model building process shows that all predictor sets entered contributed significantly to prediction of the impact of informal communication on decision making except for the Participant and Target Agencies set. Within the Professional Role Expectations set, only the Collab variable contributed significantly. The more important collaboration was to the participant’s role, the greater the impact of informal communication on decision making tended to be. The Criterion-scaled Participants predictor, as expected, explained a significant amount of variance in impact of informal communication over and above what the more general group-level classification and associated role expectations could account for. Higher Fair Play Trust predicted greater impact of informal communication on decision making. Not unexpectedly, higher frequency of informal communication very strongly predicted greater impact of those communications. Higher Relational Comfort Trust predicted greater impact of informal communication on decision making, and this relationship was somewhat stronger than for Fair Play Trust. No interaction variables were found significant, indicating the independence of the two dimensions of trust. The entire regression model predicted about 65% of the variability in impact of informal communication.

Table 4.

Hierarchical Regression Analysis Predicting Impact of Informal Communication on Decision Making.

Hierarchical regression model summary^a
Model	Predictor set entered		R² change statistics
Model	Predictor set entered	Model R²	R² change	df predictors	df residual	Model 2 F-test	p
1	Participant’s Agency	.001	.001	3	966	0.61	.608
2	Professional Role Expectations	.011	.011	3	963	8.46	<.001
3	Criterion-scaled Participants	.410	.398	102	861	9.37	<.001
4	Target Agency	.411	.002	3	858	1.28	.279
5	Fair Play Trust Component (FP)	.416	.005	1	856	11.54	<.001
6	Frequency of Informal Communication (FI)	.632	.216	1	857	519.19	<.001
7	Relational Comfort Trust Component (RC)	.642	.010	1	855	23.76	<.001
8	Trust Interactions I (FP * RC, FP * FI, RC * FI)	.645	.003	3	852	2.03	.108
Model coefficients
Model	Individual predictors	Unstandardized coefficients		Standardized coefficients	Correlations		Model 2 F-test
Model	Individual predictors	Β	Std. error	β	Part	sr ²	F	p
1	Participant’s Agency: Federal	−.012	.116	−.004	−.003	.000	0.02	.883
	Participant’s Agency: NYS	−.011	.064	−.009	−.006	.000	0.09	.769
	Participant’s Agency: NYC	−.048	.069	−.034	−.022	.000	1.16	.281
2	Collaboration important to role (Collab)	.114	.037	.102	.098	.010	23.05	<.001
	Science information critical to role (SciCrit)	−.006	.023	−.008	−.008	.000	0.15	.695
	Information quality adequate for role (InfoAdeq)	.011	.026	.014	.014	.000	0.47	.493
3	Criterion-scaled Participants	1.000	.039	.640	.631	.398	9.37	<.001
4	Target Agency: Federal	.054	.071	.024	.019	.000	0.87	.352
	Target Agency: NYS	−.022	.050	−.017	−.011	.000	0.28	.596
	Target Agency: NYC	.026	.054	.019	.012	.000	0.35	.554
5	Fair Play (FP)	.073	.026	.074	.069	.005	11.54	<.001
6	Frequency Informal Communication (FI)	.469	.020	.525	.465	.216	519.19	<.001
7	Relational Comfort (RC)	.104	.020	.118	.099	.010	23.76	<.001
8	FP * RC	.038	.021	.040	.035	.001	2.99	.084
	FP * FI	−.025	.029	−.021	−.017	.000	0.70	.402
	RC * FI	.047	.025	.042	.035	.001	3.00	.083

Dependent variable: Impact Informal. Note. Bold font denotes statistically significant predictors and values.

Table 5 reports the hierarchical regression analysis for predicting the impact of formal communication on decision making. The summary of the model building process shows that all predictor sets entered contributed significantly to prediction of the frequency of formal communication except for Target Agency, Fair Play Trust, and all of the trust component interaction sets. If the participant’s agency was the Federal Government, a NYS agency or a NYC agency, this predicted a significantly lower average frequency of formal communication compared to Non-NYC Local agencies (the reference category). Within the Professional Role Expectations set, the Collab and InfoAdeq variables contributed significantly. The more important collaboration was to the participant’s role and the more adequate the information provided by other agencies was for their role, the greater the impact of formal communication tended to be. The criterion-scaled Participants predictor, as expected, explained a significant and large amount of variance in impact of informal communication over and above what the more general group-level classification and associated role expectations could account for. Higher frequency of informal communication predicted greater impact of formal communication on decision making. Higher Relational Comfort Trust also predicted greater impact of informal communication on decision making. Greater impact of informal communication on decision making significantly predicted greater impact of formal communication. Greater frequency of formal communication, not surprisingly, strongly predicted higher impact of formal communication on decision making. As in the previously reported model, none of the interaction variables were found significant. The entire regression model predicted about 58% of the variability in impact of formal communication.

Table 5.

Hierarchical Regression Analysis Predicting Impact of Formal Communication on Decision Making.

Hierarchical regression model summary^a
Model	Predictor set entered	Model R²	R² change statistics
Model	Predictor set entered	Model R²	R² change	df Predictors	df Residual	Model 2 F-test	p
1	Participant’s Agency	.014	.014	3	966	9.10	<.001
2	Professional Role Expectations	.033	.020	3	963	13.10	<.001
3	Criterion-scaled Participants	.334	.301	102	861	5.93	<.001
4	Target Agency	.337	.003	3	858	2.01	.111
5	Fair Play Trust Component (FP)	.338	.001	1	856	2.75	.097
6	Frequency of Informal Communication (FI)	.432	.094	1	857	189.27	<.001
7	Relational Comfort Trust Component (RC)	.453	.021	1	855	41.91	<.001
8	Trust Interactions I (FP * RC, FP * FI, RC * FI)	.454	.001	3	852	0.70	.551
9	Impact Informal Communications (II)	.498	.044	1	851	88.14	<.001
10	Trust Interactions II (FP * II, RC * II)	.498	.000	2	849	0.10	.907
11	Frequency of Formal Communication (FF)	.576	.078	1	848	155.85	<.001
12	Trust Interactions III (FP * FF, RC * FF)	.579	.004	2	846	3.69	.025
Model coefficients
Model	Individual predictors	Unstandardized coefficients		Standardized coefficients	Correlations		Model 2 F-test
Model	Individual predictors	Β	Std. error	β	Part	sr ²	F	p
1	Participant’s Agency: Federal	−.291	.113	−.092	−.082	.007	13.52	<.001
	Participant’s Agency: NYS	−.214	.062	−.170	−.110	.012	24.33	<.001
	Participant’s Agency: NYC	−.166	.068	−.117	−.078	.006	12.23	<.001
2	Collaboration important to role (Collab)	.114	.036	.104	.100	.010	20.11	<.001
	Science information critical to role (SciCrit)	−.025	.022	−.038	−.036	.001	2.61	.107
Model coefficients
Model	Individual predictors	Unstandardized coefficients		Standardized coefficients	Correlations		Model 2 F-test
Model	Individual predictors	Β	Std. error	β	Part	sr ²	F	p
	Information quality adequate for role (InfoAdeq)	.063	.026	.081	.078	.006	12.23	<.001
3	Criterion-scaled Participants	1.017	.049	.579	.548	.301	5.93	<.001
4	Target Agency: Federal	.145	.074	.064	.051	.003	5.32	.021
	Target Agency: NYS	.036	.052	.029	.018	.000	0.69	.407
	Target Agency: NYC	.063	.057	.046	.029	.001	1.74	.187
5	Fair Play (FP)	.038	.027	.039	.037	.001	2.75	.097
6	Frequency Informal Communication (FI)	.281	.022	.320	.307	.094	189.27	<.001
7	Relational Comfort (RC)	.151	.025	.174	.144	.021	41.91	<.001
8	FP * RC	.003	.025	.003	.003	.000	0.02	.893
	FP * FI	−.012	.035	−.010	−.008	.000	0.13	.720
	RC * FI	−.033	.031	−.030	−.025	.001	1.26	.263
9	Impact Informal Communication (II)	.309	.034	.314	.209	.044	88.14	<.001
10	FP * II	−.023	.054	−.018	−.010	.000	0.20	.654
	RC * II	.011	.046	.011	.006	.000	0.07	.788
11	Frequency Formal Communication (FF)	.361	.027	.328	.278	.077	155.41	<.001
12	FP * FF	−.010	.048	−.006	−.005	.000	0.05	.823
	RC * FF	.114	.041	.077	.059	.003	7.00	.008

Dependent variable: Impact Formal. Note. Bold font denotes statistically signifant predictors and values.

Figures 4 and 5 provide a succinct diagrammatic summary of the two hierarchical regression analyses. Figure 4 summarizes the outcomes from the model predicting the impact of informal communication on participant decision making. Figure 5 summarizes the outcomes from the model predicting the impact of formal communication on participant decision making. The arrowheads point to the dependent variable in each regression model and the overall R² associated with each model is shown in bold, adjacent to the relevant dependent variable. The strength and direction of each relationship is shown by the standardized regression coefficient attached to each predictor–dependent variable link; all coefficients shown are significant at p < .01. The hierarchical predictor sets are separated by short dotted lines and the change in R² associated with the addition of that predictor set to each regression model is shown in solid-line boxes; all R² shown are significant at p < .01. The order of set entry is down the left side of the figure.

Figure 4.

Summary of significant hierarchical regression relationships for predicting impact of informal communication.

Figure 5.

Summary of significant hierarchical regression relationships for predicting impact of formal communication.

These figures illustrate the importance of both informal communication and trust in facilitating learning and adjustment in collaborative networks. In each, the frequency of informal communication has a pronounced effect on mutual learning and adjustment through formal and informal channels. Furthermore, informal communicative impact, itself largely determined by informal communication frequency, substantially influences mutual learning taking place through formal channels. The figures also illustrate that, although the two dimensions of trust promote mutual learning and adjustment, the existence of a long-term professional relationship (Relational Comfort) is the most important predictor for both dimensions of learning and adjustment. These findings are significant despite controlling for a wide range of potential confounding variables, including home jurisdiction, target jurisdiction, and existing attitudes about collaboration.

The interviews indicated ways in which Relational Comfort and informal communication can have the pronounced effects on mutual learning and adjustment across agencies identified in this analysis. Several participants suggested that formal policy processes are accompanied by informal communications for the purposes of discussing the parameters of an issue and building understanding and consensus. As stated by a staff member of the NYS Environmental Protection Bureau in the Office of the Attorney General: “Climate change is too inter-disciplinary, too complex. Energy efficiency cuts across many different organizations and areas of expertise. So you need formal and informal relationships.” A staff member at the NYC Department of Environmental Protection explained that the formal state-level task forces provided an opportunity for him to development relationships with people he would otherwise not know, facilitating further (less formal) collaboration. Civil servants at other agencies echoed these sentiments, explaining that formal mechanisms are often used as ways to introduce young staff to the people they will continue to work with on a long-term basis.

It was through these long-lasting relationships and informal contact that the process of mutual information sharing unfolded. Interviewees revealed that one way in which informal communication with members of other agencies enables learning and influences decision making is through the education of staff on the complexities associated with the problems being faced. In the words of a staff member in the NYS Department of Environmental Conservation’s Office of Climate Change:

I think that has happened to a very large extent. For me personally, I came into this with no background in climate science, and no particular expertise in climate adaptation or mitigation, nor has anyone we’ve talked to [in these other agencies]. We’re always sharing information from our various [professional] backgrounds.

When asked how they and others learn about the science of climate change, he responded: “We talk to each other, we talk to other experts.” Indeed, the NYS Department of Environmental Cooperation’s Office of Climate Change coordinates a six-agency informal climate change adaptation working group which seeks ways of developing ways of implementing the recommendations of formal task force reports in accordance with existing authority. Our analysis and interviews suggest that informal entities of this sort are important for building capacity among formal decision-making channels by developing solutions to problems and sharing information. Repeatedly, interviewees explained that knowing people from other agencies, and being able to contact them easily and informally, were crucial for the complex learning-oriented work they perform.

Conclusion

This article has sought to contribute to a dialogue on the capacity of collaborative public management networks to facilitate mutual learning and adjustment across agencies working to address complex environmental challenges. Although interagency coordination is important for a variety of service delivery and long-term management problems faced by governments, climate change–related policy is a particularly instructive example of a dynamic problem in need of a coordinated and integrative solution. Its intersectoral suite of challenges, characterized by temporal uncertainty, an overlapping suite of governmental tasks, and a need for a wide range of policy-relevant knowledge, suggest that policy makers and civil servants will need to find enduring ways of collaborating that cultivate a diversity of interpersonal connections and sources of expertise.

Mutual learning and policy integration are, of course, complex and multifaceted problems with many possible angles for inquiry. Our approach has been to focus on behavioral and organizational processes rather than an administrative or policy outcome. This involved, first, assessing the collaborative network’s properties in terms of connectivity (i.e., the extent to which civil servants in different agencies communicate with one another), the nature of this connectivity in terms of whether it occurs through formal or informal means, and the distribution of two dimensions of trust. Second, we examined the effects of these network properties on two measures of interagency influence, controlling for a range of variables. This operationalization of mutual learning and adjustment is based on subjective respondent assessments about whether formal or informal communication with civil servants from other agencies had caused them to act differently in their professional role than they would have otherwise.

Our analysis revealed that formal communication among staff at different agencies is utilized more often than informal communication and that interagency relationships are more characterized by a feeling of “fair play” than by “relational comfort.” Yet our predictive models indicate that it is informal communication and Relational Comfort that are the most important in facilitating coordination. Our interviews with New York civil servants yielded some insights into how these processes may occur—notably by facilitating the sharing of information and by building capacity.

These findings reinforce Imperial’s (2005) and Agranoff’s (2007) similar observations about the role of informal networks performing these tasks. Yet they also provide more rigorous empirical foundation about the pathways through which informal networks interact with formal and enhance learning and coordination, and offer clues about the dimension of trust that tends to be the most important. Furthermore, this research suggests that informal networks function separate from (although interacting with) formal networks. Rather than informal as merely an early phase of relationships that later become formalized (as suggested by Abram, Mahaney, Linhorst, Toben, & Flowers, 2005; and Imperial, 2005), they are an integral and indispensable part of the policy implementation process. This finding was supported by our qualitative interview data.

On trust, our findings confirm those of Edelenbos and Klijn (2007), that it facilitates cooperation in collaborative public management networks. However, it also builds on this research. Edelenbos and Klijn reported survey findings of network participants’ assessments of trust’s influence on cooperative arrangements in the network as a whole. Because participants in our survey identified specific agencies as referents for trust and communicative impact, our findings include details about the distribution of trust and a more robust assessment of its effects. Furthermore, we disaggregate trust; our dimensions of Fair Play and Relational Comfort effectively operationalize Stern and Coleman’s (2015) distinction between “procedural” trust and “affinitive” trust, respectively, which they identify as being relevant to the management of complex environmental issues. Our findings provide some much needed empirical evidence to this discourse, suggesting that affinitive trust is of greater importance to civil servants than trust based on positive procedure-oriented assessments of how fair other agencies are in their dealings. This has potentially substantial implications for the management of collaborative public management networks seeking to respond to complex environmental issues. It also suggests the need for scholars to consider whether studies of trust development in public organizations based on the perceived trustworthiness of another person or trust transferability (see Lambright et al., 2010) are examining the types of trust most relevant to the administrative policy making and implementation setting.

The importance of informal communication and a Relational Comfort on interagency coordination, coupled with the underutilization of informal communication in relation to formal communication, suggest that affinitive trust and the informal dimension to management may be an underutilized resource in the governance of complex environmental and other intersectoral issues. These are not qualities that can be instituted effectively through bureaucratic rules concerning transparency. Interagency working groups, “climate czars,” or other similar arrangements are important in getting the relevant players acquainted and working with one another through formal channels, but they are likely to require significant informal “back-channeling” to be effective—what Klijn et al. (2010, p. 1069) refer to as “connecting strategies.” Indeed, in a recent study on the effectiveness of different categories of collaborative network management strategies for the perceived outcomes of networks addressing environmental issues, Klijn et al (2010, p. 1069) found “connecting strategies” the most effective. These included “initiating new series of interactions, coalition building, mediation, appointment of process managers, removing obstacles to cooperation, [and] creating incentives for cooperation.” Similarly, Edelenbos and Klijn (2007, p. 42) showed that interorganizational network trust “requires constant nurturing through process management” to develop. Actions such as actively encouraging staff to work away from their desks and embracing opportunities to meet others in different agencies at various levels of government working on similar and different work would be valuable here. For example, federal and state civil servants would likely benefit from being both formally and informally supported to develop strong professional relationships with the local officials necessary for the implementation of climate change–related policy—especially “home-rule” states like New York.

These findings offer important insights, applicable to a range of public management issues, concerning the role of informal and ephemeral dimensions of collaborative networks in developing the capacity (i.e., Agranoff and McGuire’s concept of “groupware”) for mutual learning and adjustment among actors in different organizations. Our findings can also inform public administration strategies to develop well-functioning collaborative networks in order to underpin the sustainable governance of complex environmental management challenges internationally (see Healy et al., 2014; Selin & VanDeveer, 2009). Toward developing robust “theories of sustainability management” (Starik & Kanashiro, 2013, p. 7), we suggest that studies of environment-related decision making and learning through collaborative networks could benefit from engaging with the environmental policy integration literature (Jordan & Lenschow, 2010; Lafferty & Hovden, 2003; Nilsson & Persson, 2003; Persson, 2009). Both literatures share a concern for policy learning and coordination, and the advanced set of measures the environmental policy integration literature has developed in assessing the state of integration as a dependent variable can be used to examine the policy outcomes of a range of network properties. Environmental policy integration, for its part, would benefit from the enriched understanding of collaborative network dynamics within public administration.

Footnotes

Appendix

Table A3.

Attitudinal Control Variables.

Measure	Question answered on a 5-point Likert-type agreement scale (scaled so 1 = strongly agree and 5 = strongly disagree)
Collab	Working with other government agencies is an important part of my role.^a
SciCrit	Scientific information is critical to my role.^a
InfoQual	The information I receive from other agencies falls short of my needs.
	I am very satisfied with the scientific information available to me to perform my role.^a

To facilitate interpretation, the scaling directionality for these items was reversed so that 5 equaled the most positive perception/outcome and 1 equaled the most negative perception/outcome.

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: This work was supported by the Social Science and Humanities Research Council of Canada (grant number 430–2011-0644) and the IBM Center for the Business of Government, Washington DC.

Notes

Author Biographies

Owen Temby is an assistant professor in the Department of Political Science at The University of Texas Rio Grande Valley. His current research focuses on air pollution politics and transboundary ecosystem governance. His recent work has appeared in Review of Policy Research, Environment, Planning Perspectives, Journal of Policy History, and Journal of Great Lakes Research.

Jean Sandall is manager of customer design, research and innovation at the Department of Transport and Main Roads, Queensland. She contributed to this program of research while undertaking a postdoctoral fellowship in the Department of Natural Resource Sciences at McGill University. Her research interest lies in understanding and harvesting the drivers of innovation in government.

Ray Cooksey is emeritus professor in the UNE Business School at the University of New England and is a fellow of the Australian and New Zealand Academy of Management. His research areas include managerial decision making, complexity theory, statistics, and research methods. He is the author of Judgment Analysis: Theory, Methods & Applications (Academic Press), several other books on social research methods and statistics, and more than 65 refereed publications.

Gordon M. Hickey is an associate professor and William Dawson Scholar in the Department of Natural Resource Sciences at McGill University, Canada. His international research applies mixed-method techniques to explore the institutional processes affecting sustainable natural resource-related policy making and implementation, with a particular focus on integrating scientific knowledge for innovation. His work has appeared in journals such as Social Studies of Science, Science and Public Policy, Ecological Economics, Regional Environmental Change, Land Use Policy, and World Development.

References

Abram

F. Y.

Mahaney

H. A.

Linhorst

D. M.

Toben

Flowers

(2005). Interorganizational collaboratives for children of prisoners: One that succeeds, another that struggled. Journal of Community Practice, 13(1), 31-47.

Agranoff

(2006). Inside collaborative networks: Ten lessons for public managers. Public Administration Review, 66(Suppl. 1), 56-65.

Agranoff

(2007). Managing within networks: Adding value to public organizations. Washington, DC: Georgetown University Press.

Agranoff

McGuire

(2001). Big questions in public network management research. Journal of Public Administration Research and Theory, 11, 295-326.

Agranoff

McGuire

(2003). Collaborative public management new strategies for local governments American governance and public policy. Washington, DC: Georgetown University Press.

Allen

C. R.

Gunderson

L. H.

(2011). Pathology and failure in the design and implementation of adaptive management. Journal of Environmental Management, 92, 1379-1384.

Arnott

D. C.

(2007). Research on trust: A bibliography and brief bibliometric analysis of the special issue submissions. European Journal of Marketing, 41, 1203-1240.

Aucoin

(1995). The new public management: Canada in comparative perspective. Montreal, Quebec, Canada: Institute for Research on Public Policy.

Babbie

(2012). The practice of social science research (13th ed.). Belmont, CA: Wadsworth.

10.

Bailey

C. A.

(2007). A guide to qualitative field research. Thousand Oaks, CA: Pine Forge Press.

11.

Bardach

(2001). Developmental dynamics: Interagency collaboration as an emergent phenomenon. Journal of Public Administration Research and Theory, 11, 149-164.

12.

Barnett

M. N.

Finnemore

(1999). The politics, power, and pathologies of international organizations. International Organization, 53, 699-732.

13.

Berardo

(2009). Processing complexity in networks: A study of informal collaboration and its effect on organizational success. Policy Studies Journal, 37, 521-539.

14.

Börzel

T. A.

(1998). Organizing Babylon—On the different conceptions of policy networks. Public Administration, 76, 253-273. doi:10.1111/1467-9299.00100

15.

Brass

D. J.

Galaskiewicz

Greve

H. R.

Tsai

(2004). Taking stock of networks and organizations: A multilevel perspective. Academy of Management Journal, 47, 795-817.

16.

Bryner

G. C.

Duffy

R. J.

(2012). Integrating climate, energy, and air pollution policies. Cambridge: MIT Press.

17.

Burch

(2010). Transforming barriers into enablers of action on climate change: Insights from three municipal case studies in British Columbia, Canada. Global Environmental Change, 20, 287-297.

18.

Chiles

T. H.

McMackin

J. F.

(1996). Integrating variable risk preferences, trust, and transaction cost economics. Academy of Management Review, 21, 73-99.

19.

Cohen

West

Aiken

(2003). Applied multiple regression/correlation analysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Lawrence Erlbaum.

20.

Coleman

J. S.

(1988). Social capital in the creation of human capital. American Journal of Sociology, 94(Suppl.), S95-S120.

21.

Craik

Studer

VanNijnatten

(Eds.). (2013). Climate change policy in North America: Designing integration in a regional system. Toronto, Ontario, Canada: University of Toronto Press.

22.

Crozier

(1964). The bureacratic phenomenon. Chicago, IL: University of Chicago Press.

23.

Cummings

L. L.

Bromiley

(1996). The organizational trust inventory (OTI). In Kramer

R. M.

Tyler

T. R.

(Eds.), Trust in organizations: Frontiers of theory and research (pp. 302-330). Thousand Oaks, CA: Sage.

24.

Delgado-Márquez

B. L.

Hurtado-Torres

Aragón-Correa

J. A.

(2012). The dynamic nature of trust transfer: Measurement and the influence of reciprocity. Decision Support Systems, 54, 226-234.

25.

Edelenbos

Klijn

E.-H.

(2007). Trust in complex decision-making networks: A theoretical and empirical exploration. Administration & Society, 39, 25-50.

26.

Ferrin

D. L.

Dirks

K. T.

Shah

P. P.

(2006). Direct and indirect effects of third-party relationships on interpersonal trust. Journal of Applied Psychology, 91, 870-883.

27.

Field

(2013). Discovering statistics using IBM SPSS statistics (4th ed.). Thousand Oaks, CA: Sage.

28.

Gerring

McDermott

(2007). An experimental template for case study research. American Journal of Political Science, 51, 688-701.

29.

Gibbons

J. A.

Sherwood

R. D.

(1985). Repeated measures/randomized blocks ANOVA through the use of criterion-scaled regression analysis. Educational and Psychological Measurement, 45, 711-724.

30.

Healy

R. G.

VanNijnatten

López-Vallejo

(2014). Environmental policy in North America: Approaches, capacity, and the management of transboundary issues. Toronto, Ontario, Canada: University of Toronto Press.

31.

Heckscher

(1994). Defining the post-bureaucratic type. In Heckscher

C. C.

Donnellan

A. M.

(Eds.), The post-bureaucratic organization: New perspectives on organizational change (pp. 14-62). Thousand Oaks, CA: Sage.

32.

Henry

A. D.

Lubell

McCoy

(2011). Belief systems and social capital as drivers of policy network structure: The case of California regional planning. Journal of Public Administration Research and Theory, 21, 419-444.

33.

Holling

C. S.

Meffe

G. K.

(1996). Command and control and the pathology of natural resource management. Conservation Biology, 10, 328-337.

34.

Imperial

M. T.

(2005). Using collaboration as a governance strategy: Lessons from six watershed management programs. Administration & Society, 37, 281-320.

35.

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.

36.

Isett

K. R.

Provan

K. G.

(2005). The evolution of dyadic interorganizational relationships in a network of publicly funded nonprofit agencies. Journal of Public Administration Research and Theory, 15, 149-165.

37.

Jordan

Lenschow

(2010). Environmental policy integration: A state of the art review. Environmental Policy and Governance, 20, 147-158.

38.

Klenk

N. L.

Hickey

G. M.

MacLellan

J. I.

(2010). Evaluating the social capital accrued in large research networks: The case of the Sustainable Forest Management Network (1995-2009). Social Studies of Science, 40, 931-960.

39.

Klijn

E.-H.

Steijn

Edelenbos

(2010). The impact of network management on outcomes in governance networks. Public Administration, 88, 1063-1082.

40.

Lafferty

Hovden

(2003). Environmental policy integration: Towards an analytical framework. Environmental Politics, 12(3), 1-22.

41.

Lalor

B. M.

Hickey

G. M.

(2014). Strengthening the role of science in the environmental decision-making processes of executive government. Organization & Environment, 27, 161-180.

42.

Lambright

K. T.

Mischen

P. A.

Laramee

C. B.

(2010). Building trust in public and nonprofit networks: Personal dyadic, and third-party influences. American Review of Public Administration, 40, 64-82.

43.

Liebeskind

J. P.

Oliver

A. L.

Zucker

Brewer

(1996). Social networks, learning and flexibility: Sourcing scientific knowledge in new biotechnology firms. Organization Science, 7, 428-443.

44.

McGuire

Agranoff

(2011). The limitations of public management networks. Public Administration, 89, 265-284.

45.

Miles

M. B.

Huberman

A. M.

(1984). Qualitative data analysis: A sourcebook of new methods. Beverly Hills, CA: Sage.

46.

Mulroy

E. A.

Shay

(1998). Motivation and reward in nonprofit interorganizational collaboration in low-income neighborhoods. Administration in Social Work, 22(4), 1-17.

47.

New York City Panel on Climate Change. (2010). Climate change adaptation in New York City: Building a risk management response. Annals of the New York Academy of Sciences, 1196, 1-354.

48.

New York City Panel on Climate Change. (2013, June). Climate risk information 2013: Observations, climate change projections, and maps. Retrieved from http://www.nyc.gov/html/planyc2030/downloads/pdf/npcc_climate_risk_information_2013_report.pdf

49.

New York City Panel on Climate Change. (2015). Building the knowledge base for climate resiliency. Annals of the New York Academy of Sciences, 1336, 1-150.

50.

New York State 2100 Commission. (2013). Recommendations to improve the strength and resilience of the Empire State’s infrastructure. Retrieved from http://www.governor.ny.gov/sites/governor.ny.gov/files/archive/assets/documents/NYS2100.pdf

51.

New York State Climate Change Action Council. (2010, November). Climate change action plan interim report (Report No. 11-9-10). Retrieved from http://www.dec.ny.gov/docs/administration_pdf/irpart1.pdf

52.

New York State Sea Level Rise Task Force. (2011). Report to the legislature. Retrieved from http://www.dec.ny.gov/docs/administration_pdf/slrtffinalrep.pdf

53.

Nilsson

M. N.

Persson

A. S.

(2003). Framework for analysing environmental policy integration. Journal of Environmental Policy & Planning, 5, 333-359.

54.

Nooteboom

Berger

Noorderhaven

N. G.

(1997). Effects of trust and governance on relational risk. Academy of Management Journal, 40, 308-338.

55.

O’Toole

L. J.

(1995). Rational choice and policy implementation: Implications for interorganizational network management. American Review of Public Administration, 25, 43-57.

56.

Pedhazur

E. J.

(1977). Coding subjects in repeated measures designs. Psychological Bulletin, 84, 298-305.

57.

Persson

(2009). Different perspectives on EPI. In Nilsson

Eckerberg

(Eds.), Environmental policy integration in practice: Shaping institutions for learning (pp. 25-47). Sterling, VA: Earthscan.

58.

Provan

K. G.

Beagles

J. E.

Mercken

Leischow

S. J.

(2013). Awareness of evidence-based practices by organizations in a publicly funded smoking cessation network. Journal of Public Administration Research and Theory, 23, 133-153.

59.

Provan

K. G.

Kenis

(2008). Modes of network governance: Structure, management, and effectiveness. Journal of Public Administration Research and Theory, 18, 229-252.

60.

Provan

K. G.

Milward

H. B.

(1995). A preliminary theory of interorganizational network effectiveness: A comparative study of four community mental health systems. Administrative Science Quarterly, 40, 1-33.

61.

Provan

K. G.

Milward

H. B.

(2001). Do networks really work? A framework for evaluating public-sector organizational networks. Public Administration Review, 61, 414-423.

62.

Rethemeyer

R. K.

Hatmaker

D. M.

(2008). Network management reconsidered: An inquiry into management of network structures in public sector service provision. Journal of Public Administration Research and Theory, 18, 617-646.

63.

Ring

P. S.

Van De Ven

A. H.

(1994). Developmental processes of cooperative interorganizational relationships. Academy of Management Review, 19, 90-118.

64.

Rosenzweig

Solecki

DeGaetano

O’Grady

Hassol

Grabhorn

(Eds.). (2011). Responding to climate change in New York State: The ClimAID integrated assessment for effective climate change adaptation (Report No. 11-18). Albany: New York State Energy Research and Development Authority. Retrieved from http://www.nyserda.ny.gov/climaid

65.

Sabatier

P. A.

Jenkins-Smith

H. C.

(1999). The advocacy coalition framework: An assessment. In Sabatier

P. A.

(Ed.), Theories of the policy process (pp. 117-166). Boulder, CO: Westview Press

66.

Selin

VanDeveer

S. D.

(2009). Climate leadership in northeast North America. In Selin

VanDeveer

S. D.

(Eds.), Changing climates in North American politics: Institutions, policymaking, and multilevel governance (pp. 111-136). Cambridge, MA: MIT Press.

67.

Seppänen

Blomqvist

Sundqvist

(2007). Measuring inter-organizational trust—A critical review of the empirical research in 1990–2003. Industrial Marketing Management, 36, 249-265.

68.

Shrestha

M. K.

(2013). Self-organizing network capital and the success of collaborative public programs. Journal of Public Administration Research and Theory, 23, 307-329.

69.

Starik

Kanashiro

(2013). Toward a theory of sustainability management: Uncovering and integrating the nearly obvious. Organization & Environment, 26, 7-30.

70.

Stern

M. J.

Coleman

K. J.

(2015). The multidimensionality of trust: Applications in collaborative natural resource management. Society & Natural Resources, 28, 117-132.

71.

Temby

Rastogi

Sandall

Cooksey

Hickey

G. M.

(2015). Interagency trust and communication in the transboundary governance of Pacific salmon fisheries. Review of Policy Research, 32, 79-99.

72.

Thomson

A. M.

Perry

J. L.

(2006). Collaboration processes: Inside the black box. Public Administration Review, 66(Suppl. 1), 20-32.

73.

Uiterkamp

A. J. M.

Vlek

(2007). Practice and outcomes of multidisciplinary research for environmental sustainability. Journal of Social Issues, 63, 175-197.

74.

VanNijnatten

Craik

(2013). Designing integration: The system of climate change governance in North America. In Craik

Studer

VanNijnatten

(Eds.), Climate change policy in North America: Designing integration in a regional system (pp. 5-34). Toronto, Ontario, Canada: University of Toronto Press.