Abstract
This study explores the circumstances under which certain collaborative tools are adopted, and whether some tools are typically used in combination with others. We share the view of other scholars that collaboration practice is ahead of scholarship. Accordingly, we ground our analysis and conclusions on the observations provided by a sample of public managers who participate actively in collaborations. Findings from interviews with managers about the use of collaborative tools in their jurisdictions demonstrate that certain tools are used together, and that collaborations can be understood along three dimensions—structure of the collaboration, shared governance arrangements, and commitment of both parties to the collaboration. For researchers, this finding provides a foundation to comprehend, compare, and analyze collaborations across myriad policy domains. For practitioners, this result illustrates that collaborative tools are not interchangeable and are typically employed in three coherent groupings. For researchers and practitioners, the findings dispute common assumptions that greater collaboration (i.e., employing more tools) is productive and suggest that the emphasis might be more usefully placed on selecting and using the appropriate and parsimonious combination of tools to generate public value.
Keywords
Public agencies use collaboration to respond to problems, deliver services, and advance governance (Stone, Crosby, & Bryson, 2013). Public managers use a wide range of “tools” as part of the collaborative process: They may share information, facilities, staff, or programming. They may work with their partners to formulate common goals or measurable outcomes and share decision-making authority or power. Sometimes partners rely on mutual trust, commitment, or contractual arrangements to facilitate or enable joint action.
Public managers select and use these tools in different ways to allow government, nonprofit, and for-profit agencies to work together to address public problems, deliver goods and services, and enhance governance in ways that cannot be accomplished by working alone. Collaborations can be designed to be relatively permanent and enduring (e.g., reconfigure a service delivery network) or temporary or ad hoc (e.g., complete a one-time project). Thus, collaborative tools can be relatively informal (e.g., sharing information, facilities, staff, etc.) and rely on shared social norms, or they can become institutionalized in formal contractual agreements (e.g., an interlocal agreement or contract; Agranoff & McGuire, 2003; Blair & Janousek, 2013; Imperial, 2005).
Our conception of “tools” differs from Salamon’s (2002) “policy tools” and, more recently, Scott and Thomas’s (2017) “collaborative governance tools.” Salamon (2002, pp. 8, 14) describes the increasingly collaborative nature of government action, the shift from “government” to “governance,” and the transition from competition to collaboration as defining features of cross-sectoral relationships. He argues that these changes precipitate the need for identifying the policy instruments or “tools of public action” that offer governmental and nongovernmental actors various options for addressing public problems, including regulation, contracting, grants, direct loans, loan guarantees, insurance, vouchers, and tax expenditures. Thus, Salamon (2002) shifts the “the unit of analysis in policy analysis and public administration from the public agency or the individual public program to the distinctive tools or instruments through which public purposes are pursued” (p. 9). In our view, Salamon (2002) presents general approaches to policy problems, rather than discrete tools that public managers use in collaboration—which constitute our primary interest in this study.
In their discussion of “collaborative tools,” Scott and Thomas (2017) also adopt a policy-oriented framework. Scott and Thomas (2017, p. 191) explore “why public managers choose to devote public resources to collaborative governance” and investigate how Salamon’s policy tools can be implemented through collaborative structures and processes. They describe collaborative tools as “methods for initiating and supporting inter-organizational collaboration. Managers use participation incentives, formal agreements, resource sharing, deliberative forums, and other means to shape and incentivize collaborative actions” (Scott & Thomas, 2017, p. 193; emphasis added). Scott and Thomas’s (2017, p. 192) conceptualization of “tool” helps explain the choice made by public managers to use the “collaborative toolbox” over other means of solving public problems. Whereas Scott and Thomas (2017) and others (e.g., Imperial, Prentice, & Brudney, 2018) explore from a policy orientation why public managers choose to collaborate, in this article, we seek to uncover from a management orientation how managers work together when collaboration is the chosen policy approach.
Thus, we examine the managerial tools that are the building blocks of joint action, and our central research question explores whether some tools are typically employed in conjunction with other tools to “get things done.” Given the various factors influencing the choice of whether to collaborate (Scott & Thomas, 2017) and the multitude of public purposes for which collaboration is employed, good reason exists to suggest that the practice of collaboration is a multidimensional phenomenon (Agranoff & McGuire, 1998, 2003; Bowman & Parsons, 2013; Imperial, 2005; Thomson & Perry, 2006; Thomson, Perry, & Miller, 2009). The conceptual and rhetorical argument made by previous researchers that collaboration is multidimensional motivates our search for those underlying dimensions. In this article, we examine empirically whether collaboration is a multidimensional practice by assessing the extent to which fundamental tools of collaboration are used (or not used) in combination.
Despite sustained attention in the literature, scholars are stymied in their efforts to elucidate collaborative processes and to offer managerial prescriptions generalizable to various contexts and policy areas. One explanation may be the multidimensional nature of collaboration, which gives rise to a complex set of interactions and choices among governmental and nongovernmental actors in different institutional settings. As a result, Bryson, Crosby, and Stone (2016) conclude, “collaboration practice is generally ahead of collaboration scholarship” (p. 658). Similarly, McGuire (2002) contends, “The practice of managing across governments and organizations outpaces empirical description and theoretical explanation” (p. 599). Moreover, Bryson et al. (2016) suggest that “because cross-sector collaborations are so complex and dynamic and operate in such diverse contexts, it is unlikely that research-based recipes can be produced” (p. 658).
What we do know about collaboration is embedded in research oriented around particular policy domains or patterns of collaboration. For example, much scholarship focuses on specific policy settings, such as emergency management (e.g., Bowman & Parsons, 2013; Kapucu, 2006; McGuire & Silvia, 2010), economic development (e.g., Agranoff & McGuire, 1998, 2003; Feiock, Steinhacker, & Park, 2009; Hawkins, 2011), environmental protection (e.g., Imperial, 2005; Imperial & Kauneckis, 2003), and health (e.g., Mullin & Daley, 2009). In addition, researchers have studied particular collaborative governance arrangements, such as interlocal agreements (e.g., Delabbio & Zeemering, 2013; Thurmaier & Wood, 2002), joint ventures (e.g., Feiock et al., 2009; Hawkins, 2011), and partnerships (e.g., Hilvert & Swindell, 2013; Shaw, 2003). Over the last two decades, scholars have offered varying advice grounded in data from collaboration that occurred for various purposes in different contextual settings. Consequently, the field has accumulated knowledge about the use of tools in selected policy domains and contexts, yet generalizable information and prescription are lacking. This gap is problematic because public managers, particularly at the local level, work across a wide range of policy settings in which the nature of those activities can vary greatly (e.g., Agranoff & McGuire, 1998, 2003).
Our study addresses this gap by focusing on public managers who work across policy settings and examines the different tools they use when collaborating. Recognizing the multidimensional and contextual nature of collaboration, we do not expect that a single theory will explain all aspects of the process (Menzel, 1987). Nevertheless, we strive for a generalizable empirically grounded explanation. We do not focus on particular policy areas or collaborative arrangements but allow a broader determination of collaborative processes to be made by the practitioners themselves. Taking guidance from the scholars who contend collaboration practice is ahead of scholarship, we adopt an exploratory approach that allows practice to guide and inform our inquiry by examining managers’ reports of their use of a variety of collaborative tools across diverse collaborations. Consequently, we develop theory grounded in the practice of collaboration and let the experience of public managers guide our inquiry.
The result is a multidimensional framework of collaboration emanating from public managers’ use of different collaboration tools. The empirical basis of our study consists of semi-structured telephone interviews conducted with more than 80% of the county managers in a growing southeastern state, who are engaged actively in collaborations with nonprofit organizations, for-profit firms, and other government agencies for the delivery of services. In the interview, the county managers characterized their collaborations with other organizations using items that reflect the collaborative tools found in the literature. Based on dimensional analysis of their observations, the findings indicate that government collaborations with other organizations are classified along three dimensions: structure of the collaboration, shared governance arrangements, and commitment of both parties to the collaboration. In addition to identifying empirically derived dimensions along which collaborations can be constructed and understood, we find that these dimensions are related to various managerial and contextual factors, as well as to collaborative performance. Our findings provide guidance not only to public managers regarding when they may want to consider employing different tools of the “collaborative toolbox” but also to researchers interested in describing and explaining the multidimensional nature of government collaborations.
Defining Collaboration
A prominent obstacle to theory building on collaboration is the lack of a common definition (Prentice & Brudney, 2016). Two competing worldviews contribute to the definitional dissonance. The first sees collaboration in the broadest terms to capture the rich variety of interorganizational relationships, ranging from relatively simple activities associated with communication, cooperation, and coordination to highly complex shared governance arrangements (e.g., Imperial, 2005). The other perspective conceives of collaboration in somewhat narrower terms as requiring a higher level of collective action than simpler forms of cooperation (e.g., Foster & Meinhard, 2002; Thomson & Perry, 2006). Some scholars exclude highly complex forms of collaboration such as contracting by arguing that they are best studied as distinct phenomena (e.g., Donahue & Zeckhauser, 2011; Hilvert & Swindell, 2013; Lee & Hannah-Spurlock, 2015), while others ignore the relatively simple ways that organizations collaborate (e.g., information sharing) to get things done (e.g., Foster & Meinhard, 2002). Prior research demonstrates that public managers typically make no such distinctions (Prentice & Brudney, 2016).
We follow Phillips, Lawrence, and Hardy (2000) and others who suggest defining collaboration broadly to capture the full range of activities and interorganizational relationships undertaken by organizations. Like Bardach (1998), who builds on Moore (1996), we define collaboration as any joint activity by two or more organizations intended to create public value by working together rather than separately. This definition is sufficiently inclusive to encompass the entire range of collaborative tools and distinguishes collaboration from markets or hierarchical control mechanisms (Lawrence, Hardy, & Phillips, 2002; Powell, 1990). According to this conception, collaboration is the product of an interactive process involving a set of autonomous, purposive actors who use shared rules, norms, or organizational structures to act or make collective decisions and work together (Imperial, 2005; Wood & Gray, 1991). It includes formal and informal arrangements as well as ad hoc interactions and more enduring joint activities. Importantly, this broad conception is consistent with the way managers in our study defined collaboration (Prentice & Brudney, 2016) and maintains the broad focus we incorporate in framing the results.
When government organizations are involved, these joint activities aim to generate public value. Collaboration scholars (e.g., Bardach, 1998; Delabbio & Zeemering, 2013; Dyer & Singh, 1998; Feiock et al., 2009; Huxham, 1996; Klijn & Teisman, 2005; Zaheer & Venkatraman, 1995) conceptualize public value in many ways such as collaborative advantage, relational rents, reduced transaction costs, and surplus value. Although this area of scholarship needs further development, scholars appear to coalesce around the argument that collaboration in the public sector should occur only when it generates some form of value that exceeds direct bureaucratic action (e.g., improved efficiency or performance, reduced cost of service delivery, etc.).
The Collaboration Paradox
Research shows that cross-sector collaboration is complex, synergies are highly contingent, and success depends on many factors (Gazley, 2017; Vangen, 2017). The pursuit of increasingly esoteric paradigms for explaining collaborative activity produces paradoxical recommendations that complicate the practice of collaboration (Brudney, Prentice, & Harris, 2018; Connelly, Zhang, & Faerman, 2008; Ospina & Saz-Carranza, 2010). As a result, scholars are still largely unable to use theory to inform the practice of collaborative governance.
For example, public managers are encouraged to pursue collaborations with partners who share the same goals and culture. Bolstered by heightened interconnectivity and mutual understanding, collaborations among homogeneous partners are found to reach agreement more easily, establish trust more readily, and work more smoothly. Collaborations absent shared goals and culture are apt to experience incompatible working practices, interpersonal discord, and collaborative inertia (Vangen, 2017). Yet, managers are likewise encouraged to pursue collaborations with partners who possess unique skills that complement, rather than duplicate, their capabilities. Collaborations among similar partners may breed competition, and “differences between organizations—including their expertise, assets, know-how, priorities, cultures and values—constitute unique resources that, when brought together, create the potential for synergies and collaborative advantage” (Vangen, 2017, p. 2). In addition, partnerships lacking diversity may produce overly facile approaches insufficient to address complex (wicked) problems (Bryson, Crosby, & Stone, 2006). Thus, the calculus managers face in developing collaborations is complicated, and researchers’ advice often strikes practitioners as equivocal or paradoxical (Brudney et al., 2018; O’Leary & Vij, 2012).
Vangen (2017) characterizes collaborations as “webs of overlapping, dynamic hierarchies and systems that comprise competing designs and processes that are necessary to achieve desired outcomes” (p. 3). The perplexing nature of collaboration stems from at least two sources. First, what is necessary to build and sustain collaborative processes when they are initiated can be quite different from when the collaboration has matured. Sustaining collaborative interaction over longer periods likely requires different tools than short-lived efforts (Mendel & Brudney, 2018). Unfortunately, most research, including this study, tends to be cross-sectional in nature, and the life cycle aspect of collaborative processes has received much less attention (Bowman & Parsons, 2013; Imperial, Johnston, Pruett-Jones, Leong, & Thomsen, 2016). Second, the decision to collaborate is intertwined with the related choice of determining how best to work with the chosen partners, that is, selecting the right tools (Kwon & Feiock, 2010), which likely depends on the institutional setting in which the collaboration occurs (Thomson & Perry, 2006). Thus, any advice forthcoming must be highly contingent. To make this advice more practical in this research, we present a dimensional approach to collaboration, which we elaborate below.
A Dimensional Approach to Collaboration
By contrast to scholars who place collaboration along a continuum, arguing that higher intensity, integrative collaborations yield the greatest value creation (e.g., Austin, 2000; Austin & Seitanidi, 2012; Bailey & Koney, 2000; Bryson et al., 2006; Rondinelli & London, 2003), we agree with Thomson and Perry (2006) that collaboration is a multidimensional construct. Based on a review of the literature and previous theories and definitions of collaboration, Thomson and Perry (2006) developed a model of collaboration consisting of five dimensions: two are structural (governing and administering), two are dimensions of social capital (mutuality and norms), and one is an agency dimension (organizational autonomy). Using survey data collected from directors of a large national service program incorporating 56 indicators of collaboration, Thomson et al. (2009) subsequently refined and validated this multidimensional model of collaboration.
Whereas Thomson et al. (2009) employed confirmatory factor analysis to test Thomson and Perry’s (2006) theoretical model/dimensions of collaboration, in this research, we pursue a more inductive, grounded theory approach. We start with a parsimonious list of indicators (i.e., tools) and use principal components analysis (PCA) to uncover the latent collaborative dimensions as reflected in the responses of our public administrator sample. Verifying whether collaboration practice is indeed multidimensional offers the basis for exploring more fully prescriptions regarding how best to collaborate.
Conceiving of collaboration as occurring across various dimensions, as opposed to unidimensional, allows for recognition of various continua of collaborative activity. From a unidimensional perspective, researchers are limited to advising practitioners regarding the extent (i.e., more or less) of collaboration. However, greater collaboration does not necessarily produce greater public value. To the contrary, doing more (i.e., using more tools than is necessary to get the job done) can increase transaction costs (e.g., Ostrom, Schroeder, & Wynne, 1993) and reduce the public value that would otherwise have been created. From a multidimensional perspective, researchers can acknowledge the different forms of collaborative practice (i.e., different combinations of collaborative tools) and propose using the combination of tools that is appropriate to a given context. Consequently, this study explores whether some collaborative tools are used in combination with others and the circumstances under which certain tools are adopted and others are ignored.
The Collaborative Toolbox
In the absence of a well-defined theory, we employed a quantitative grounded theory methodology first to identify the tools in the collaborative toolbox, and then to distill interview data gathered from practitioners to guide our understanding of the use of these tools (Glaser & Strauss, 1967; Strauss & Corbin, 1990). Even a cursory review of the vast collaboration literature reveals myriad ways that officials in governmental and nongovernmental organizations interact in collaborative processes. Given the exploratory nature of this study and our desire to develop theory grounded in these data, our approach to identifying the tools for analysis was similar to developing an initial coding list for conducting qualitative research. Exploring literature from different policy domains (e.g., Agranoff & McGuire, 1998, 2003; Bardach & Lesser, 1996; Blair & Janousek, 2013; Bowman & Parsons, 2013; Bryson et al., 2006; Donahue & Zeckhauser, 2011; Gray, 2000; Imperial, 2005; Lee & Hannah-Spurlock, 2015; McGuire & Silvia, 2010; Scott & Thomas, 2017; Thomson et al., 2009), we examined selected studies that emphasized or identified different ways organizations work together. Rather than develop an exhaustive list of potential tools, or exploring variations within a tool, our approach was to ensure that each tool was conceptually distinct and described in language familiar to public managers. Eliminating overlap across the tools was important methodologically to reduce potential collinearity, thereby facilitating our use of PCA to identify dimensional properties.
This process yielded the parsimonious listing of 11 collaborative tools enumerated in Table 1: sharing facilities, staffing, or programming; developing mutual or shared goals, trust, or commitment in the collaboration; the willingness to share information, decision making, or power; the existence of measurable performance outcomes; and the use of a contract to govern collaborative relationships. Although this list may not represent an exhaustive accounting of possible collaborative tools, all of these tools appear with regularity in the literature.
Common Collaboration Tools.
Our expectation is that public managers select different combinations of tools depending on several factors—what they wish to accomplish, with whom they decide to work, contextual factors, and so on. Developing an improved understanding of the multidimensional nature of the collaborative toolbox and the relationship to the contextual environment in which these processes occur is the central objective of this study. Our analysis illuminates whether certain constellations or groupings of tools coalesce or occur together and, thus, may provide the dimensions along which public-sector collaborations might be organized and better understood.
Method
This study uses a quantitative grounded theory methodology intended to move interactively between theory development and data analysis. This inductive process implicitly acknowledges the limitations of current collaboration scholarship for building deductive hypotheses and places an emphasis on learning from practice. We began by reviewing the scholarship on collaboration to identify tools commonly used or described in the literature (Table 1). Next, we conducted in-depth interviews with respondents who actively engage in interorganizational collaboration and asked them to indicate the extent to which the tools we identified occurred in their collaborations. Based on the responses, we then used PCA to assess possible dimensions of government collaboration at the ground level. Using relevant scholarship as a guide, we identified and interpreted the emerging collaboration dimensions inductively. Finally, we analyzed these findings further by testing relationships between the collaboration dimensions and a series of managerial, contextual, and effectiveness variables constructed from the practitioner interviews and archival data sources.
The unit of analysis for our empirical inquiry is counties in the state of North Carolina. We focus on counties for several reasons. County governments provide an ideal laboratory for examining cooperative activities (Parks, 1991), and these entities are at the forefront of new opportunities and challenges associated with collaboration (Delabbio & Zeemering, 2013). County governments are also active players in local governance in many policy domains (Smith, 2007), and therefore their leadership is often instrumental in advancing collaborations. Because counties are not equally positioned to engage in collaboration, we expected variation across our dependent variables (Abernathy, 2012; Delabbio & Zeemering, 2013).
According to Bowman and Parsons (2013), “Although single-state studies have limited generalizability, North Carolina counties offer much variability in terms of demographics and social factors, thereby lessening the limitation (Hoyman & McCall, 2010) and providing a defensible test bed for exploration . . .” (p. 65). North Carolina is the ninth most populous state in the United States, with high in-migration and a mix of large urban centers and rural communities. Across many indicators, North Carolina is a large, complex, growing state that faces pressures for service delivery similar to those encountered by many other states and, thus, public administrators are likely to engage actively in collaboration with other sectors of the economy. Consequently, North Carolina has proven to be a useful state for exploring collaboration at the county level (e.g., Abernathy, 2012; Bowman & Parsons, 2013).
Data Collection
We collected the primary data for our analysis via telephone interviews with county managers in North Carolina in 2014. These respondents are the chief administrative officers of their jurisdictions, with budgets well into the millions of dollars and primary service delivery responsibilities for the county. County managers oversee general county government activities, public safety, economic and physical development, human services delivery, cultural and recreational programming, education, and debt servicing. North Carolina county managers on average oversee annual expenditures in excess of US$131 million and are responsible for serving 98,000 citizens on average (North Carolina Department of the State Treasurer; U.S. Census Bureau).
We obtained Human Services (institutional review board [IRB]) approval to contact all 100 county governments in North Carolina a maximum of 5 times in a mixed-methods approach comprising telephone and email communication. This effort resulted in the completion of 84 telephone interviews. Our sample consists of 74 county managers, six interim county managers, and four assistant county managers for a response rate of 84%. For convenience, we refer to the sample as “county managers.” Our interviews with the county managers confirmed that they are knowledgeable and active in collaborations for the delivery of public services in partnership with nonprofit, for-profit, and other government organizations.
Operationalization
In our semi-structured telephone interviews, we asked respondents about their professional backgrounds and positions, their conceptions of collaboration, and several questions regarding their county’s collaborative efforts. Given this study’s inductive and grounded approach, respondents were not cued with a uniform definition of collaboration. Rather, we sought to learn whether county managers would offer a clearer conception of collaboration than the disparate definitions of collaboration offered in the academic literature (see above). The following question asked respondents to define collaboration: “People talk a lot about county government collaboration with other organizations, such as nonprofit organizations, private businesses, and other government agencies. How do you define collaboration?” Responses were transcribed verbatim and read back to the respondent. The interviewer then probed for elaboration, recorded all respondent comments, and confirmed the full description.
Analysis of these responses demonstrates the broad scope and inclusive character managers ascribe to the practices and purposes of collaboration (Prentice & Brudney, 2016). In their definitions of collaboration, county managers typically referred to collaboration as a means for getting things done (via formal and/or informal processes), and described many tools of collaboration (e.g., sharing goals/mission, sharing resources, sharing decision making, contracting) and reasons for collaborating (e.g., increase public benefit, efficiency, mutual benefit, range of services, access to expertise; Prentice & Brudney, 2016). The broad and inclusive definitions presented by respondents informed our equally comprehensive conception, delineated in the “Defining Collaboration” section above.
Collaboration tools
Immediately following the question asking county managers to define collaboration, we asked them to estimate the proportion of their county’s collaborations that used one or more of the tools listed in Table 1: “There is no single, correct definition of collaboration. But most definitions of government collaboration with other organizations involve one or more of the following characteristics. About what proportion of your county’s collaborations have the following?” For each of the tools, the county managers chose from the following options: “none,” “one third or less,” “one third to two thirds,” “two thirds or more,” or “all.” Responses were coded 0 through 4, respectively, for each tool.
Managerial variables
Esteve, Boyne, Sierra, and Ysa (2013) show that the personal characteristics of top managers may affect organizational strategy and influence interorganizational collaboration. In addition, McGuire and Silvia (2010) find that past experiences, future concerns, and educational attainment influence the frequency with which local governments collaborate with other public organizations. Accordingly, county managers were asked about their tenure in the current position (tenure), total number of years in government employment (government work), work experience in the for-profit and nonprofit sectors (for-profit work and nonprofit work, respectively), level of education (education), major course of study (education functional track), and preferred collaboration partner (preferred partner).
County government variables
Previous county-level studies of collaboration have used a variety of variables to capture contextual factors thought to motivate collaboration (e.g., Bowman & Parsons, 2013; Delabbio & Zeemering, 2013; Feiock et al., 2009; Kwon & Feiock, 2010; McGuire & Silvia, 2010; Mullin & Daley, 2009). We tested several variables to capture the wealth and spending pattern of each county government, including data from the North Carolina Department of State Treasurer (www.nctreasurer.com) on county government’s total expenditures, education expenditures, debt service expenditures, human services expenditures, general government expenditures, public safety expenditures, and other expenditures. 1
The remaining county government variables relate to the scope of current collaborations and emanate from data collected in the interviews. We asked county managers to estimate the number of collaborations in which the county engages (collaboration amount), the percent of their collaborations that are with partners from each sector (nonprofit partner, for-profit partner, and government partner—where the sum totals 100%), and the program area with the most collaborations (human services or other). 2
County demographic variables
Previous county-level collaboration studies incorporate a series of demographic variables to represent contextual factors that may influence collaborative processes (e.g., Bowman & Parsons, 2013; Delabbio & Zeemering, 2013; Feiock et al., 2009; Kwon & Feiock, 2010; McGuire & Silvia, 2010; Mullin & Daley, 2009). We obtained the following county demographic data: population, population density, population change, race, median household income, tax base, percent inhabitants below poverty, percent with bachelor’s degree, percent voted Obama, and metro status from the U.S. Census Bureau.
Collaboration effectiveness variables
Managers presumably choose to collaborate because the county realizes some benefit (actual or perceived) from doing so (e.g., Feiock et al., 2009; Imperial, 2005; McGuire & Silvia, 2010). Accordingly, the final set of variables allows us to examine the consequences of different dimensional configurations for perceived collaboration results. We asked county managers whether they measured the success of their collaborations (measure success). We also asked whether, compared with not having a partnership, their county’s collaborations had led to more, less, or about the same level of effectiveness across six performance measures (citizen benefits, red tape, efficiency, service delivery speed, access to expertise, and flexibility). The appendix presents descriptive statistics for all variables.
Findings
The findings are organized in three parts. First, we explore whether the common collaboration tools culled from scholarship (Table 1) converge into dimensions empirically. The results of the PCA presented in Table 2 demonstrate that the collaboration tools form three distinct (independent) dimensions. Second, we tried to understand how the dimensions of collaboration identified in the PCA are associated with managerial, county government, and county demographic variables. The bivariate correlation analysis presented in Table 3 indicates that collaborations take shape primarily based on managerial factors, that is, managerial characteristics, rather than aggregate county government or demographic characteristics. Finally, we explore the implications of different dimensional configurations for collaborative success.
Component Loadings for Collaboration Tools (N = 84).
Note. Extraction method: Principal components analysis. Rotation method: Varimax with Kaiser normalization. Bold Values indicate the component on which the variable loaded the highest.
Bivariate Correlations Between Collaboration Components and County Manager Characteristics, County Government Characteristics, and Collaboration Effectiveness (N = 84).
Note. Only statistically significant correlations are shown. Correlations between the collaboration components and the following characteristics did not achieve statistical significance: managerial (government work, for-profit work, education, education functional track), county government expenditures (total, education, debt service, human services, general, public safety, and other), county demographic (population size, population density, population change, metro, percent White, median household income, percent below poverty, county education, Obama, tax base).
p < .10. **p < .05. ***p < .01 (two-tailed).
PCA
We used dimension identification and reduction techniques to examine whether some or all of the 11 collaboration tools converge into fewer underlying dimensions. For this purpose, we employed PCA, using principal component extraction with Varimax rotation 3 of the 11 collaboration tools identified in Table 1. The analysis yielded a three-component solution with a simple, and highly interpretable, structure: Only three components have an eigenvalue greater than 1.0; each tool loads clearly (and uniquely) on one component; and no significant cross-loadings on the components emerge. As found in prior literature, these results provide strong support for the conclusion that collaboration is a multidimensional construct (Agranoff & McGuire, 1998, 2003; Bowman & Parsons, 2013; Imperial, 2005; Thomson & Perry, 2006; Thomson et al., 2009). Table 2 presents the results of the PCA.
Table 2 reveals that the collaboration tools “load” (correlate with) or cluster on the dimensions in theoretically meaningful ways. The first component, which we label structure, captures operational aspects of a collaboration where transaction costs can be relatively low. The emphasis of these tools is on getting things done by sharing resources or crafting agreements that ensure the provision of some good or service, that is, primarily on service provision, rather than on nurturing and maintaining an ongoing relationship with the collaborative partner.
The second component, which we label shared governance, captures the extent to which the partners are on an equal footing in the collaboration—that is, whether the partners have the same amount of power and decision-making authority. Arguably, from an institutional standpoint, these collaborative tools are the most complex: They take longer to develop and require much more complicated rules and norms to structure the joint processes needed to allow the collaborations to endure and produce over an extended period. Consequently, developing and maintaining these collaborations likely imposes the highest transaction costs, so that we would expect their use to be limited to situations where the partners perceive sufficient benefits to justify such investments.
The third component, which we label commitment, captures the extent to which the partners in the collaboration are dedicated to the joint enterprise and to one another. As with the shared governance dimension, the tools that load highly on the commitment dimension likely require a greater investment in the collaboration and a higher level of shared engagement by the partners to allow two or more organizations to work together to determine and advance a common or shared interest. This dimension implies a higher level of transaction costs to initiate and maintain the collaboration than is required for the structural dimension.
The key finding that collaboration falls along three dimensions demonstrates that managers rely on different combinations of tools to get things done, presumably to accomplish different managerial purposes and generate different types of public value. The choice of tools likely reflects the nature of the joint work that managers need to accomplish, their familiarity and preferences regarding different tools, the contextual setting, and the expected duration of the collaborative activity. It is conceivable that managers aim to use the combination of tools needed to get the job done to maximize the expected return from their investment in the collaborative process and avoid using tools unnecessarily to minimize transaction costs.
Correlation Analysis: Managerial, County Government, and County Demographic Variables
The results of the PCA demonstrate that collaboration is a multidimensional construct and raise the possibility that each dimension of collaboration relates differently to managerial and contextual factors. To explore the possibility that different conditions give rise to different forms of collaboration, we derived principal component scores from the dimensional analysis and performed correlational analysis between the scores and a wide range of managerial, county government, and county demographic variables that appear in the collaboration literature. 4
Previous county-level studies of collaboration suggest that managerial and contextual characteristics influence choices related to whether to collaborate and the form of those collaborations. To test these relationships, we examined bivariate correlations between each of the three collaborative dimension scores and the various managerial, county government, and county demographic variables. Table 3 presents those correlations that achieve statistical significance.
Analysis of the correlations between the principal component scores for the three dimensions and the managerial, county government, and county demographic characteristics suggests that collaboration is not the product of a deterministic process whereby certain environmental conditions necessitate how organizations collaborate. Rather, collaboration appears to be a predominantly situational and contextual practice. Among the many managerial, county government, and county demographic characteristics found in previous literature and analyzed here, only a handful find statistical significance with the component scores for the three dimensions. For the background variables, only county managers’ current and past work experience (tenure, nonprofit work experience, and years of experience) and preferred collaboration partner (preferred partner) are associated statistically with the underlying dimensions of the collaborative toolbox.
Our findings show that county managers who have been in the position longer tend to emphasize the structural components of collaboration (r = .18, p < .10), suggesting that with greater experience as the top governmental administrator in the county comes the realization that collaborations require organization, formalization, measurable performance outcomes, and resource allocations by both parties. Nonprofit work experience also appears significant: Whether county managers had experience working in the nonprofit sector (r = .22, p < .05) and, if so, the number of years they worked in the sector (r = .28, p < .01) are related significantly to the structure dimension. It is possible that firsthand experience in the nonprofit sector, where organizations enjoy flexibility but are plagued by resource constraints, gives these managers an appreciation of the value of pursuing collaborations that have an adequate structure. Correspondingly, county managers who expressed a preference for partnering with the nonprofit sector or the public sector rather than with the for-profit sector were more likely to report that their county’s collaborations ranked highly on structure (r = .20, p < .10).
The county government characteristics do not fare as well as the managerial characteristics with respect to substantiating statistical relationships with the dimensions of collaboration. Although it might seem reasonable to assume that variables such as how much a county government spends and where it directs its expenditures would influence how it engages in collaboration, the bivariate correlations shown in Table 3 belie such expectations. These findings are consistent with other county-level studies of collaboration that find inconsistent statistical relationships between collaboration and county-level spending, education, income, and population variables (e.g., Bowman & Parsons, 2013; Delabbio & Zeemering, 2013; Feiock et al., 2009; Kwon & Feiock, 2010; McGuire & Silvia, 2010; Mullin & Daley, 2009). As we observed earlier, collaboration seems to be a situationally determined practice.
Nevertheless, we did find that the frequency with which county managers partner with organizations from particular sectors of the economy and the policy or program area in which they collaborate are related to the shared governance dimension. County governments that collaborate more with other government organizations are more likely to report having shared governance arrangements (r = .30, p < .01). Conversely, county governments that collaborate more with for-profit organizations are less likely to report having such arrangements (r = −.22, p < .05). Presumably, this finding reflects a preference for using contracting in the latter situations. Power and decision making are more readily shared in collaborations between governmental agencies, and less likely to be shared by government agencies that partner with for-profit firms. Because county governments are unlikely to cede power to other governmental units, relationships must be based on shared governance; by contrast, counties can use the structural tools such as contracting (with nonprofit organizations) and sharing resources to collaborate with nongovernment entities. In addition, county governments that report human services as the program area in which they have the most collaborations are less likely to report having shared governance arrangements (r = −.31, p < .01), again likely as a result of partnering with nonprofit organizations, which are prevalent in this policy domain.
Aggregate-level county demographic characteristics do not appear to be related statistically to the different dimensions of the collaborative toolbox. This finding likely occurs because collaboration is not a deterministic process (Scott & Thomas, 2017) but a strategic response used to solve problems or find ways to improve service delivery. We do not consider this result altogether surprising. In our judgment, collaboration does not emerge spontaneously due to demographic factors or community characteristics. Instead, collaboration emanates from interactions among public managers and potential partners who make independent and collective judgments about the advantages of partnering over working independently (Hilvert & Swindell, 2013; McGuire, 2002). Even when mandated by a funder or other governmental entity to collaborate, managers typically still have great discretion regarding the form and character these mandated relationships might take.
In sum, we sought to understand how managerial, county government, and county demographic factors might relate to the structure, shared governance, and commitment dimensions of county collaborations. Yet, few of these correlations attain statistical significance. Although we can classify the collaboration tools into meaningful dimensions according to the results of the PCA, the conditions that give rise to the different types of collaboration appear both situational and variable. We return to this issue in the discussion.
Collaboration Effectiveness
The bivariate correlations presented in Table 3 indicate that the structure and shared governance dimensions are related statistically to certain managerial and county government characteristics. It is the commitment dimension, however, that is related most strongly to the effectiveness of the collaboration.
The correlations in Table 3 show that counties with larger numbers of collaborations rating highly on the commitment dimension more frequently measure success of their collaborations (r = .20, p < .10), and report less red tape (r = −.19, p < .10), greater citizen benefits (r = .26, p < .05), greater efficiency (r = .36, p < .01), faster service delivery speed (r = .30, p < .01), greater access to expertise (r = .20, p < .10), and greater flexibility (r = .23, p < .05). By these results, achieving a high level of “jointness” in the collaboration—that is, achieving high willingness of the partners to share information, high commitment by top managers from each partner organization, high mutual trust, and shared goals—is integral to collaborative success, a finding consistent with much collaboration scholarship (e.g., Shaw, 2003).
By contrast, incorporating elements of structure or shared governance into collaborative relationships is not enough to ensure their effectiveness. Many of the tools associated with the structural dimension are largely oriented around the necessity to get things done, rather than the effectiveness measures that focus on ways to enhance collaborative governance. The tools associated with the shared governance dimension by necessity impose higher levels of transaction costs, which reduce some of these benefits of collaboration but presumably allow for other benefits that compensate for the additional costs.
Collaboration Complexity
The final statistical analysis explored how counties configure their collaborations across the three dimensions simultaneously and the implications of these arrangements for the effectiveness of collaboration. To create different collaboration configurations, we divided the principal component scores for each of the three dimensions at the median and assigned a code of “high” to component scores falling above the median and a code of “low” to component scores falling below the median. From these scores, we created an index of collaboration complexity based on the number of “high” scores registered across the structure, shared governance, and commitment dimensions. In our view, counties with three “high” ratings—that is, those emphasizing the structural, shared governance, and commitment dimensions of their collaborations—engage in more complex collaborations (10.7% of the counties), whereas those with three “low” ratings engage in less complex collaborations (8.3% of the counties). Falling between these poles, 44% of the counties have somewhat complex collaborations (with scores on one of the collaboration dimensions surpassing the median), and the remaining 36.9% of the counties rate moderately complex (with scores on two of the collaboration dimensions surpassing the median). The collaboration complexity index ranges on a scale from 0 (least complex) to 3 (most complex).
We used analysis of variance (ANOVA) techniques and examined whether collaboration complexity is related to collaboration effectiveness, as assessed by the variables tested earlier. In fact, counties with more of their collaborations rating higher in complexity reported achieving lower levels of efficiency in those collaborations (r = −.28, p < .01). In addition, county governments with most of their collaborations occurring in the human services domain report greater collaboration complexity (r = .23, p < .05), perhaps as a consequence of the intricacies of arranging, operating, and assessing collaborations in this policy area.
Although the collaboration effectiveness measures are strongly and consistently related to the commitment dimension of collaboration, these findings suggest that the benefits we tested are attenuated as collaborations increase in complexity. One possible explanation for this finding is that collaborations that place greater emphasis on shared commitment may be more likely to achieve higher collaborative success, but they may also be more inefficient, perhaps because achieving higher levels of commitment requires greater investment in collaborative processes. This explanation would help to explain Thomson et al.’s (2009) finding that it is rare to find collaborations scoring high on all five of their collaboration dimensions. It is also consistent with other research arguing that county managers avoid crafting collaborative governance arrangements that have excessively high transaction costs (Delabbio & Zeemering, 2013; Feiock et al., 2009).
Discussion
Much of the literature on collaboration concentrates on particular policy areas (e.g., education, environment), focuses on one type of collaborative arrangement or tool (e.g., contracting), and on particular collaboration partners (e.g., government–nonprofit partnerships). By contrast, in this research, the policy area of the collaboration, the tools that public managers might use, and their choice of partners (other governments, nonprofit agencies, or for-profit firms) are left broad, to allow public administrators to describe the nature of their collaborations more fully. We share the view of other scholars expressed earlier that collaboration practice is ahead of scholarship. Accordingly, we ground our empirical analysis and conclusions directly on the observations provided by a large sample of county managers who participate actively in collaborations in a growing southeastern state.
Based on the managers’ accounts of the use of the various collaborative tools, collaboration occurs along three dimensions—structure of the collaboration, shared governance arrangements, and commitment to the collaboration. For researchers, this finding provides a strong foundation to comprehend, compare, and analyze collaborations across myriad policy domains. Building on their own reports, for practitioners, this result indicates that the collaborative tools available to them can be distilled into three dimensions or coherent groups of tools that tend to cluster together.
The first dimension emanating from our analysis corresponds to the structural or operational dimension of collaboration; it encompasses such collaboration basics as sharing facilities, staff, and programming, the use of a contract, and specification of outcomes. These tools are analogous to the operational-level collaborative activities identified by Imperial (2005) in his study of watershed-based collaborations, in which local governments collaborated with other governmental and nongovernmental entities to share facilities, staff, funding, and equipment to undertake restoration, improvement, education, and monitoring projects. Intended to solve immediate problems, many of these collaborations were short-lived. However, when a large number of these projects needed to occur over a relatively long period, other tools were used to help support and guide these projects (Imperial, 2005). Thus, the tools in the structure dimension can be combined with tools in the shared governance or commitment dimensions to accomplish the joint work that needs to be performed over a protracted period of collaboration.
The second dimension of collaboration emerging from our analysis pertains to shared governance and encompasses sharing power and decision making over the collaboration. These arrangements are highly sensitive, involving substantial transaction costs, which can entail the creation of new entities to govern and manage the collaboration. For example, Imperial’s (2005) watershed study describes cases in which the partners formed partnerships that functioned as organizations in their own right with members sharing power and decision-making authority over the watershed’s governance.
The third dimension emanating from our analysis of the county mangers’ reports signals another way to move toward a more enduring collaboration arrangement, which we label “commitment.” This dimension taps the degree to which the partners are willing to invest in building a long-lasting relationship through, for example, sharing information, coming to trust one another, and reaching agreement on goals. Public managers who use these tools take on additional transaction costs with the aspiration of achieving greater collaboration effectiveness. Imperial’s (2005) watershed study provides numerous examples in which governmental and nongovernmental actors shared information and made commitments to follow a new set of shared goals/missions related to addressing watershed problems.
To understand the implications of the dimensional nature of the collaborative toolbox, we offer the metaphor of fixing a car. Car repairs come in many different forms and entail various levels of cost and complexity. A car can have assorted problems, including damage from a collision, faulty transmission, a flat tire, worn brake pads, and so on, and the repair for each problem involves some similar, but mostly different, combinations of tools and equipment. For example, a tire change requires a lug wrench, jack, and new tire, whereas an oil change requires a socket wrench, oil filter wrench, jack, oil drain pan, funnel, and new oil filter. Some repairs occur quickly and require few tools (e.g., oil or tire change), while other repairs take longer and require many (e.g., engine rebuild). In short, different combinations of tools are deployed for different purposes, and one ought not use more tools than is necessary to complete the job.
Whereas much research embraces the implicit or explicit assumption that collaboration falls along a unidimensional continuum and the oft-noted prescription that more (i.e., greater intensity) collaboration is better, our PCA shows that a multidimensional perspective is more fitting. Furthermore, the ANOVA findings demonstrate that greater collaboration (i.e., increased collaborative complexity) does not yield more benefits. These findings are intuitive in our car repair example: One does not need an oil filter wrench to change a tire and should therefore avoid the increased transaction costs of acquiring and utilizing unnecessary tools.
From the data available to us, we cannot discern the purposes for which different combinations of collaborative tools are used. However, our finding that collaboration is multidimensional and that some tools are often used in conjunction with certain others suggests that managers do not view the tools as interchangeable but use them selectively to address different problems. The tools used by the mechanic cluster into groups needed for different types of car repair, just as the collaborative tools cluster into different dimensions. We expect that public managers rely on the tools they need to get work done and to achieve the best return on their investment in collaborative processes.
When collaboration scholarship focuses on particular policy settings or on particular tools or types of partners, it may fail to develop a complete understanding of the full range of tools, and how they can be combined to solve problems in other policy domains (Hilvert & Swindell, 2013). This study tries to overcome that limitation by examining the accounts of how actors (e.g., county managers) collaborate across different policy domains with a full range of tools. Further research on the use of tools in different policy settings is important for theory building as well as practice. Table 4 provides an apt example. It shows how the dimensions of collaboration surface differently in studies of watersheds versus economic development.
Combining Tools Into Governance Strategies.
While this study contributes to our understanding of tools and their dimensional properties, additional inquiry is necessary to explore the full range of the collaborative toolbox and the situations in which different combinations of tools are more or less appropriate. What our findings do suggest is that public mangers will (and should) structure their collaborations differently depending on their goals for the relationship as well as the trade-offs encountered between transaction costs and results. Collaborations appear situationally determined and contingent on the factors at hand. Like Thomson and Perry (2006), we find that collaboration does not vary along a single continuum where greater investment may lead, at least hypothetically, to better outcomes or the creation of greater public value (e.g., Austin, 2000; Bailey & Koney, 2000). Rather, value generation is likely contingent on selecting the most appropriate and parsimonious combination of tools. Our findings intimate that county managers should not aim to develop collaborations that rank highly across all dimensions, as if such a configuration represented the ideal (cf., Thomson & Perry, 2006). To the contrary, our study suggests that more complex collaborations may suffer lapses in efficiency, likely as a result of higher transaction costs. It is facile to assume a “one best way” to practice collaboration, or that greater or more complex collaboration is somehow “better” (i.e., generates more public value) than less involved or complicated collaboration. To the contrary, our findings suggest that managers should employ the combination of tools needed to get the job done to maximize their expected return from their investment in the collaborative process (Imperial et al., 2016).
Our view is that public managers should engage in collaboration only when it creates public value that cannot be achieved by working alone (Bardach, 1998). As that determination is highly individual and situational, we would not expect to observe across counties many strong statistical relationships to government characteristics or contextual variables (see Table 3). As stated, collaboration is not a deterministic process driven by contextual conditions (Scott & Thomas, 2017, p. 194). Therefore, one would not expect that a set of demographic or financial characteristics would serve as drivers that require, or strongly suggest, the use of a particular collaboration tool (or set of tools). County managers have many choices with respect to working with other governmental and nongovernmental actors in collaborative relationships. Our analysis of their reports concerning the use of tools from the collaborative toolbox shows that application of certain tools tends to be associated with other tools in meaningful ways, which gives rise to and represents the structure, shared governance, and commitment dimensions of collaboration. These dimensions can help both researchers and practitioners form, understand, and compare collaborations across different policy areas, partners, and settings.
Conclusion
The findings of this study indicate that public managers use a range of tools when working with other public and nongovernmental organizations to get things done. The tools cluster along three dimensions, which we interpret as structure, shared governance, and commitment. These findings offer several implications for future research. First, rather than develop grand theories and complex frameworks for mapping collaboration processes, researchers need to understand the situational contingencies that give rise to managers’ decisions about which tools to combine for different purposes with different partners across different contextual settings. Researchers’ inability to uncover and comprehend these contingencies is the principal reason that collaboration scholarship lags practice (Bryson et al., 2016; McGuire, 2002). In our view, for theory to inform practice, it must derive from these situational roots. A deterministic approach that views collaboration as the by-product of a set of exogenous conditions in the environment (e.g., demographic or governmental characteristics) appears nonproductive. Although the contextual environment may affect the decision to collaborate, our findings suggest that the characteristics and preferences of public managers and their selection of partner (government, nonprofit, business) have greater influence over how they collaborate.
Second, the finding that more complex collaborations (i.e., those that use more tools) are not necessarily “better” than less complex collaborations creates the need for more nuanced prescriptions. The fact that the tools cluster into different dimensions suggests that the dimensions may have a connection to different types of public value that a manager seeks to achieve through partnership. Unfortunately, although scholars generally agree that collaboration can yield additional value beyond that which can be created by working alone, little agreement exists regarding how collaboration can be structured to generate greater value in particular settings and contextual environments. A better understanding of the public value concept and how it is linked to different combinations of tools would allow us to provide more useful advice to practitioners.
Although the collaborative toolbox offers a promising first step toward improving our understanding of the tools and how they are used in combination, we need to develop the construct more fully, apply it to other settings, and examine different groups of public managers. Moreover, we may have overlooked tools, such as sharing monetary resources and, perhaps, monitoring partners to the collaboration. In this initial inquiry, we were not able to explore variations in the tools (e.g., the different ways to share programming, facilities, decision making, or power), gradations (e.g., how much trust is needed), or continua (e.g., high vs. low power sharing). In addition, theory development would benefit from examination of the use of the tools over the collaboration life cycle (Imperial et al., 2016); some combinations of tools may be best suited to partnership formation, while others may prove ideal for sustaining long-term relationships.
Nevertheless, the strength of the toolbox approach and the resulting dimensional framework lies in its parsimony and flexibility: It distills public collaborations to the core tools associated with how government actors work with others to get things done. The approach can be applied in diverse policy settings to analyze relatively simple forms of collaboration over short periods, or highly complex forms of collaborative relationships over longer periods. The collaborative toolbox also relies on terminology that reflects the way practitioners interact and think about collaboration. Although it is desirable to use theory to inform practice, we believe a more productive approach with respect to collaboration is to develop theory grounded in the observed practice of public managers.
The findings of this study also allow us to offer advice to public managers considering collaboration as a strategy to address mutual problems. Much as Behn (2003) observes that different performance measures fulfill different purposes, our results suggest that different collaborative tools may be employed to accomplish different purposes. We find that no one combination of tools is likely to work best in all settings. Thus, before determining how to collaborate, public managers in conjunction with their partners should think critically about the nature of their joint work, the kind of public value they want to generate, the expected duration of the joint activities, and the level of investment they are willing to make in the collaborative process.
Public managers should be strategic in choosing which tools to use and with whom to partner. For example, the findings suggest that when government agencies partner, they tend to use tools from the shared governance dimension, but they are less likely to use these same tools when partnering with private-sector actors (for-profits and nonprofits). Recognizing that greater collaboration does not necessarily lead to better outcomes draws attention to the strategic nature of collaboration and the costs associated with the use of each tool. In sum, it is equally important to identify and adopt the right combination of collaborative tools to ensure that they are not used unnecessarily and time and effort not wasted, as to achieve the public value that is the fundamental pursuit of the collaboration.
Footnotes
Appendix
Descriptive Statistics for Variables Used in the Analysis.
| Variable name | Operational measure | M/% | SD | Range |
|---|---|---|---|---|
| Collaboration tools | ||||
| Ordinal variables (0 = none, 1 = one third or less, 2 = one third to two thirds, 3 = two thirds or more, 4 = all) | ||||
| Facilities | 1.26 | 0.92 | 0-4 | |
| Staff | 0.96 | 0.94 | 0-4 | |
| Programming | 1.49 | 0.95 | 0-4 | |
| Measurable outcomes | 1.51 | 1.11 | 0-4 | |
| Contract | 1.77 | 1.05 | 0-4 | |
| Decision making | 1.90 | 0.97 | 0-4 | |
| Power | 1.54 | 0.95 | 0-4 | |
| Goals/mission | 2.74 | 1.08 | 0-4 | |
| Trust | 2.76 | 0.93 | 0-4 | |
| Commitment | 2.91 | 0.84 | 1-4 | |
| Information | 2.81 | 0.95 | 0-4 | |
| Managerial characteristics | ||||
| Tenure | Number of years manager has been in current position | 6.07 | 6.60 | 0.2-35 |
| Government work | Number of years working in government sector | 22.16 | 9.89 | 1-35 |
| For-profit work | Number of years working in for-profit sector | 4.90 | 6.91 | 0-41 |
| Nonprofit work | Number of years working in nonprofit sector | 0.84 | 4.10 | 0-34 |
| Education | Highest level of formal education (1 = less than high school, 2 = high school diploma or GED, 3 = some college or associates degree, 4 = college graduate, 5 = some graduate school, 6 = completed graduate school) | Less than high school = 0% |
3-6 | |
| Education functional track | Series of dichotomous variables measuring functional track as law, business, public administration, or management related | Law = 4% |
0-1 | |
| Preferred partner | Binary variable (1 = nonprofit or government, 0 = for-profit) | Nonprofit = 18.3% |
0-1 | |
| County government characteristics | ||||
| Total expenditures | Average annual total expenditures (5-year average, 2009-2013) | US$131,904,199 | US$181,309,185 | US$12,094,057-US$1,263,425,623 |
| Education expenditures | Average annual education expenditures (5-year average, 2009-2013) | US$38,278,763 | US$62,593,858 | US$1,142,632-US$404,384,684 |
| Debt service expenditures | Average annual debt service expenditures (5-year average, 2009-2013) | US$15,870,945 | US$28,291,756 | US$385,390-US$210,544,792 |
| Human services expenditures | Average annual human services expenditures (5-year average, 2009-2013) | US$27,661,348 | US$38,420,716 | US$3,031,362-US$271,767,002 |
| General expenditures | Average annual general government expenditures (5-year average, 2009-2013) | US$10,259,616 | US$12,243,726 | US$1,965,506-US$82,717,541 |
| Public safety expenditures | Average annual public safety expenditures (5-year average, 2009-2013) | US$22,801,239 | US$24,637,175 | US$2,499,012-US$151,037,100 |
| Other expenditures | Average annual other expenditures (5-year average, 2009-2013) | US$17,032,284 | US$21,271,680 | US$2,419,233-US$172,657,620 |
| Collaboration amount | Percent of collaborations with partners from each sector | |||
| Nonprofit | 37.0 | 19.6 | 5-90 | |
| For-profit | 14.0 | 14.2 | 0-60 | |
| Government | 49.0 | 20.5 | 0-95 | |
| Human services | Program area with most collaborations. Dichotomous variable (1 = human services, 0 = other) | Human services = 53.6% |
0-1 | |
| County demographic characteristics | ||||
| Population size | Number of county residents (2013) | 98,480 | 151,807 | 4,109-990,977 |
| Population density | Number of persons per square mile (2010) | 195 | 260 | 10-1,756 |
| Population change | Percent population change in county (2010-2013) | 0.83 | 3.09 | −6.8 to 9.3 |
| Metro | Dichotomous variable (1 = metro, 0 = nonmetro) (2013) | Metro = 46.0% |
0-1 | |
| Percent White | Percent of population identifying as white (2012) | 68.75 | 17.64 | 27.0-94.5 |
| Median household income | Median household income of county residents (2012) | 41,674 | 7,634 | 30,031-65,826 |
| Percent below poverty | Percent of population falling below poverty line (2012) | 18.85 | 4.75 | 7.4-31.9 |
| County education | Percent of population holding bachelor’s degree or higher (2012) | 19.37 | 8.76 | 8.7-55.2 |
| Obama | Percent of population voting Obama (2012) | 43.94 | 12.40 | 23.5-75.8 |
| Tax base | Average annual tax base (5-year average, 2009-2013) | US$80,928,146 | US$120,745,115 | US$7,029,277-US$895,163,815 |
| Collaboration effectiveness | ||||
| Measure success | County measures success of collaborations (1 = yes, 0 = no) | 0.58 | 0.50 | 0-1 |
| Effectiveness | County collaborations compared with not having partnerships yield (1 = less, 2 = about the same, 3 = more) | |||
| Benefits to citizens | 2.83 | 0.46 | 1-3 | |
| Red tape | 1.8 | 0.76 | 1-3 | |
| Efficiency | 2.73 | 0.57 | 1-3 | |
| Service delivery speed | 2.56 | 0.55 | 1-3 | |
| Access to expertise | 2.76 | 0.46 | 1-3 | |
| Flexibility | 2.50 | 0.63 | 1-3 | |
| Collaboration complexity | ||||
| Collaboration complexity index | Collaboration configurations that rate highly (above median) on each dimensions: 3 = most complex (highly rated on all dimensions), 2 = moderately complex (highly rated on two dimensions), somewhat complex (highly rated on one dimension), least complex (not highly rated on any dimensions) | Least complex = 8.3% |
0-3 | |
Note. Percentages, rather than means and standard deviations, are reported for categorical variables; GED = general educational development.
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.
