The Pennsylvania State University. The Graduate School. School of Public Affairs NETWORKS IN CONTEMPORARY PUBLIC ADMINISTRATION: A DISCOURSE ANALYSIS

The Pennsylvania State University The Graduate School School of Public Affairs NETWORKS IN CONTEMPORARY PUBLIC ADMINISTRATION: A DISCOURSE ANALYSIS A Dissertation in Public Administration by T. Aaron Wachhaus 2008 T. Aaron Wachhaus Submitted in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy May 2008

ii The thesis of T. Aaron Wachhaus was reviewed and approved* by the following: Steven Peterson Professor of Politics and Public Affairs Director, School of Public Affairs Dissertation Advisor Chair of Committee Goktug Morcol Associate Professor of Public Affairs Robert Munzenrider Associate Professor of Public Administration Holly Angelique Associate Professor of Community Psychology *Signatures are on file in the Graduate School

iii ABSTRACT An abstract of the dissertation of T. Aaron Wachhaus, jr. for the Doctor of Philosophy in Public Administration presented February 2008. Title: Networks in Contemporary Public Administration: A Discourse Analysis Network concepts and theories have been of increasing interest to public administration scholars over the past twenty years. We frame organizations, inter-and intra-organizational collaborations, and theories of public policy processes in network terms; we use techniques of network analysis to examine the functioning of public organizations, and as a measure of the success of public-private collaborations; we also frame governance issues in network terms and search for effective mechanisms for network management. However, it is not clear that public administration scholars share a common understanding of network concepts, nor a common language for talking about networks. In this dissertation, I analyze the discourse on the concept of network in the contemporary public administration literature, identify characteristics used to define and discuss networks, and propose that by drawing from those characteristics already in use in the literature researchers will be able to communicate their specific interests in networks clearly and relate their work to a larger body of related scholarship.

iv TABLE OF CONTENTS List of Tables......v List of Figures.... vi Introduction......1 Chapter 1. NETWORKS IN PUBLIC ADMINISTRATION: A REVIEW..9 Network Scholarship in Public Administration........10 Defining Network... 30 Chapter 2. DISCOURSE ANALYSIS.....33 Introduction... 33 Discourse Analysis....34 Discourse Analysis in Public Administration... 39 Chapter 3. DISCURSIVE DEFINITION..62 The Importance of Definition...63 Gerring s Framework for Definitional Analysis... 69 Strategies of Definition: The Impact of Minimal Definition......80 Summary......90 Chapter 4. METHODS......92 The Analysis of Network in Contemporary Public Administration.93 Chapter 5. ANALYSIS/FINDINGS...112 Literature Survey....113 Identification of Attributes......116 Categorization of Attributes....120 Towards a Minimal Definition of Network...150 Chapter 6. SIGNIFICANCE & LIMITATIONS....154 Scope of the Dissertation.... 155 The Framework... 158 The Current Analysis......161 Conclusion......167 References...169 Appendix A: Use of Attributes by Journal...183 Appendix B: Surveyed Articles..209

v List of Tables Table 1.1 Keyword Returns by Journal..15 Table 3.1 Attributes Identified In the Definitional Analysis of Ideology By Category. 87 Table 4.1. Rank In Each Sub-Area For Selected Journals..100 Table 5.1 Number of Articles Surveyed by Journal 115 Table 5.2 Single and Repeat Authors Among the Surveyed Literature..116 Table 5.3 Network Attributes Listed Alphabetically..117 Table 5.4 Network Attributes in Descending Order of Appearances.118 Table 5.5 A Definitional Framework For Networks...120

vi List of Figures Figure 1.1 Surveyed Articles by Year Published.11 Figure 3.1 Attributes of Minimal and Maximal Definitions of Ideology..89 Figure 4.1 Adaptation of Gerring s Method...111 Figure 5.1 Number of Surveyed Articles Published by Year.115

1 INTRODUCTION Networks, and network concepts, have become increasingly popular. The internet, described as an enormous network, has been credited with changing the work and ushering in the age of the network (Castells, 2000, 2001). This new age is characterized by a proliferation of communication networks, including the internet, but also encompassing telephone and cell phone networks as well as email. We are moving towards having a single device to manage our connections to all of these networks as well as moving towards having the networks notify us of incoming messages and information, enabling constant connection to a comprehensive communication network. Social networks abound, both online in websites such as Myspace, Facebook, and LinkedIn devoted to increasing personal connections among participants, as well as through any number of organizations designed to facilitate contacts in the real world. Networking has entered our lexicon as a verb to denote these activities. Social networks have also been seen as a lens for understanding and improving business structures (Cross, 2004; Jones, Hesterly, & Borgatti, 1997) as well as theories of the policy process (Kickert, Klijn, & Koopenjan, 1997; Marsh, 1998; Sabatier & Jenkins-Smith, 1988). Much of our physical infrastructure has been described in network terms, including our power grids and transportation systems (Barabasi, 2002). Globalization is largely dependent on networks for transporting goods and information quickly and efficiently around the world (Bamberg, 1990). Network concepts inform scholarship

2 in a number of fields such as physics (Graessley, 2004; Wilson, 2004), and computer science (Barabasi, 2002, ch.11; Dorogovtsev & Mendes, 2001; and Newman, 2003). Network concepts have been utilized in many areas of public administration as well (e.g., Agranoff & McGuire, 2001b; Bressers & O Toole, 1998; Keast, Mandell, Brown & Woolcock, 2004; van Bueren, Klijn, & Koppenjan, 2003). Social network analysis techniques have been utilized as an analytic tool for evaluating organizational structure (Borgatti & Cross, 2003; Cross, 2004; Cross, Parker, & Sasson, 2003). Network concepts have grounded an approach to viewing organizations the network organization has emerged as a distinct organizational type (Agranoff & Yildiz, 2006; Considine & Lewis, 1999; Jones, Hesterly, & Borgatti, 1997; Weiner, 1990). Additionally, networks ground theories of the policy process (Bressers & O Toole, 1998; Marsh, 1998; Sabatier & Jenkins-Smith, 1993), and ground perspectives on theories and practice of governance (Considine & Lewis, 1999; Kickert, Klijn, & Koopenjan, 1997; Sorensen & Torfing, 2005). Given the scope of applications of network concepts, it is not surprising that a number of criticisms of networks have also been raised. Olsen (2006) rejects notions that networks are transforming public administration, arguing that the traditional bureaucratic perspective has at least as much utility as networks, has not been lessened by the advent of networks, and has an extensive body of well-developed scholarship behind it. Networks, on the other hand, have been the subject of a smaller and much more recent corpus of work, leading Olsen to question whether they represent a new wave in public administration or merely a passing fad. More than a decade earlier,

3 Dowding (1995) had argued that network concepts had not been developed sufficiently to have explanatory power; rather, they exist solely as mental constructs, metaphors that may guide our thinking. Borzel s (1998) survey of the state of the network literature in public policy came to a similar conclusion, noting that definitions and uses of network exist in a Babylonian array. The analogy to the tower of Babel is an apt description of the state of the network literature, as the central lessons of the story revolve around the power of language: Mankind was engaged in the construction of an enormous tower that would reach into the heavens, putting men on an equal footing with God. While the tower itself was an impressive structure, the true power of the effort was linguistic the people would make a name for themselves (Gen. 11:4), gaining political power and renown through unity in an echo of the power associated throughout the Bible with the power of God s Name. God s response was equally linguistic: rather than destroy the physical structure, He destroys the structure of language - confusing the speech of men, fragmenting their previously united language so that no common understanding was possible. Lacking clear language, collaboration on the tower ceases; the fall of the tower itself is not described in Genesis as it is a consequence of the destruction of meaningful language. The lesson of Babel that clear language and a common understanding is a necessary foundation for any collective endeavor - provided the impetus for this dissertation. Nobel laureate George Thomson defined concepts as ideas which receive names (1961, p.4). Gerring has more recently pointed out that in order to

4 talk about anything at all one must call it by a name (2001, p. 20). Moreover, he pointed out, it matters not only that we name our concepts, how we define our concepts is critical as well (p. 38). As Sartori has put it bad language generates bad thinking (1984, p. 15). 1 The issue here is whether, lacking a common language, we can build a conceptual edifice of lasting stability? Thus, my central research question is simply, Is there a common definition for networks currently in use among public administration scholars? Based on previous assessments (Agranoff &McGuire 2001; Borzel, 1998; O Toole 1997a) as well as my own reading of network literature, I did not anticipate finding a widely used definition. Accordingly, my follow-up questions are If there is no common definition, can common elements among the various definitions of network be identified? and Can I propose a definition based on these common elements? In the next chapter I begin to address the question of whether there is a common definition of networks by reviewing literature. I first discuss work that criticizes the fragmented nature of the network literature. My own review of recent network literature emphasizes the lack of any unified meaning and highlights several topic areas around which the literature tends to congregate. As a first step towards answering my follow-up questions, I analyze literature in order to identify common attributes used to refer to networks and to assess the level of coherence or fragmentation in that literature. I do so using techniques of discourse analysis. 1 This, in turn, echoes Orwell s observation that if thought corrupts language, language can also corrupt thought (1946).

5 Discourse analysis is itself quite a broad term, encompassing a variety of approaches to examining language. In general, discourse analysis studies the ways that language is used to communicate meaning; it focuses on everyday, practical language, or language-in-use (as opposed to the study of the formal structures of a language). In chapter three I establish the approach to discourse analysis used in this dissertation. I draw on the theoretical and methodological approaches to discourse analysis developed by Jonathan Potter, Stephanie Taylor, and Margaret Wetherell (Wetherell, 2001; Potter & Wetherell, 1987, 2001; Taylor, 2001) as well as the foundational linguistic work of Wittgenstein (1951, 2001). I follow the view that language is purposive and constructive and that meaning is open to interpretation. I therefore believe that how we talk about networks shapes how we think about networks. The question of network definitions is of critical importance, as the lack of a common definition will hinder the coherence of any field of network studies. I also provide a context for my research in chapter three through an exploration of discourse analysis in public administration. Discourse analysis has been put forth as an approach to understanding public administration (White, 1999) as well as a necessary complement to the rational analysis of public policies (Fischer, 1980, 1995, ch. 1). Scholars have also asserted that public policies and public management may be analyzed as narratives that the stories that we tell and the language that we use to tell them are an important means of understanding what goes on in the public sector (McSwite, 1997; Roe, 1994). Discourse analysis has also been used as a technique for analyzing specific policies (Outshoorn, 2002; Yanow, 1995, 1996). I see my own

6 research closely related to Miller and Jaja s discourse analysis of public administration symposium articles (2005). Their use of discourse analysis to investigate whether public administration scholarship makes sense as a story (whether there a consensus as to what public administration is, or should be, about) resonates with my own research question (1995, p. 729). I also see my dissertation as a direct response to Hay and Richards call for a discursive definition of public administration networks (2000). Their study of British civil service networks concludes that current analytic definitions of networks are often conflicting and do not encompass network concepts well. They suggest that a discursive definition may better communicate network concepts. I introduce the methodological basis for my analysis in chapter 4. I follow a framework for the definitional analysis of abstract social science concepts developed by John Gerring (1997; 2001). Gerring s concern with the proliferation of definitions for abstract social science concepts is similar to the concerns raised by Hay and Richards. In response to the problem of multiple competing definitions, Gerring developed a framework for the definitional analysis of abstract concepts that proceeds in three broad steps. First, a term of interest must be identified and a body of literature is identified and surveyed, and relevant passages are collected. Second, attributes referring to the term are identified and classified into logically related categories. Finally, a definition is constructed from a reduced set of attributes common throughout the literature. The utility of this definition lies in that fact that it has been constructed from elements present in a body of discourse. Thus, rather than extending a concept s

7 meaning by adding descriptors, this approach brings coherence by defining in terms that enjoy a level of consensus among those using a concept. Since Gerring s work is largely conceptual and proceeds according to the three general steps outlined above, I discuss the particular methods I use to adapt Gerring s framework to the analysis of public administration networks in chapter five. I defend my decision to limit the literature survey to articles published over the past 20 years within ten peer-reviewed public administration journals, for example. I then detail my methods for the particulars of the framework, including the identification and classification of attributes, the means of determining a sufficient intensity of analysis of the literature, and my selection criteria for determining the set of common attributes from which I construct a definition. The results of this analysis are presented in chapter five. I identify 56 attributes in the surveyed literature that are used to refer to networks. I concur with earlier assessments that there is no agreement as to a meaning for networks, as no attribute was utilized in even half of the literature, and as nearly half (26 of the 56 attributes) occur in less than ten percent of the surveyed articles, I conclude that the level of fragmentation of meaning is high. I categorize those 30 attributes that occur in more than ten percent of the literature, and briefly discuss each in turn. I conclude my analysis with a discussion of the difficulties attendant upon any attempt to construct a definition in light of the degree to which the literature reflects idiosyncratic definitions and the lack of a common base of shared meaning on which to draw. However, I do

8 suggest five attributes that may serve as a foundation for bringing a greater degree of definitional coherence to the concept of network. I conclude with a final chapter addressing the significance as well as several factors limiting this study. In my view, the most significant factor is the degree to which there is no agreement as to what a network is. Although the impetus for my research came from a sense that scholars were, to an extent, talking past one another (a position bolstered by similar claims in the literature [esp. Agranoff &McGuire 2001; Borzel, 1998; O Toole 1997a]), I did expect to see that there was at least a majority consensus somewhere in the literature. It is my firm belief that no meaningful work in this area can be done until basic issues, such as meaning, are resolved. The lack of consensus also may be seen as the largest factor limiting this analysis. Faced with minority opinions throughout the literature, one can legitimately question any attempts at constructing a definition based on usage in that literature. This situation does, I believe, strengthen the need for work towards a consensus of meaning.

9 CHAPTER 1 NETWORKS IN PUBLIC ADMINISTRATION: A REVIEW Our research questions give priority to discourse, in any form, and ask about its construction in relation to its function. J. Potter and M. Wetherell 1897, p. 6 Discourse analysis provides a method for systematically examining the exchange of ideas. Regardless of the medium of communication (e.g., spoken, written, graphically depicted), discourse entails a cycle of receiving and responding to messages. Before moving to the analysis of the discourse, it is first necessary to establish the arena and the subject of discourse as well as assess the level of reception and response. In other words, before analyzing discourse, we must understand a) what we re talking about b) how and where we re talking and c) whether we re listening to each other. The last question, in particular, relates to my research questions proposed in the previous chapter. Are we listening to each other? The public administration network literature has been criticized for being fragmented. Is there a coherent body of discourse (with an exchange of ideas back and forth among scholars) or are network researchers talking past each other? I suggest that one measure of this body of discourse is to assess the level of consensus as to the definition of what a network is. Is there agreement as to what a network is - to what the characteristics of networks are?

10 The discourse on networks is a conversation carried out through the written word and accessed either in print, or increasingly, electronically (how and where we are talking). In particular, the literature I survey for my analysis consists of peerreviewed academic journal articles, all of which are available online as well as in printed editions. The ease with which online databases facilitate instantaneous searches through vast amounts of material informs my literature review. I draw on search protocols by organizing my review of the literature by keyword search returns. The first question that of what we re talking about makes up the bulk of this chapter s review of the network literature. I begin, however, with literature that asserts the increasing interest in, and importance of network concepts to public administration. Network Scholarship in Public Administration Increased Interest in Networks I do not know of any empirical survey of network-focused scholarship in public administration. However, there is a sense among some researchers that networks are an area of growing interest to both scholars and practitioners (see discussions in Agranoff, 2007; Kettl, 2002; O Toole, 1997a). My own survey of the public administration network literature bears this out. I reviewed articles published in ten public administration journals during the past 20 years and identified 127 articles

on networks. 2 Figure 1.1 displays the number of these articles appearing in each of the past 20 years. 11 Figure 1.1 Surveyed articles by year published. 25 20 15 10 5 0 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 Number of articles published As shown in Figure 1.1, there has been an increase of published journal articles on networks. In particular, the past seven years has seen a marked increase in network scholarship. Agranoff and McGuire (2001b) call for scholarship on what they see as the big questions in network management; they ground their call on the increased use of networks in public management. In seeking to develop an empirically derived knowledge base for network management, they assert that network management requires a new paradigm, separate from the hierarchical-organizational authority 2 A list of the journals surveyed may be found in Table 1.1. The rationale for the selection of these journals is discussed in chapter four.

12 paradigm framing bureaucratic management (p. 297). 3 Network management is growing to dominate public organizations and governance; the authors assert that a new paradigm for understanding the tools and techniques of network management must be developed. O Toole also sees public administration as increasingly taking place in network settings (1997a). Not only do networks represent a common state of affairs in public administration, O Toole asserts, they are likely to increase in both number and importance. However, in his view, networks are not taken seriously enough: practitioners are ill-equipped to deal with the network-centric reality of modern public administration, while scholars have not sufficiently addressed network issues. In order to rectify this situation, O Toole proposes a multi-pronged approach designed to frame practical, descriptive, conceptual, and empirical agendas of network management in public administration. Imperial (2005) reiterates the growing interest in networks, asserting that more researchers recognize the ubiquitous nature of network relationships, the roles they play in social and organizational life, and their importance to policy implementation (p. 281). Imperial s study of the effectiveness of network governance as a strategy for watershed management grows directly out of O Toole s call for increased empirical data on networks, and aims to provide both descriptive as well as comparative results. 3 Compare with Agranoff s more recent view of networks. While he characterizes networks as fundamentally different from hierarchical organizations and founded on a different base (exchange relations), he does not explicitly call for the development of a separate paradigm. (2007, ch. 2).

13 The above articles note that networks are of growing importance to modern public administration, emphasizing that networks are increasingly common and are well suited to the demands of the complexity of modern public administration. Therefore, it is expected that the use of networks will increase, as will research into networks. An additional line of scholarship expresses this sentiment in much stronger terms. Kettl (2002, 2000), for instance, traces the growth of networks in public administration over the past 20 years and asserts that much of the work of public administration has occurred via networks. He states that it is impossible to dispute the existence and importance of these networks. It is easy to bet on their permanence (p. 494). The central challenge for public administration, according to Kettl, is to integrate the network-based governance and delivery of services with a government designed and structured around notions of hierarchical authority and vertical relationships. While Agranoff and McGuire (2001b) note the increased use of networks and suggest that network management was in need of a paradigm of its own, they can also be more emphatic: In comparing four models of management in a federal system, they declare that The age of the network has arrived, supplementing previous group, hierarchy, and bureaucratic eras (2001a, p. 677). Thus, while network management may be in search of an operating paradigm (i.e. misunderstandings regarding the causes and applicability of networks remain), they assert that it already has come to be the dominant mode of governance. Moreover, they maintain that

14 network management has not yet reached its full ascendancy, but will increase over time and become the hallmark of the information age. Although O Toole observes a surge of interest in networks and notes that public administration is increasingly occurring in a networked setting (1997b), he laments that networks are not being taken seriously (1997a). His 2004 work on cooptation and the problems that networks can cause opens by noting that network research and literature has burgeoned in recent years. In it, he also asserts that networks are now a crucial feature of the public administration landscape (O Toole & Meier, 2004). Organizing the Surveyed Literature Networks have been widely utilized across public administration. As Table 1.1, above, shows, the growth of interest in networks is relatively recent among public administration scholars. I grouped articles on the basis of keywords in either the article title or abstract. Table 1.1 shows the distribution of articles in journal for the most commonly appearing keywords.

Table 1.1 Keyword returns by journal 4 Policy Governance Management Analysis Administration & Society (A&S) 7 3 3 3 American Review of Public 4 5 4 Administration (ARPA) 4 International Journal of Public Administration (IJPA) 3 5 4 Journal of Policy Analysis & Management (JPAM) 2 2 Journal of Public Administration 6 6 10 8 Research & Teaching (JPART) Policy Sciences (PS) 7 2 2 3 Policy Studies Journal (PSJ) 9 1 1 3 Public Administration Quarterly (PAQ) 1 3 Public Administration Review 16 5 16 15 (PAR) Public Performance & Management Review (PPMR) 1 15 As may be seen in Table 1.1, these articles account for just under three quarters of all the literature I surveyed (93 articles). Although in several cases two or three of the remaining articles shared a focus, no other sizable groups emerged. Networks and policy. Networks and Policy are present in the title or abstract of 52 articles, or just over 40% of the literature I surveyed. These are not exclusive categories as several of the articles utilized overlapping sets of keywords. However, my object is not to construct a typology, but to show how the literature is dispersed. The fuzziness of these areas highlights the need for conceptual clarity where do the differences and 4 See chapter four for my criteria for selecting these journals.

16 similarities between governance and policy lie with regards to network concepts, for example. A substantial portion of these articles focus on theory, some at a very broad level. For instance, Agranoff and McGuire argue that the American federal system itself is a network, and that the processes of American government may be best understood using network concepts. They further assert that the processes of federalism have merged with public administration itself: public administration has become networking (2001a, my emphasis). It follows, therefore, that implementing network theories and practices will provide the next management paradigm for public administration itself. Indeed, Agranoff and McGuire herald the age of the network contrasting this approach against earlier eras in public management including hierarchical and bureaucratic. Miller (1994) begins from the position that networks form a type of social structure distinct from markets and hierarchies. He notes, however, that while critiques of market- and hierarchy-based practices abound, these modes remain dominant in the public sphere while network-based practices remain relatively undeveloped. To show how a network perspective informs public management, Miller interprets a single case (water basin restoration) in light of each of these three models of management. He argues that while each approach to this case has its own strengths, the network approach is more open. That is, it includes a more diverse range of actors, acknowledges that interaction and exchange among these actors is an inherent part of public administration, and focuses directly on the meanings made by these interactions

17 rather than seeking to situate them within pre-established frameworks. In doing so, public administration becomes better able to incorporate dissenting views and so becomes more democratic and responsive. Klijn (1996) examines the conceptual uses, value, and limitations of policy networks. In focusing on the conceptual strengths and limitations of policy networks, Klijn aligns himself with the scholarship discussed above that cited the need for a wider body of theory as well as increased empirical investigation to properly frame networks (e.g., Agranoff & McGuire, 2001b; O Toole, 1997). For Klijn, network concepts work well as descriptors of policy making, but lack the clarity and coherence necessary to move beyond the level of description. In particular, Klijn notes that there are a wide range of conceptions and definitions of what a policy network is. Specifically, he argues that network definitions have focused on structural elements of networks and have attempted to give a static picture of what is essentially a dynamic relationship: the relationship between structure and action has been insufficiently emphasized in conceptual work thus far 5. He further notes that, while network concepts stress the importance of context, there is as yet no theoretical framework for understanding the role of context in networks. Klijn suggests that adopting a strategic (i.e. action-oriented, behavior based), rather than structural, view of policy networks may yield valuable insight into network management techniques. 5 A structural view of networks may be quite broad, defining networks as a set of interdependent relationships fostering resource exchange and bound by a system of rules; or, it may be more narrowly focused on a particular network, addressing, for example, the number of actors in the network, the forces maintaining the network, or the particular system of rules in place. Klijn views network structure as the pattern of relations among actors in a network, but notes that the patterns themselves ( action ) has not been investigated how do these patterns form, what keeps the relations stable?

18 Borzel s review of the policy network literature reiterates Klijn s conclusion that the lack of definitional clarity has limited the usefulness of network concepts (1998). After providing an extensive survey of scholarship on policy networks from British, German, and American sources, Borzel concludes that there is little agreement as to what a network is, what its proper role is (as an analytic or theoretic tool), and even whether or not networks exist. She, too, notes that the development of a research agenda providing shared purpose and direction to scholars might strengthen the contributions of network scholarship. A number of pieces work towards building a stronger theoretical framework for networks by focusing on a particular aspect of networks or on a specific perspective or theoretical approach. For example, Blom-Hansen (1997) advocates a new institutional perspective on policy networks, while O Toole and Meier explore the dark side of public network management (2004). Blom-Hansen s new institutional approach is similar to Klijn s (1996) call for a strategic, or action-based, approach to networks. Blom-Hansen defines institutions as the rules constraining the actions of the participating actors (p. 669), and applies institutional theory to policy networks. In other words, Blom-Hansen attempts to incorporate a theory of actors behaviors into policy networks. Cashore and Vertinsky (2000) take a neoinstitutional view of networks, focusing on action and behavior on an organization (or network) level. Where Blom-Hansen examines the behavior of individuals within networks, Cahsore and Vertinsky look at factors that may affect the behavior of the network as a whole. Like O Toole and Meier, they conclude that the external forces in

19 the environment in which a network functions may impact the behavior of the network. O Toole and Meier (2004), on the other hand, address the political contexts in which policy networks function. They point out that networks are often viewed as neutral structures that happen to be made up of political actors and occur in political environments. In this model, networks remain unaffected by their political environment and continue to function for the greater public good. O Toole and Meier explore the possibility that politics may influence the behavior of network managers, steering a network away from neutrality towards a politically biased strategy. This work shares with both Klijn (1996) and Blom-Hansen (1997) an emphasis on the behavior of actors in a network, rather than just the structural aspects of networks. Just over half of the articles on policy networks are case studies that present a wide array of applications of network concepts in a variety of settings. Six pieces focus on issues of environmental policy: Whether looking at agricultural (Greer, 2002; Toke & Marsh, 2003), industrial impact (Daugbjerg, 1998; Van Bueren, Klijn, & Koppenjan, 2003), or other environmental issues, there is consensus in these articles that a network approach is able to deal with the complexity of environmental policy issue in an open and democratic manner by including a broad range of concerned actor. Other scholars consider issues of democracy (Rynck, 2006) and social capital (Broadbent, 2000) within policy networks. Finally, issues of governance in policy networks are analyzed in four cases. Damgaard (2006) looks at four Danish municipalities employment policies to see whether the existence of policy networks

20 leads to network governance as a municipal strategy. Bache (2000) examines the connections between policy networks and mode of government, concluding that, at least as regards economic policy in Yorkshire and the Humber, a resurgence of policy networks has not significantly affected government structures. Cloke, Milbourne, & Widdowfield (2000) explore the emergence of policy networks to deal with homelessness in Taunton. Networks and governance. Network and governance returned 25 articles that focus on the relevance of network theory to governance issues, explore particular aspects of governance using network concepts, or use a network perspective to analyze case studies. There is some overlap between the governance and policy network pieces. For example, Agranoff and McGuire s comparison of public management techniques discusses governance issues in a network context (2001a). Similarly, Cloke, Milbourne and Widdowfield s study of homelessness analyzes how policy networks operate in local governance settings (2000). The theoretical approach to network governance differs from the theoretical work on policy networks. The policy network literature noted the increasing importance of network concepts to the policy process and called for the development of a coherent theoretical framework for applying network concepts to public policy. While the governance scholarship refers to a shift from government to governance, there is no call to develop a theory of governance or to develop a

21 framework incorporating networks into governance. Rather, network governance scholars are content to describe the close association of networks to governance, and to consider the impacts that network concepts may have on governance practices, rather than developing an empirical foundation for exploring those impacts. Kettl, for instance, speaks of both government and governance but does not explicitly relate these two concepts (2000). He describes networks of communication, information, and transportation as having transformed governance, and describes the impact of networks on governance, but does not suggest that a network theory of governance would be of practical help. Similarly, Goodsell s new vision for public administration (2006) acknowledges that governance is replacing government as an activity, and that governance is closely associated with networks. Both networks and governance are accepted as facts of public life requiring thoughtful management rather than theoretical issues requiring a framework for inquiry. Stoker (2006) takes much the same approach, noting that network governance has emerged and that the pressing issue is to determine the best management techniques as well as how to implement the techniques in government settings still accustomed to hierarchy and bureaucracy. Considine and Lewis (1999) consider the impact of networks and network governance on frontline bureaucrats and identify a number of strategies for more efficiently working in a network setting. Cloke et al. (2000) examine the importance of partnerships in policy network as a governance practice. Agranoff and McGuire analyze the challenges involved in managing multiple networks across levels and units

22 of government. Most recently, Graddy and Chen (2006) have examined factors that may influence the size and scope of service delivery networks. Finally, note that work addressing network governance theory or addressing properties of governance networks makes up only half of the policy governance articles. Ten of the twenty articles on network governance are case studies. Some of these are quite broad in scope, seeking to illustrate network governance in governmental processes. Bowman (2004), Sorensen (2006), and Considine and Lewis (2003) each apply a network governance perspective to the examination of governments. Considine and Lewis compare governance models in four countries, while Bowman investigates interactions between American states, and Sorensen examines how Danish politicians have been affected by network governance processes. Other cases are more narrowly focused on particular programs. Wheeland (2003) studies a community strategic plan in South Carolina. Musso, Weare, Oztas, and Loges (2006) look at community governance networks in Los Angeles. Cloke et al. (2000) analyze homelessness in Taunton, England. Comparing network governance scholarship and policy network scholarship. The work on policy networks provides a range of theory and application, as does work on network governance. As noted, policy network theorists have called for a coherent theoretical framework on which to hang network research, a call that is absent from the network governance articles. This may be, in part, due to the tendency of policy network scholars to approach networks as structural entities. Descriptive

23 work lays out the extent of policy networks, details the actors in a given network, and maps out their relationships. Empirical work also focuses on mapping networks and investigating the characteristics that make up the ties between actors in a network. Policy network scholars are concerned with questions such as: how do networks form? What brings (and holds) actors together in a network? Why do networks fail? Network governance scholarship, on the other hand, approaches networks from an organizational, rather than structural, perspective. Thus, network governance scholarship focuses on questions of management, and looks for techniques that will allow for governance networks to function more effectively. The focus is less on the network itself than with the ways that governance networks operate in a given context. Descriptive work maps not the network itself, but the ties between the network and its surroundings. For example, Wheeland (2003) looks at the results of a network governance strategic plan ten years into the program; Bevir and Rhodes (2001) and Considine and Lewis (2003) use a network governance perspective to comment on government systems; Bevir and Rhodes to interpret the British government; Considine and Lewis to facilitate comparisons between models of governance. There is overlap between these two keyword areas several articles return searches under governance as well as policy. However, there are a number of works that treat governance as separate from policy. I have indicated briefly some differences between network governance and policy research. Clarity of meaning is essential in areas such as these, where it is not always clear where one domain ends and another begins. I have found no work that confronts the question of whether

24 network holds the same meaning in the governance literature as it does in the policy literature. If there is a single meaning, the differences mentioned here become problematic; if there is not a single meaning, the areas of overlap between governance and policy need to be clarified. Clarity of meaning and consistent use of terminology is necessary for meaningful scholarship. Networks and management. A fourth set of articles focuses on networks and management. Network management is the newest 75% of these articles were published in the past five years. McGuire (2002) sees network management as a critical component of public administration, but one that lacks a robust theoretical framework. He proposes a multifaceted research agenda similar to that called for by O Toole (1997a), noting that empirical work describing networks and investigating various aspects of networks must be complemented by research describing and explaining behavior and choicemaking in networks. To that end, he begins to develop what may be seen as an analog to traditional management s POSDCORB tasks of activation, framing, mobilizing, and synthesizing. McGuire s work on this topic has stimulated some research. Meier and O Toole s work with large-n samples (2003, 2005) is in direct response to McGuire s assertion that network research should be conducted in small-scale settings. In framing a proactive management perspective on network management, Goerdel (2006) also indirectly takes up McGuire s project by examining behaviors in networks.

25 However, most network management research consists of empirical studies - either case studies of network management in specific settings, such as public schools (Meier & O Toole, 2001; 2003; 2005; Mossberger & Hale, 2002) or homeland security (Caruson & MacManus, 2006; Wise, 2006) or the examination of a specific aspect of network management, such as tenure within a system (Juenke, 2005) or program performance (Meier, O Toole, & Goerdel, 2006). Networks and analysis. A number of scholars view networks as an analytic tool. While policy network research focuses on the structural aspects of networks and governance scholars see networks as a form of organizing public management, this work uses network concepts as a tool for analysis. Klijn (1996) argues that network-based analysis may bring new insights into the policy process by focusing on strategic interaction among policy actors rather than the structure of their relationships. In this instance, the supposition is that a network perspective may provide new insights in the analysis of public policy. However, by network analysis scholarship, I refer to a specific use of network analysis. Choi and Brower capture this meaning succinctly: Network analysis includes various concepts and techniques designed to analyze relational data. Unlike attribute data that focus on the characteristics or features particular to individual actors, relational data are based on the connections that relate one actor to another. (2006, p. 654) In other words, network analysis is oriented towards the description and analysis of strategic behavior and action rather than structure.

26 A range of specific techniques have been developed to analyze relational data in networks; these articles utilize these techniques to examine networks in the public sphere. While policy network research calls for development of a theoretical framework and a greater base of empirical network research, and network governance scholarship seeks new techniques for managing in networks, network analysis scholarship aims to provide specific methods for examining relationships in networks. 6 In their introduction to a symposium on networks in public administration, Bogason and Tooning (1998) highlight the need for tightened network definition and scholarship by collecting a range of meanings for the term, including a) the study of relationships between institutions; b) a new way of analyzing traditional governance arrangements; and c) the study of public sector structures from a perspective emphasizing interdependence, diversity, fragmentation, and the need for consensus (p. 205). While each of these definitions utilizes networks as a means of analysis, it is not at all clear what the relationship between them ought to be. Agranoff and McGuire (1998) analyze the relationships between networks for economic development in 237 cities. They examine the types of networks in place and the variations among these networks as well as the strategic interaction between networks. They compare management techniques of single organizations to those of interdependent networks and discuss the implications for networks as a setting for 6 Social Networks, a peer-reviewed journal, is devoted to network analysis.

27 designing and implementing public policies. Again, the network itself has become the means of analyzing management techniques. 7 Most of this work, however, is directed at specific uses of network analysis. Provan, Veazie, Staten, and Teufel-Shone (2005), for instance, present ways in which network analysis may be used by community leaders to build community by strengthening relationships between public and non-profit organizations. The majority of articles, however, detail the application of network analysis to specific cases, Milward and Provan (1998) in mental health networks, Choi and Brower (2006) in emergency management, and Daugbjerg (1998) in Danish and Swedish environmental policymaking, among others. Remaining articles. The four keywords I have identified within the surveyed literature account for just under 75% of the total articles collected (93 of 127). I could not identify other keywords that accounted for groups of more than two or three articles in the remaining literature. As the classification by keyword is intended to highlight trends within network scholarship rather than to make a complete accounting of all network research, I present some brief observations on the remaining literature in order to give a sense of the range of topics and approaches covered. Several articles work at a basic level, providing descriptions of networks, observing how particular networks in the public sphere function, or raising basic 7 Agranoff s most recent book (2007) extends the depth of analysis of public management networks by examining the management processes of fourteen networks in the public sphere.

28 questions on topics including the formation and stability of networks as well as forces that may bring actors together in a network. O Toole (1997b) provides descriptions of public networks. Hall and O Toole (2000) search for evidence of networks in national programs, while Hall and O Toole (2004) examine how common regulatory networks actually are. Finally, Olsen (2006) is one of the few scholars to question the rise of networks, how widespread networks may be, and whether networks may be said to have supplanted more traditional forms of management. Some works explore structural aspects of networks. Recall that some, such as Klijn (1996), have advocated a strategic, dynamic view of networks and have urged scholars to examine networks as relations between actors. Others, however, have focused on structural aspects of networks not on the relationships themselves, but on the means by which actors come together, exchange resources, and interact. Graddy and Chen (3006) examine the impact of specific structural aspects of a network (e.g., size, resource dependency, diversity, and asymmetry) on the ability to deliver social services. Hall and O Toole (2004) examine the formal structures for making rules in regulatory networks. Hay and Richards (2000) compare structural views of networks with a view of networks as a dynamic social practice. Keast, Mandell, Brown, and Woolcock (2004) make a distinction between networks and network structures. Network is a more general term used to denote informal connections between people, while network structure applies to a more formal arrangement in which actors are aware that collaboration on broad goals is the only means of progress.

29 Summary Networks have been of increasing interest to public administration scholars; the past twenty years have seen explosive growth in published research in this area. Network concepts have been applied to a wide range of subjects and cases in public administration. In order to organize the public administration network literature, I used a keyword search to identify groups within the literature surveyed for this dissertation. Nearly 46% of the articles surveyed dealt with policy networks, either applying network concepts to theories of the policy process, or analyzing case studies from a network perspective. Network governance articles focus on the management of networks in the public sphere, although there is development of theory in this area as well. Network management as a more general topic occupied a number of scholars; research in this area consists almost entirely of case studies. Finally, the use of network analysis constitutes a fourth group of articles. Work in this area used network analytic techniques to investigate relational data. Recall that these four keyword areas account for 75% of surveyed literature. Thus, from this review of the literature, I conclude that while there are identifiable areas of shared interest, much of the literature does not cohere around a particular subject or approach. Given this condition, is it possible to speak of a network literature in public administration? That is to say, is there a common understanding of what a network is or what network scholarship constitutes among public administration scholars? Is there a shared set of methods for studying networks or a common approach to network research? Before introducing my own research on the