The State of Network Research in Public Administration

Networks and Research Streams

There remains a wide disagreement on the definition of networks. A few scholars use networks as a metaphor to describe interorganizational relationships (Isett et al., 2011) or as organizational forms that are different from markets or hierarchies (Powell, 1990). Some scholars maintain that it is not even necessary to provide a universal definition of networks (e.g., Borgatti & Foster, 2003). Networks can be generally defined as a set of nodes or actors and relationships between these nodes in sociology, network sciences, and many other disciplines (Borgatti, Everett, & Johnson, 2013). In the field of public administration, networks are defined either as interorganizational collaboration arrangements or as new governance structures designed to achieve a common goal that cannot be achieved (or that cannot be achieved effectively) by one single organization (Agranoff & McGuire, 2001; Koliba et al., 2010; O’Toole, 1997). Despite the differences, most definitions of networks in public administration highlight the importance of collective action, common goals, and relationships between organizations (Provan et al., 2007). The majority of network studies in public administration are focused either on the interorganizational collaborations to achieve management or policy goals that are beyond the scope of a single organization, or on the network governance structure and process that differs from traditional bureaucratic structure and involves nonstate stakeholders in policy-making and implementation. Although some scholars use networks and SNA interchangeably, the latter refers to analysis methods for studying structures and interactions within social structures (Scott, 2013). The variations in defining networks connote the wide applicability and scope of the concept of networks in the field.

To further disentangle the complexity of networks in public administration, researchers categorize network research into different research streams based on the research foci. Rethemeyer and Hatmaker (2008) maintained that existing policy network research and collaborative network research are related but different. Policy network research focuses on networks involving traditional policy makers such as public agencies and legislative officials, along with nontraditional players such as private entities, interest groups, and nonprofits that have an interest in specific policy domains. Collaborative network research, however, concerns the provision and delivery of public goods and services or the implementation of public programs (Rethemeyer & Hatmaker, 2008). Isett et al. (2011) reflected on the current status of scholarship on networks in public administration and added governance networks to the two network research streams discussed by Rethemeyer and Hatmaker (2008). Governance networks focus on the coordination and governance processes to achieve common goals. Similarly, Lecy et al. (2013) identified three domains of network research in public administration: policy formation networks, governance networks, and policy implementation networks. Policy formation networks received relatively little attention compared with the other two domains and also noted that advanced network analysis remains limited in the field. Whole service delivery networks (Provan & Lemaire, 2012), or policy implementation networks (Lecy et al., 2013), are most prevalent in public administration.

Challenges in Network Research

Although network research has gained momentum in the past few years, some fundamental challenges remain to be addressed (Isett et al., 2011). More research is needed to reach a consensus on definitions of key terms, improve conceptual clarity, and clearly define theoretical frameworks of network research (Lecy et al., 2013). Isett et al. (2011) summarized the conceptual concerns facing public administration network scholarship. They noted a lack of agreement on the definition of networks and units of analysis, inconsistency in terminology, insufficient attention given to informal networks, and difficulty in defining network boundaries and collecting network data. Lecy et al. (2013) found that several articles failed to describe and define the networks clearly and that very few articles clarified their units of analysis and network boundaries. These issues may pose hurdles to advancing network scholarship within the discipline.

To address these challenges, most of these review articles highlight the need to advance network research methods and apply proper analytical tools and methods. Significant improvement has been made in understanding networks in public administration from theoretical and conceptual perspectives; however, more work is needed to enhance the methodological rigor of social network research (Robinson, 2006). The ensuing section reviews SNA in public administration.

SNA in Public Administration

SNA is an analysis method and tool used to analyze theoretical constructs and concepts that are defined as relational processes and outcomes (Wasserman & Faust, 1994). A social network consists of both nodes and ties. Nodes, or actors, within a network can represent individuals, groups, organizations, communities, and nations that make up the networks. The relationships between nodes or actors are linked through ties. These ties can indicate communication between nodes, information exchange, formal contractual relations, or friendship ties between nodes. The relationships between nodes or actors can be either formal (legal/contractual) or informal (based on trust and understanding, or interpersonal relationships; Binz-Scharf, Lazer, & Mergel, 2012; Borgatti et al., 2013; Provan et al., 2007; Scott, 2013). These various types of relationships can form different types of networks, including but not limited to affiliation networks, resource exchange networks, mentorship networks, advice networks, information exchange networks, friendship networks, shared belief networks, and formal coordination networks.

SNA presents ways to analyze dynamic relationships between various actors and to examine complex social processes and various types of interactions within social systems. Both intraorganizational and interorganizational relationships can be studied using SNA. In public administration, the unit of analysis can be government employees within a department or an agency, or organizations within a city or a county, or counties within a state, or even states. Networks are being studied using SNA to illustrate the collaborative, cooperative, and conflicting relations between actors involved in the policy process and in policy output (Ingold, 2011).

SNA has been applied to a limited extent in public administration when compared with other fields such as sociology. Yet, with a movement toward collaborative governance and management through networks, there is ample room for the application of SNA in theory and praxis. To understand collaborative processes, challenges of collective action, and evaluate network outcomes and performance, public administration researchers need to adopt both an egocentric micro approach (common in organizational science) and a macro approach to studying whole networks (Provan & Lemaire, 2012). The past few years have seen an increasing amount of network research and SNA use, in particular. Yet simpler measures of SNA are mostly utilized in the field, and advanced network measures and analytic methods remain underused when compared with neighboring disciplines such as sociology and political science (Lecy et al., 2013).

Berry et al. (2004) suggested that intellectual cross-fertilization between disciplines such as sociology, political science, and public administration is required to build strong theoretical and methodological frameworks for the understanding and application of networks. Recently, Provan and Lemaire (2012) have drawn lessons and ideas from social and business network literature to advance the network literature in public administration. A recent content analysis of network publications conducted by Lecy and his colleagues suggested that few articles that focus on networks actually utilize SNA methods and measures; those that did apply SNA usually used simple measures, such as centrality. Their sample of 82 articles was chosen from among frequently cited journal articles (Lecy et al., 2013). To further understand the status of SNA in public administration, a closer examination of SNA research is needed. This article focuses on methodological issues of network research and reviews a more comprehensive list of articles that apply SNA in public administration. This article describes the types of networks studied in public administration research, discusses methodological advancements and challenges, and identifies research gaps for future network research.

Topics in Public Administration Network Research

Previous reviews of network research in public administration have grouped various types of networks into three categories: policy networks, governance networks, and collaborative networks (Isett et al., 2011). As Figure 2 suggests, our review of the 81 SNA articles shows that half of the articles fall under the governance networks research stream. Articles in the policy networks and collaborative networks streams each account for nearly one third of the total number, respectively. In addition, 7.5% of the articles fall under both policy and governance networks research, and approximately 9% fall under both governance and collaborative networks research.

OPEN IN VIEWER

Figure 2.

SNA utilization in network research streams in public administration.

In addition to examining the broad research streams, we took one step further to look closely at what specific management and policy issues are studied through the lens of SNA within each research stream. Nearly 85% of the 81 SNA articles (68) center on the following management issues or policy domains: health and human services, regional or community alliances, economic development, emergency management, environmental policy, education policy, urban planning and development (Figure 3).

OPEN IN VIEWER

Figure 3.

Management issues and policy domains of network research in public administration.

It is interesting to note that some of these management or policy issues are closely clustered around a specific network research stream identified in the literature. The network visualization of the 81 articles in Figure 4 shows that governance networks clearly stand out as the network research stream that is utilizing SNA. Moreover, the topic of regional and community alliances is the most popular topic addressed through governance network research. Within the policy networks research stream, economic development is the most popular policy area of study. Another important finding from this network is that SNA research on emergency management is restricted to collaborative and governance networks and does not currently fall under the policy networks research stream.

OPEN IN VIEWER

Figure 4.

Network research streams and topics in network research.

A closer look at the specific topics shows that the topic in public administration that is most intensively studied through SNA has been health and human services. Among these 15 articles (18.5%), 6 are on community mental health service provider networks (Huang & Provan, 2007; Milward et al., 2009; Provan & Huang, 2012; Provan, Huang, & Milward, 2009; Provan et al., 2004; Rethemeyer, 2007). Provan and Milward’s research on collaboration networks of health service organizations has been widely recognized in the field. Besides community mental health networks, Provan, Beagles, Mercken, and Leischow (2012) have recently examined the network of a public smoking cessation program. Other topics covered under health and human services are relocation of the disabled (Poole, 2008), behavioral health of children (Bunger, 2012), federal health policy for Medicare prescription (Heaney, 2006), and public health cooperation on the U.S.–Mexico border (Collins-Dogrul, 2012). This research focuses on how health and human services are administered and managed through networks. Relevant policies and programs are studied, and their effectiveness is evaluated by the performance of service delivery networks.

The second most commonly researched topic utilizing SNA is regional and community alliances and collaboratives, with 13 articles (16%). This stream of research includes various subtopics such as interlocal agreements (ILAs; LeRoux & Carr, 2010), elite interaction and interlocal networks (Moore, Sobieraj, Whitt, Mayorova, & Beaulieu, 2002; Ruigrok, Peck, & Keller, 2006; Vidovich & Currie, 2012), community development and community collaboratives (Nowell, 2009; Pope & Lewis, 2008; Shea, 2011; Varda, 2011). Other subjects covered under this research stream include Danish mayor networks (Villadsen, 2011), NGO networks with government (Neal, 2008), European cohesion policy (EU partnership principle; Jordana, Mota, & Noferini, 2012), and local government contractual networks (Andrew, 2009).

Of the 81 articles that we reviewed, 10 focused on economic development issues. Five articles studied local economic development collaboration and networks (Feiock, Lee, & Park, 2012; Feiock, Lee, Park, & Lee, 2010; Hawkins, 2010; I. W. Lee, Feiock, & Lee, 2012; Y. Lee, Lee, & Feiock, 2012), while the rest explored subtopics of economic development, such as creation of employment avenues (Jokisaari & Vuori, 2010), regulation of markets (Fischer, Ingold, Sciarini, & Varone, 2012), innovation policy (Caloffi & Mariani, 2011; Cao & Prakash, 2011), and international sustainability networks (Zeemering, 2012).

Emergency management—in particular, disaster response networks—was discussed in 10 (12%) of the 81 articles identified as using SNA. Of the 10 articles, 6 were written by Kapucu and his colleagues. Besides discussing emergency response networks pertaining to certain events, such as 9/11 and Hurricane Katrina, some articles discussed the role of the public sector in emergency management (Kapucu & Van Wart, 2006), interstate partnerships in disasters (Kapucu, Augustin, & Garayev, 2009), emergency management at the local level (Choi & Brower, 2006), and communication networks (Kapucu, 2006a).

More recently, topics studied through SNA have become more diverse. Topics on environmental management and government climate policy are being addressed and described using a network approach. All nine articles on environmental management and climate policy using SNA were published in 2010 or later. Recent articles published in public administration journals on environmental management and climate policy are on the Swiss climate policy (Ingold, 2011; Ingold & Varone, 2012), watershed policy and partnerships (Jasny, 2012; Weible, 2011), and natural resource management (Robins et al., 2011). Other interesting and diverse topics using SNA include advocacy coalitions in carnivore management systems (Matti & Sandström, 2011), natural heritage areas (NHAs; Laven, Krymkowski, Ventriss, Manning, & Mitchell, 2010), and rural water supply and sanitation (Shrestha, 2012, 2013). Other topics identified through this research are education policy and reform (seven articles), urban planning and development (six articles), human resource management (three articles), nonprofit management (three articles), and research and development (two articles). Compared with network research in the 1990s or early 2000s, SNA research has been recently used in a much broader management and policy context.

Units of Analysis and Levels of Analysis

The unit of analysis in most of the SNA publications in public administration has predominantly been organizations and agencies. Approximately 72% of the articles that we reviewed use organizations as the unit of analysis. Some examples are disaster response agencies (Kapucu, Arslan, & Collins, 2010; Kapucu & Demiroz, 2011; Kapucu & Garayev, 2012; Kapucu & Van Wart, 2006; Marcum, Bevc, & Butts, 2012), agencies providing health and human services and serving SMI patients (Huang & Provan, 2006; Milward et al., 2009; Provan & Huang, 2012; Provan et al., 2004), and agencies and organizations working for economic development in metropolitan regions (Hawkins, 2010; I. W. Lee, Feiock, & Lee, 2012; Y. Lee, Lee, & Feiock, 2012). Among these studies, individuals or organizational representatives were often surveyed to collect organizational network data. For instance, Henry et al. (2011) surveyed key stakeholders to specify their partnerships involved in land use and transportation planning in four regions of California. Feiock et al. (2012) surveyed administrators and elected officials in four counties in Florida to understand the collaboration between city governments in economic development.

Relatively few network studies in public administration (14% of the 81 articles reviewed) have focused on individuals as the unit of analysis. Vardaman, Amis, Dyson, Wright, and Randolph (2012) surveyed public school teachers to understand how teachers’ position in a network influenced their perception of changes. Chen and Krauskopf (2012) examined how organizational merge affected the formal and informal networks within a nonprofit organization. As Isett et al. (2011) suggested, research on informal networks remains limited in existing literature. Even fewer network studies in public administration have communities or countries as the unit of analysis. There are, however, three exceptions (4%). Shrestha (2012, 2013) studied how internal and external social capital can influence the success of community water projects in Nepal. Cao and Prakash (2011) examined the policy diffusion of ISO 9000 Quality Standards across nations. The units of analysis in the remaining 10% of the 81 articles do not fall under single categories of individuals, organizations, communities, or countries. Some of the articles have mixed units of analysis, while other articles do not provide sufficient information to clearly specify their units of analysis. Overall, the number of studies with foci on individuals, communities, or nations is much smaller than the number of studies focusing on interorganizational networks.

Although the units of analysis are largely organizations, the levels of analysis can vary from individual nodes (ego), to ties between two nodes (dyadic) or three nodes (triadic), to substructures, and even to complete systems at the whole-network level of analysis (Borgatti et al., 2013). The SNA approach allows the researcher to examine the same management problem or policy issue across different levels of analysis (Provan & Milward, 2001). However, the node-level analysis (54%) and whole-network-level analysis (56%) are common in public administration to describe the power or position of an organization, to analyze interorganizational interactions, and to understand the performance, governance, and structural characteristics of a network (Kapucu & Demiroz, 2011; Provan et al., 2007; Provan & Lemaire, 2012). By contrast, triadic-level analysis and substructure-level analysis remain limited. Of the 81 articles reviewed, 17 included triadic measures in their network analysis and 23 had measures on substructures of networks. Future research can delve further into the substructures of networks to understand the subgroups within a network, the grouping of powers, or other clustering effects. Researchers can conduct more multilevel analysis of networks to fully take advantage of SNA.

Data Collection Methods

Data for SNA have been collected using both primary data collection methods (e.g., field surveys, online survey questionnaires, face-to-face structured and semistructured interviews) and secondary data (e.g., archival data from newspapers, news reports, situation reports, online company profiles, databases). The methods categories discussed below are not mutually exclusive. For instance, surveys, interviews, and documents may be used in the same article, which denotes a mixed-methods design.

Administering surveys is the most frequently used method of collecting network data in public administration research. Of 81 articles, 50 (62%) used surveys to collect network data. Twenty-nine articles (38%) used secondary data collection methods that involved archival data. Content and document analyses were conducted to analyze newspapers and situation reports, annual company reports, and government reports. Of the 29 articles that used archival data, 16 also used primary methods of data collection, such as surveys, questionnaires, and interviews, to obtain additional information on the networks being studied. In many cases, secondary data are used only as a means to develop a list of network actors to be surveyed or interviewed as part of the methodology. Of the 81 articles, 43 (53%) collected network data from interviews, 9 of which were specified as semistructured interviews. Interviews are commonly applied as a supportive method to achieve triangulation in network research or to get qualitative data and rich information about networks. Other data collection methods are rarely used. Four articles used observational data, and only one article used focus groups for data collection.

Overall, there is an increasing use of mixed methods of data collection. Of the 81 articles, 28 (35%) used both quantitative and qualitative methods to collect network data. However, the majority of articles relied mostly on survey data or qualitative interviews. Studies relying heavily on survey data require participants to recall information pertaining to networks and interactions before a policy change, an emergency, or the implementation of a program. As a consequence, researchers may face the challenges of selection bias and the internal validity of the results. On the other side, collecting secondary data and archival data may help produce less biased data; however, the challenge of getting information on the complete network datasets is difficult. Hence, more future research should consider integrating quantitative approaches with qualitative approaches. Another challenge in the field is to study the sustainability, maturation, and evolution of networks. Systematic longitudinal designs and analysis remain largely missing. These findings lend support to previous research that identified the need to integrate quantitative and qualitative designs in network research and to conduct more longitudinal analysis of network change and evolution (Berry et al., 2004; Provan et al., 2007; Provan & Lemaire, 2012). As the field continues to grow, more diverse approaches to collect network data are needed to further advance network research in public administration.

Key Measures and Analytical Methods

The content analysis further identified the key network measures used for network analysis. Most articles used both whole-network measures, such as density and centralization, along with node-level measures, such as degree centrality and betweenness centrality (see Borgatti et al., 2013, or Scott, 2013, for more information about these measures and advanced analytic techniques). Although most articles used measures that applied to both the node-level analysis and whole-level analysis, many authors did not clearly spell out the level at which analysis was carried out or the rationale behind using measures that belong to different levels of analysis.

The use of centrality measures is common in social network research in public administration. Of the 81 articles, 52 (65%) used some measures of centrality, the most commonly used being degree centrality, betweenness centrality, and eigenvector centrality. Degree centrality and eigenvector centrality measures are used to measure the position, power, or resource accessibility of individual nodes within networks (e.g., Choi & Brower, 2006; Huang & Provan, 2007). Betweenness centrality is used to understand how a node influences the flow of resources or information within networks (e.g., Ansell, Reckhow, & Kelly, 2009) or the brokering role of individual nodes (e.g., Ingold, 2011; Ingold & Varone, 2012; Jokisaari & Vuori, 2010). The second most commonly applied measure was network density (a measure of whole networks), which is found in 32 (40%) of the articles. A total of 15 (19%) articles also used network centralization. Network density and centralization are used to measure the cohesion or connectedness of networks (Borgatti et al., 2013) and describe the structural characteristics of networks (e.g., Milward et al., 2009).

Group-level network measures were applied to a limited extent. A few articles applied triad network measures (e.g., Andrew, 2009; Feiock et al., 2012; Feiock et al., 2010; Fischer et al., 2012; Henry, Lubell, & McCoy, 2012; Y. Lee, Lee, & Feiock, 2012; Oliver & Montgomery, 2008; Varda, 2011). The key measure used for analyzing triads is transitivity effect (for directed data, transitivity measures whether the connection between Nodes A and B and the connection between Nodes B and C can lead to a connection between Nodes A and C). Analysis of the subgroups in existing network studies has been limited. Three articles included clique analysis, and three included cluster analysis. Kapucu et al. (2010) and Kapucu et al. (2009) applied clique analysis to identify the subset of organizations that work closely within disaster response networks. Ansell et al. (2009) conducted clique analysis to identify the key stakeholder groups in the context of urban school reform. Cluster analysis is also used to classify communities into different groups according to the “similarity of network connections” (Shrestha, 2012, p. 15) and to cluster the stakeholders of environmental management into different groups based on their beliefs (Weible, 2011).

Another important measure for analyzing the network position and network substructures is structural equivalence. Structural equivalence examines “the direct connections of an actor to others in the network” (Borgatti et al., 2013, p. 207). For undirected data, two nodes are structurally equivalent when they are connected with the same other actors. For directed data, two nodes are structurally equivalent when they send ties to the same actors and receive ties from the same actors in the network (Borgatti et al., 2013). Eight articles applied structural equivalence/blockmodeling using CONCOR to understand the roles played by different stakeholder groups, to measure trade competition, to identify advocacy coalitions, and to evaluate the impact of similar network positions on the adoption of a new program, cooperative behavior, and individual perception about victimization at the workplace (e.g., Cao & Prakash, 2011; Ingold, 2011; Ingold & Varone, 2012; Jokisaari & Vuori, 2010; Lamertz & Aquino, 2004; Lambright, Mischen, & Laramee, 2010; Rethemeyer, 2007; Robins et al., 2011).

Besides descriptive network measures, researchers have begun to use more advanced statistical analysis techniques such as Quadratic Assignment Procedure (QAP; for example, Bunger, 2012; Chen & Krauskopf, 2012), Exponential Random Graph Modeling (ERGM; for example, Feiock et al., 2010; Robins et al., 2011), and Stochastic Actor-Oriented Models (SAOMs; for example, Andrew, 2009; Fischer et al., 2012). Table 2 provides a detailed list of research topics studied using these inferential network analysis models. Advanced SNA techniques such as QAP were used in nine articles that compare and analyze the relationships between different network datasets (e.g., Bunger, 2012; Chen & Krauskopf, 2012; Jasny, 2012; Lamertz & Aquino, 2004; Y. Lee, Lee, & Feiock, 2012; LeRoux & Carr, 2010; Matti & Sandström, 2011; Provan & Huang, 2012; Stephens, Fulk, & Monge, 2009). Based on the product–moment correlations of random permutations on the rows and columns of the network matrices, QAP can correct for autocorrelation (Borgatti, Everett, & Freeman, 2002). Multiple regression with QAP (MRQAP) enables researchers to test the level of association between interdependent network variables (Krackhardt, 1988; van Duijn & Huisman, 2011).

OPEN IN VIEWER

Table 2.

Statistical Network Analysis Models.

SNA models	Topics
QAP: Examine the correlation between two network matrices through a random permutation approach (Borgatti, Everett, & Johnson, 2013).	The relationships between perceived trustworthiness, competition, and administrative cooperation among nonprofit organizations (Bunger, 2012)
	Relationship between the workflow network, problem-solving network, mentoring friendship network, and socio-emotional support network (Chen & Krauskopf, 2012)
	The relationships between donation behavior and congressional voting, between organizational type and participation in collaboration forums (Jasny, 2012)
	The influence of informal and formal status on victimization perception (Lamertz & Aquino, 2004)
	The impact of perceived competitive or cooperative relationships on the formation of informal policy networks for economic development (Y. Lee, Lee, & Feiock, 2012)
	The relationships between interpersonal networks of elected officials and public administrators, and the formation of interlocal agreements (LeRoux & Carr, 2010)
	The influence of perceived influence and perceived belief on the formation advocacy coalitions in the context of environment management (Matti & Sandström, 2011)
	Whether the ties between health and human service organizations for information sharing, contracts, influence, service referrals, and reputation change over time (Provan & Huang, 2012)
	The relationship between interpersonal communication and interpersonal trust, and the interorganizational partnership building (Stephens, Fulk, & Monge, 2009)
ERGM: Predict or explain the existence of a network tie by modeling dependence among network ties and examining the microconfigurations of a network (Lusher, Koskinen, & Robins, 2012; Robins, Bates, & Pattison, 2011).	The impact of actor attributes and microconfigurations on collaboration in governmental innovation (Robins et al., 2011)
	Predict the emergence of collaboration ties in the context of economic development based on the substructural characteristics of the network (Feiock, Lee, Park, & Lee, 2010)
	The influence of belief clusters and social capital on collaboration in regional land use and transportation planning (Henry et al., 2010)
	How the reciprocity and social clustering structures influence the formation of collaboration ties among local governments to address economic development issues (I. W. Lee, Feiock, & Lee, 2012)
	The influence of the microconfigurations on the formation of collaboration ties in managing national heritage (Laven, Krymkowski, Ventriss, Manning, & Mitchell, 2010)
	The influence of network density on the participation in collaboration forums (Jasny, 2012)
	The influence of the structural configurations of the network on collaboration in the context of adult basic education policy networks (Park & Rethemeyer, 2012)
SAOMs: Explain the change of network ties and evolution of network structures over time (Snijders, van de Bunt, & Steglich, 2010).	The change of contractual ties between local governments for service integration over time (Andrew, 2009)
	The evolution and structural changes of policy networks in the Swiss telecommunication industry over two decades (Fischer, Ingold, Sciarini, & Varone, 2012)

Note. QAP = quadratic assignment procedure; ERGM = exponential random graph modeling; SAOMs = stochastic actor-oriented models.

ERGM was used in seven articles to study the emergence of network ties or the stochastic process of network formation by examining the microconfigurations or structural characteristics of a network, such as reciprocal ties, in two-stars, out two-stars, two paths, three-cycles, and transitive triplets (e.g., Feiock et al., 2010; Henry et al., 2011; Jasny, 2012; Laven et al., 2010; I. W. Lee, Feiock, & Lee, 2012; Park & Rethemeyer, 2014; Robins et al., 2011). SAOMs are developed for analyzing longitudinal network data and explaining the evolution of network structures over time (Lubell, Scholz, Berardo, & Robins, 2012; Snijders, van de Bunt, & Steglich, 2010). Our analysis suggests that advanced SNA methods are mostly utilized in the study of policy and governance networks and are underutilized in the study of implementation or collaborative networks.

Recent years have seen a gradual movement from simple descriptive network analysis and visual mapping to more inferential analysis and theory testing. The increasing use of advanced analysis techniques implies that social network research in public administration is moving beyond the early stage that uses simple measures and analysis techniques to describe characteristics of nodes, subgroups, or networks. In fact, with recent advancement in statistical network analysis, multivariate techniques such as QAP, ERGM, and SAOMs have been built into the SNA package (Borgatti et al., 2013). With these tools, researchers can study the relationships between multiple network matrices, examine the influence of endogenous network structures and microconfigurations on the formation of networks, and explore the evolution of network structures. Researchers can conduct more theory-driven explanatory network research as well as comparative and longitudinal network research.

SNA Software Programs

UCINET (Borgatti, Everett, & Freeman, 2002) is the most widely used software for SNA in public administration. This is consistent with the finding of Knoke and Yang’s (2008) research that suggested UCINET, with its frequently updated and improved versions, is the most popular software for network analysis. Our results show that, of the 81 SNA articles short-listed, 50 (62%) used UCINET. Seventeen articles (21%) did not mention the software used in carrying out SNA. Two articles used the software VISONE (Brandes & Wagner, 2004). This software is used for interactive visualization of networks and sociograms and for analyzing whole networks. For visualization, the more common approach in the majority of the articles was to use the NetDraw subroutine in UCINET.

Three articles used the StOCNET software package for advanced statistical analysis of networks. This software is a comprehensive tool and includes higher level analysis for structural equivalence/blockmodeling using the BLOCKS and SIENA routine. SIENA (merged into R) is known as the analysis of repeated measures on networks and uses the same process as ERGMs. This package was developed by Stokman, Snijders, and van Duijn at the University of Groningen in the Netherlands (Ingold & Varone, 2012; Ripley & Snijders, 2011). SIENA is sophisticated software that allows multiple waves of longitudinal data in analyzing networks (Ripley & Snijders, 2011; Snijders, 2005). The development of these various software programs has allowed researchers to further explore social network research in public administration.