Structural Development Of Public Management
This chapter describes the concepts and theories that lay the foundation for a theory of structural development of public management networks. This proposed theory is represented by a set of structural hypotheses describing the development of public management networks over time. These hypotheses were developed using a multitheory mutlilevel modeling approach (Contractor, Wasserman, & Faust, 2006) that incorporates relevant sociological, organization science and public management theories. This theory will then be tested using the techniques of confirmatory network analysis that include random graph models, such as Markov random graph models (Frank & Strauss, 1986; A. Strauss & Ikeda, 1990) and the p* family of models (Anderson, Wasserman, & Crouch, 1999; Pattison & Robins, 2002; Pattison & Wasserman, 1999; Robins, Elliott, & Pattison, 2001; Wasserman & Pattison, 1996).
Most of the previously mentioned studies (Isett & Provan, 2005; Milward, Provan, Fish, Isett, & Huang, 2010; Provan, Huang, & Milward, 2009; Provan, Isett, & Milward, 2004) used to understand the development of public management networks suffer from a one level-one theory problem. In other words, these studies are focused on one level of the network and employ one established theory normally derived from sociology or organizational theory. However, public management networks are relational systems that should be studied at the different levels (individual actor, dyadic, triadic and the whole network) via the lens of the various theories that explain the emergence of public management networks.
Having reviewed the literature on communication, organizational and interorganizational networks, Monge and Contractor (2003) spotted nine groups of theories explaining the emergence, development, management and sustainability of organizational networks. They include:
1) Theories of self-interest
2) Theories of mutual interest and collective action
3) Contagion theories
4) Cognitive theories
5) Cognitive consistency theories
6) Exchange and dependency theories
7) Homophily theories
8) Proximity theories
9) Evolutionary and coevolutionary theories
Each of these theories, along with their mechanisms and implications for network development and management are presented in Table 1. A researcher studying public management networks in particular cases can employ diverse social theories that use similar theoretical mechanisms to get similar explanations of network development but at completely different levels of social network analysis, i.e., at the individual actor, dyadic, triadic and whole network or global level (Contractor et al., 2006).#p#分页标题#e#
Given the limited development of theories about the evolutionary development of public management networks and their scarce empirical base, it is important to remember that "different theoretical mechanisms sometimes offer complementary as well as contradictory explanations at the same level of analysis" (Contractor et al., 2006, p. 683). However, these theories are an abundant source that can help public management scholars understand the structural development of collaborative and public management networks of all types, as identified by Agranoff (2007).
Individual value maximization
Investments in opportunities
Control of information flow
Mutual self-interest and collective action theories
Joint value maximization
Inducements to contribute
Number of people with resources and interests
Cognitive mechanisms leading to
Cognitive social structures
Similarity in perceptual structures
Cognitive consistency theories
Choices based on consistency
Drive to avoid imbalance and restore balance
Drive to reduce dissonance
Exposure to contact leading to
Social information processing
Structural theory of action
Similar positions in structure and roles
Exchange and dependence theories
Exchange of valued resources
Equality of exchange
Inequality of exchange
Complex calculi for balance
Choices based on similarity
Choose comparable others
Choose based on own group identity
Choices based on proximity
Influence of distance
Influence of accessibility
Network evolution and coevolution theories
Variation, selection, retention
Competition for scarce resources
Complex adaptive systems
Network density and complexity#p#分页标题#e#
Source: Contractor, N. S., Wasserman, S., & Faust, K. (2006). Testing multitheoretical, multilevel hypotheses about organizational networks: an analytic framework and empirical example. Academy of Management Review, 31(3), 681-703.
These social theories along with the statistical advancements of social network analysis allow one to develop testable structural hypotheses about the development of public management networks. A structural hypothesis or "structural signature" is understood here as a pattern of interactions among network members occurring in a network (DeLeon & Varda, 2009; Monge & Contractor, 2003). For instance, let's hypothesize that during the planning stage, public management networks tend to be less centralized than they are at the implementation stage, so that more voices can be heard in the development of the norms and values that will allow for effective functioning in the future. Then, one can collect data on one particular public management network and test this structural hypothesis by using the techniques of confirmatory network analysis (Wasserman & Robins, 2005). If our structural hypothesis is verified we can conclude that other public management networks of a similar type will bear similar structural characteristics.
Structural hypotheses at the different levels allow for testing the structural tendencies of a network at the different stages of network development. These tendencies reflect the structural properties at different levels that can be linked to sociological, organization science and public management theories, and can be defined by both endogenous and exogenous variables (Contractor et al., 2006). Endogenous variables are those "various relational properties of the focal network itself that influence the probability ties will be present or absent in the same network" (p. 686). They indicate that a particular network configuration in the observed network has a structural tendency proposed in the structural hypothesis. The traditional examples of specific measures of the endogenous variables at the different levels of analysis include actor structural autonomy (actor level), reciprocation (dyadic level), transitivity (triadic level), cyclicality (triadic level) and network centralization (global level). Exogenous variables are those "various properties outside the focal network that influence the probability ties will be present or absent in the focal network" (Contractor et al., 2006, p. 687). They can include the individual characteristics of the network actor such as gender, socioeconomic status, educational attainment, occupation, organizational affiliation or different network relations such as differential reciprocity (dyadic level), differential transitivity (triadic level), differential cyclicality (triadic level), and differential network centralization (global level). Each of these endogenous or exogenous variables can be linked to one of the nine groups of theories mentioned above that explain the formation, development and maintenance of the network.#p#分页标题#e#
Structural hypotheses of network development at the global level 全球一级网络发展的结构假设
Structural hypotheses at the global level allow for testing the structural tendencies of a network at different stages of development at the level of the whole network. These tendencies reflect structural properties at the network level that can be linked to sociological, organization science and public management theories. From a social network point of view, these tendencies can be explained using traditional measures of networks such as endogenous variables, including centrality, density, and transitivity.
Centrality. Centrality in public management networks can be interpreted differently depending on the measure of centrality and the theory associated with this measure. The most popular measure of centrality is degree centrality, which can be associated with degree of power distribution or power structure in policy and public management networks (DeLeon & Varda, 2009). The various stages of network development require different power structure degrees. At the planning stage, public management networks should be characterized by low centrality, so that few network actors hold central positions. From the structural point of view, this indicates a lower ability of particular network actors to have control and influence over the network (DeLeon & Varda, 2009) and a diminished ability to form bridges and fill in structural holes (Burt, 1992; Monge et al., 1998). Centrality also indicates the degree of connectedness. It is expected that the planning stage will have lower scores of network centralization, which suggests that network actors share equal connections and are able to engage in mobilizing activities (Laumann, Knoke, & Kim, 1985; Laumann & Pappi, 1976) and to act in the name of promoting the collective good (Marwell & Oliver, 1993; Marwell, Oliver, & Prahl, 1988). On the other hand, studies by Provan and Milward (1995) and Provan, Isett, & Milward (2004) suggest that the implementation stage is characterized by a high degree of centralization.  Based on these studies, the following hypothesis is proposed for the theoretical framework for understanding structural development of public management networks over time:
Density is another commonly used network property that measures the number of existing ties between the network actors compared to the maximum number of ties possible among these network actors (Wasserman & Faust, 1994). The role of density in the evolution of public management networks is still unclear, although the Provan and Sebastian (1998) study of networks of mental health agencies suggests that an increase in density is negatively associated with public management network effectiveness. But it may be concluded that in the evolution of public management networks the number of ties increases over time. Provan, Isett, & Milward (2004) concluded that collaboration was accompanied by number of ties and strengths of multiplexity of the ties over time. Therefore, it may be reasonably concluded that there will fewer connections at the planning stage than at the implementation stage. As I mentioned in chapter 2, there is a positive correlation between cohesion and density. In other words, the higher the density of a network, the more cohesive is the network. High cohesiveness is important for the normal functioning of a public management network, since high cohesiveness is associated with the presence of rules governing the behavior of network members. Since the rules of operation and interactions are primarily developed in the planning stage, I can conclude that the implementation stage will be characterized by a higher cohesiveness than that of the previous network development stage. Consequently, one can conclude that the density of a public management network will be higher at the implementation stage than at the planning stage, which can be summarized neatly in the following hypothesis for a theoretical framework for understanding the structural development of public management networks over time:#p#分页标题#e#
Hypothesis 2: As public management networks develop over time, network density tends to increase.
Transitivity is another useful structural property at the global level that can explain whether the level of hierarchy is linked theoretically to balance theory, originally developed as a theory of cognitive consistency (Heider, 1958). The interpretation of transitivity depends on the type of network. If the network represents friendship relations, then transitivity points at the consistency of choices in selecting friends. If the network is organizational or inter-organizational, then transitivity points at the degree of hierarchy in the organization (Contractor et al., 2006)
According to the balance theory, there is a patterned relationship between reciprocity and transitivity. If an organizational or interorganizational network exhibits a propensity for low reciprocity accompanied by a high transitivity, this network may be hierarchical due to the presence of the stars (Kilduff & Tsai, 2003). It is widely believed that networks are flat and that every network member enjoys the same rights and powers within the domain of that network. However, this is far from the reality that is observed by empirical studies of public management networks. In fact, public management networks are characterized by "a more interesting and 'hilly' social terrain that includes aspects of hierarchy" (Rethemeyer & Hatmaker, 2008, p. 636). The initial stages of network development require some sort of hierarchy, where the scalar vector of power is clear, with distinct power differentials between network members. The distribution of power follows the principles of hierarchical and managerial accountability necessary for the processes of planning. During the implementation stage, the power differentials between network members are expected to be leveled off since the power in the network should be shared by several centers: one center responsible for governing or "managing across" networks and another center responsible for "managing within" networks (Perri, 2006). Therefore, the following hypothesis is proposed for the theoretical framework for understanding structural development of public management networks over time:
Hypothesis 3a: As public management networks develop over time, they tend to become less hierarchical (transitivity).
Structural hypotheses at the triadic level 在三级结构假设
Structural hypotheses at the triadic level are derived theoretically from triadic census analysis and represent the endogenous variables at the triadic level such as transitivity, cyclicity and balance. These hypotheses can be tested by the means of triadic census analysis (Wasserman & Faust, 1994) and later by p* models (Anderson et al., 1999; Pattison & Robins, 2002; Pattison & Wasserman, 1999; Robins et al., 2001; Wasserman & Pattison, 1996). As I mentioned in chapter 2, triadic census investigates the various triadic network configurations that influence the probability that ties will be present or absent in a network (Contractor et al., 2006). Although social network analysists examine the sixteen components, or isomorphic classes, of triads, the focus here will be on hierachical, cyclical and fully reciprocated triads in order to develop the structural hypotheses of network development at the triadic level.#p#分页标题#e#
Hierarchical triads. Hierarchical triads are transitive triads, which are triads with ties from A to B, from B to C and from A to C. As I discussed earlier, the interpretation of transitivity is contingent on the type of relations in the network (Contractor et al., 2006). The presence of transitive triads in the affective or friendship network (when relations are friendly or sentimental) point to the consistency of relations (Heider, 1958; Holland & Leinhardt, 1975, 1981). In other words, a friend of your friend is supposed to be your friend as well, and we should definitely like the friends of our friends since we already like the latter. When the concept of transitive triads is applied to organizational networks, however, the presence of transitive triads in formal relations or in the transfer of know-how undoubtedly indicates the probability of hierarchy occurring in the network (Contractor et al., 2006). For example, Rank, Robins et al. (2010) assert that "the internal structure of a team is likely to be characterized by strong hierarchical elements" (p. 748). This hierarchy is based on various aspects of teamwork such as division of labor, assigning responsibility and authority in making decisions (Robbins & Judge, 2009). Rank, Robins et al. (2010) argue that hierarchical positions found in the organizational network need not be the same as traditional leadership roles. Informal leaders may develop as a result of expert knowledge or by applying the skills of politicking. In this case transitive triads are represented by triads with nodes sending links or ties to one node, and where this node enjoys the position of power. Of particular interest in this case are the hierarchical triads 120C and 210. The graphic representation of triads 120C and 210 can be found in Figure 2.
The structural hypothesis about transitive triads in the developmental perspective is related to the degree of hierarchy in the different stages of development. As in the case of transitivity at the global level, a higher degree of hierarchy should be exhibited at the planning stage than at the implementation stage. This can be explained by the necessities of establishing basic network structure and developing norms and rules. The structural hypothesis about transitive triads "would be supported if there were greater probabilities for graph realizations in which triads of actors in the network exhibited a high degree of transitivity" in the stage of planning (Contractor et al., 2006, p. 689). It can be succinctly written in the following hypothesis:
Hypothesis 3b: As public management networks develop over time, they tend to become less hierarchical (hierarchical triads).
Cyclical triads. Cyclical triads are triads with ties connecting the network actors in a clockwise order, an example of which is shown in Figure 2.
The presence of cyclical triads can be explained by the theory of generalized exchange (Bearman, 1997; Yamagishi & Cook, 1993). According to the theory of indirect generalized exchange, network members form relations where "each participant provides benefits to an actor in the network who does not return benefits directly to that participant" (Yamagishi & Cook, 1993, p. 237). In other words, network actors form cyclical triads, where network actor A benefits network actor B, network actor B benefits network actor C, and network actor C benefits network actor A, often exchanging resources by means of providing information, receiving information, or giving advice. And these cyclical triads may be developed around multiple relations in multirelational networks (Rank et al., 2010).#p#分页标题#e#
The role of indirect generalized exchange at the different stages of network development should be not underestimated. The research on organizational networks shows that the planning stage of network development is usually characterized by direct exchange of resources. This direct exchange implies the development of strategic norms for interactions "among participants that govern the level of mutual contributions and benefits of actors" (Rank et al., 2010, p. 758). The implementation stage, on the contrary, is characterized by indirect generalized exchange where network participants are engaged in the exchange of resources by giving and receiving information, financial support, and many other things. Therefore, the following hypothesis is proposed for the theoretical framework for understanding the structural development of public management networks over time:
Hypothesis 4: As public management networks develop over time, the flow of resources in the network increases (more cyclical triads tend to occur).
Balanced triads. As I discussed earlier, balance theory states that people prefer balanced ties in their friendship networks (Heider, 1958). They insist on mutuality and transitivity in their friendship relations, so that their affection is reciprocated and their friends are friends with others. Structurally, balance in a public management network is assessed in terms of the relationship between reciprocity and transitivity (Prell, 2012). The extremes of the combinations of reciprocity and transitivity could be detrimental for the development of the network at some stages. For example, a network with high transitivity and low reciprocity tends to be hierarchical because of the stars, who receive so many asymmetrical ties (Kilduff & Tsai, 2003). At the other extreme, a network with high reciprocity and low transitivity has a propensity to be highly decentralized, with an absence of stars and few cliques. It also has a tendency to be less effective than networks with a lower reciprocity (Provan & Sebastian, 1998). Another example would be the case of a network divided into two subnetworks that are balanced with each other. This balance is achieved by all positive linkages in one subnetwork and all negative linkages in the other subnetwork (Cartwright & Harary, 1956). In short, balance theory seeks a golden middle between reciprocity and transitivity or positive and negative ties. A graphical representation of balanced triads (102 and 300) can be found in Figure 4.
From the evolutionary point of view, balance in public management networks should vary depending on the stage of network development. It may be reasonably assumed that the first stage of development, or planning stage, is critical in terms of sustainability and vulnerability of the network. On the one hand, there should be several network centers, responsible for inviting new members into the network and developing norms and rules for the interactions of network actors. Meanwhile, the relationships of the network actors leading the formation of the network should be fully reciprocated so that strategic rules of interactions can be developed (Rank et al., 2010). Later, when the norms and rules of a public management network are developed and several centers that govern and manage the network are developed, the role of balanced triads, especially mutually reciprocated triads, is diminished. Therefore, the following hypothesis is proposed for the theoretical framework for understanding structural development of public management networks over time:#p#分页标题#e#
Hypothesis 5: As public management networks develop over time, they tend to be less balanced and stable (fewer fully reciprocated triads tend to occur).
Structural hypothesis at the dyadic level 二元结构假说
Structural hypotheses at the dyadic level are derived theoretically from the dyadic census analysis and represent the endogenous variables at the dyadic level (reciprocity). These hypotheses can be tested initially by measuring the reciprocity at the global and individual network actor levels. The final testing of the structural hypotheses at the dyadic level will be conducted by p* models (Anderson et al., 1999; Pattison & Robins, 2002; Pattison & Wasserman, 1999; Robins et al., 2001; Wasserman & Pattison, 1996)
Reciprocity, another important structural characteristic of a network, signifies the development of trust, mutual support and exchange of resources among network participants. As mentioned earlier, reciprocity is defined by the number of symmetric ties among network actors. The decision to form or join a network, according to social exchange theory, is made by calculating the number and type of resource exchanges (Blau, 1964; Homans, 1950, 1974). As deLeon and Varda (2009) argue, "Each actor in a network must see that he or she will not only benefit by collaboration but also that the overarching goal will be better achieved by working with other stakeholders" (p. 67)
The growth and sustainability of a public management network is conditional on reciprocity in the exchange of resources (Willer & John, 1997). Reciprocity is important at the planning stage since network members are engaged in the processes of mobilizing different resources and actors and setting rules, norms and values for the effective future functioning of the network. Or as Rank et al. (2010) state, "The strong preference for the formation of mutual exchange relationships among pairs of actors suggests that within the strategic decision process reciprocity norms have developed among participants that govern the level of mutual contributions and benefits of actors" (p. 758). Subsequently, at the implementation stage, the number of mutual ties will decrease since strategic actions and trust have been developed. Therefore, the following hypothesis is proposed for the theoretical framework for understanding the structural development of public management networks over time:
Hypothesis 6: As public management networks develop over time, reciprocity of ties tends to decrease.
Structural hypotheses at the individual level 个体层次结构假说
Structural hypotheses at the individual level allow one to test the structural tendencies of a network at different stages of development at the level of the individual network actor. These tendencies reflect the structural properties at the individual level that can be linked to sociological, organization science and public management theories. The structural tendencies can be defined by both endogenous and exogenous variables prescribed by multilevel, multitheory modeling (Contractor et al., 2006). Although the endogenous variables at the actor level include different measures of centrality such as degree centrality, closeness, betweenness and prestige, the structural autonomy of the network actor was selected for developing a structural hypothesis instead, since some measures of centrality were indirectly taken into account at the different levels of p* models.#p#分页标题#e#
The structural autonomy of a network actor is based on the concept of structural holes (Burt, 1992), which are represented by network actors currently without any connections. As I discussed in chapter 2, these structural holes can be "bridged" by an entrepreneurial network actor who connects the unconnected parts of the network, including subnetworks, cliques and unconnected individuals, and gain great structural advantage by doing so.
The structural autonomy of a network actor also has implications for network development. It can be assumed that network champions target the most popular network actors in the network during the planning stage via the process of activation (Agranoff & McGuire, 2001), which leads to low structural autonomy and subsequently to a low level of bridging. At this point of network development, in order to pursue the purposes of mobilizing and activation, it would be most beneficial to target network actors with many linkages (Agranoff & McGuire, 2001). These actors maintain more relations with the various network actors and are capable of mobilizing human, informational, political and material resources when needed. The implementation stage, however, sets new conditions. To ensure access to the untapped resources of various network members it is expedient to forge relations with network actors who have not been connected to anybody in the network at the previous stages rather than to popular "stars" who receive many nominations from other network members. In this way network leaders or network champions can exploit the resources, knowledge and innovations of the network actors who currently have no connections in the network. From the probabilistic point of view, the inclination to forge a tie or connection with currently isolated network actors without ties is less likely at the planning stage. One can then conclude that bridging tends to increase during the process of network development and can be measured by betweenness at the global and two-path network configurations (Snijders, Pattison, Robins, & Handcock, 2006). Therefore, the following hypothesis is proposed for the theoretical framework for understanding the structural development of public management networks over time:
Hypothesis 7: As public management networks develop over time, bridging tends to increase (betweenness and two-path configurations).
Structural hypotheses based on the exogenous variables at the individual actor level allow one to see the effect of the individual characteristics of the network actor such as gender, sector affiliation and interorganizational network experience on the probability of forming a tie at the different stages of network development. The most appropriate theory to be used in this instance is that of homophily (Coleman, 1957; Ibarra, 1992, 1993, 1995, 1997). This theory suggests that network actors tend to form ties with those actors who have similar characteristics. It may be expected that the effect of similarity on the probability of forming a tie may change over time, with the effect of homophily being stronger at some stages of network development than at others. Provided below is the rationale for structural hypotheses based on actor similarity in gender, sector affiliation and interorganizational network experience.#p#分页标题#e#
As I discussed in chapter 2, numerous studies suggest that gender significantly predicts how network actors form ties in friendship and communication networks (Brass, 1985; Ibarra, 1992, 1993, 1995, 1997). In short, male networks tend to be detrimental for women and beneficial to men because of their tendency to be homophilous (Ibarra, 1993, 1997), while women are inclined to form differentiated networks (Ibarra, 1992).
Gender similarity as a predictor of forming ties certainly has implications for network development. As a network forms, it can be expected that males have a greater propensity to form ties with males than with females. The dominant position of men in various organizations forces women to form ties with men, who tend to be in the centers of the network in the initial stages. It has been argued that centrality in the network provides the access to resources necessary for effective performance of the activities in the interorganizational networks (Miller, Lincoln, & Olson, 1981). Therefore, women tend to form more relations with the men positioned in the center of the network (Brass, Galaskiewicz, Greve, & Tsai, 2004). However, the process of implementation changes the pattern of interactions between men and women in terms of homophily. Interorganizational collaboration at this stage is more likely to occur if the network actors share status and power (Emerson, 1962; Ring & van de Ven, 1992). This implies that the power differentials between men and women should be reduced, and neither group should tend toward homophily in collaborative relations. Therefore, the following hypothesis is proposed for the theoretical framework for understanding structural development of public management networks over time:
Hypothesis 8: As public management networks develop over time, gender similarity has less effect on the propensity to form ties.
Interorganizational network experience has some bearing on the formation of ties, especially in the early stages of network development. Research on organizational and interorganizational networks suggests that interorganizational network experience has some effect on the tendency to form and diversify ties (Ahuja, 2000; Brass et al., 2004; Powell, Koput, & Smith-Doerr, 1996). Powell, Koput, and Smith-Doerr (1996), in their study of biotechnology firms, found a positive relationship between the networking experience and centrality in interorganizational networks. They also report that networking experience allows one to make more diverse connections and enriches the actor's knowledge base about his or her industry or area of work. As network actors become more skilled in interorganizational collaboration, they learn more about different ways to interact with diverse stakeholders, with the result that there is a higher probability of forming a tie. Taking these results to the organizational level, this networking experience not only results in greater knowledge of the industry but makes organizations more attractive to other organizations (Ahuja, 2000). Therefore, it is crucial to have more persons with interorganizational network experience in the planning stage since this facilitates the process of activation by reaching those potential and diverse network actors who have the necessary resources for forming the network (Agranoff & McGuire, 2001). In addition, interorganizational network experience is similar to what Senge (1990) calls the "learning organization," an organization "where people continually expand their capacity to create the results they truly desire, where new and expansive patterns of thinking are nurtured, where collective aspiration is set free, and where people are continually learning how to learn together" (p.3). This collaborative expertise is vital for a newly formed public management network, where participants have to develop the strategic norms for interactions in the later stages of development.#p#分页标题#e#
As a public management network evolves, interorganizational network experience becomes less useful and has less of an effect on the tendency of network members to forge ties. The processes of targeting network actors with generous financial, human, and informational resources are transformed into the processes for finding implementation partners or even for deactivating actors no longer needed at this stage (Agranoff & McGuire, 2001).
Thus, it can reasonably be expected that individuals with interorganizational network experience will be more inclined to form a tie in the planning stage of network development. It must also be assumed that a similarity in interorganizational network experience also results in the higher propensity to form a tie with other actors who have interorganizational experience. Therefore, the following hypothesis is proposed for the theoretical framework for understanding structural development of public management networks over time:
Hypothesis 9: As public management networks develop over time, interorganizational network experience has less effect on establishing relations.
Sector similarity is another exogenous variable at the actor level, and it too has an effect on the propensity to form ties within a sector according to the tenets of homophily theory (McPherson, Smith-Lovin, & Cook, 2001). Several studies have found that representatives of organizations from the same sector have a higher propensity to form ties with each other in interorganizational networks (Contractor et al., 2006; Shrestha & Feiock, 2009). Contractor et al. (2006) found that government organizations in the Cooperative Research and Development Agreement (CRADA) network were more likely to form ties with organizations of the same organizational type compared to interactions between government and private organizations. Public and nonprofit organizations are more likely to form relations with each other because of commonly shared attributes such as orientation on the public value, public interest, mission and stakeholders (Moore, 2000).
When networks are formed, sector similarity allows organizations of a similar type to take advantage of commonly shared routines and values for establishing new norms and values during the process of framing (Agranoff & McGuire, 2001). The implementation stage, however, returns representatives of organizations from the different sectors (public, private and nonprofit) to the modus operandi prescribed by each organizational type. It is expected then that sector identification and the subsequent organizational-type homophily prevail at the planning stage, thus making organizations of a given sector form ties within that sector. However, this conclusion will no longer hold true at the implementation stage. Therefore, the following hypothesis is proposed for the theoretical framework for understanding structural development of public management networks over time:#p#分页标题#e#
Hypothesis 10: As public management networks develop over time, sector similarity has less of an effect on establishing relations in the network.
This chapter describes the theory of structural development of public management networks represented by a set of ten structural hypotheses describing network development over time. These hypotheses were developed using a multitheory mutlilevel modeling approach that incorporates relevant sociological, organization science and public management theories in order to understand the structural development of public management networks over time. Each of these structural hypotheses reflects the changes in the patterns of interactions among network actors (structural tendencies) depending on the stage of network development.
These structural hypotheses will be tested by two types of social network analysis: exploratory network analysis (Nooy, Mrvar, & Batagelj, 2005) and confirmatory network analysis (Contractor et al., 2006). Exploratory network analysis is employed to highlight the structural properties of the Metro School network at each stage of development as well as to test the structural hypotheses at the global, dyadic and individual levels of network on a preliminary basis. The final testing of the structural hypotheses of structural development of public management networks will be performed by using the techniques of confirmatory network analysis, which include triadic census analysis (Wasserman & Faust, 1994) and exponential random graph modeling (ERGM), particularly p* models (Anderson et al., 1999; Pattison & Robins, 2002; Pattison & Wasserman, 1999; Robins et al., 2001; Wasserman & Pattison, 1996). Triadic census analysis is employed for testing structural hypotheses associated with balanced, transitive and cyclical triads. p* models are used for testing the structural hypotheses associated with such structural aspects of social and organizational networks as degree centrality, betweenness, reciprocity, density and structural autonomy as well as the effect of the individual characteristics of a network actor such as gender, sector affiliation and interorganizational network experience on the probability of establishing a tie at a particular stage of network development.
The next chapter covers the research design and methodology of the study. The logic for using a mixed methods design along with the longitudinal case in this study is explicated via the historical lenses of researching public management and policy networks. Special attention is paid to the grounded theory approach (Corbin & Strauss, 2008; Strauss & Corbin, 1998) employed for collecting and analyzing qualitative data to answer the first research question about the predominant processes of network development at different stages. The collection and analysis of social network data are also thoroughly discussed by explaining the techniques for exploratory network analysis (Nooy et al., 2005) and confirmatory network analysis (Contractor et al., 2006).#p#分页标题#e#