Structure, Function and Power in Social Networks
Network analysis concerns itself with the study of relationships among actors within a given network. New types of SNA approaches are incorporating not only actors, but also ideas or discourse, in the measured networks. One body of theory calls these ideas “actants,” to distinguish them from willfully self-propelled “actors” (Latour 2005). The formal approach of Integrated Structurational Analysis (ISA) has been proposed to integrate the various dimensions in societal processes as network vectors among the units (actors and actants) (Broadbent 1998, 2003). The present section will discuss the approach and measures developed for social networks that can be applied to this kind of integrative synthesis and analysis. The essay will then distinguish social action and discourse networks and examine their interaction.
Depending on how the term 'actor' is operationalized, actors within social networks can be individuals (micro-level social networks), groups or organizations (meso-level social networks), or states in global relationships (macro-networks). To examine the networks that underlie and produce national-level policy formation, it is often appropriate to use organizations as the actor or social unit of analysis, as done here. The relationships between actors in a social network are described as ties, and represent a point of social contact between actors within the system. The social contact can consist of any type of interaction, be it the sharing of information or ideologies, the dissemination of a norm, or the exchange of support or resources. Much social scientific research, as is typical of survey research, has been conducted on samples of individualized actors. SNA differs because it also collects information on the relations or ties among the actors. If we want to study the ties as constituting a whole system, we have to study the group of actors that could potentially have direct ties among themselves. That rules out the random selection of actors from a large population (though one can study “ego-centric” networks that way). Rather, to study systems of relationships, we have to study the patterns of ties among a set of actors susceptible to relationships.
This kind of whole network study is applied to some kind of community, such as a classroom, a town, or in this case, a “policy domain.” A policy domain refers to all the actors potentially influential upon a certain type of policy within a nation-state (or governmental area). The SNA approach takes into account both the qualities of the actors themselves, both their resources and their ideas, and the vital relationships that transfer those qualities as sanctions among the actors. The relational theory underlying the SNA approach argues that societal power is relational in that it involves the connection and mobilizations of numbers of actors and ideas. Hence, the relational ties are fundamental because they reveal the active flow of ideas and resources among actors that enable the power to affect policies and large scale societal changes.
One of the primary structural concerns within social network analysis (SNA) is identifying the “most important” or “most prominent” actors within a given social network (Wasserman and Faust 1997). The concept of actor importance is a measure of the property of actor location within a social network, with the most important actors being located in the most strategic locations within the network. Thus, an actor's role is characterized by its structural position within the network (Borgatti and Foster 2003). Actor centrality is the measure typically employed for quantifying an actor's role within a social network. A central actor is an actor with many ties to other actors within a network.
Bodin et al. (2006) note that a network actor with a high degree of centrality can effectively coordinate actors within the network during times of change. Burt (2003) characterizes these actors as “brokers” within social networks. In social networks the policy sphere, this means policies can be passed through a legislature more quickly, but Abrahamson and Rosenkopf (1997) contend this leads to centralized decision making within the network. Another implication is that actors within the network will have more limited access to other sources of information (Weimann 1982).
Another key structural feature of social networks is density within the network. Density within SNA is a quantifiable measure of connectivity between actors within the network and of the connectivity of the network as a whole. Density is not used as a measure of centrality per se; rather, it is a measure of the network's cohesion – the number of links between actors within the whole network, not on an individual basis as with centrality (Wasserman and Faust 1997). Density of a network is calculated by dividing the number of links by the number of nodes within the network. One of the structural characteristics of dense social networks is a buffering capacity referred to as redundancy (Bodin et al. 2006). In dense networks, if an actor is removed from the network, because of the many links between other actors within the network, the loss does not have as profound an effect on the overall network structure. For advocacy coalitions, this means even if a central actors is removed from the coalition, other actors can step into the position and assume the functions of the central actor (Folke et al. 2005).
Social networks with greater connectivity of knowing each other exhibit higher levels of trust among actors within the network (Granovetter 1985). Pretty and Ward (2001) theorize that greater network density increases the possibility of social control of the actors within the network, which facilitates top-down regulation of the environment by the state. Oh et al. (2004) caution that dense networks can streamline policy processes, but may also promote homogenization of both experience and knowledge. Moreover, Frank and Yasumoto (1998) caution that too many links between actors within a social network can lock certain actors into inflexible positions, making political change difficult.
A concluding example of a structural feature of social networks is modularity or betweenness. Betweenness can be measured within a social network by quantifying the distance between nodes within a network. In any given social network, groups with high internal density may be loosely connected to other groups with high internal density. This phenomenon is termed modularity, and describes groupings of actors within a social network (Bodin et al. 2006). Within a civil society, this can be characterized by businesses having dense ties to one another, but weak or peripheral ties to government ministries or environmental NGOs. The betweenness of actors within a social network is a measurement of diameter – it is the number of steps needed to reach from one node to another within then network.
A high degree of betweenness in social networks allows different blocks of actors to interpret knowledge and develop policy responses distinct from one another. This is often the case in social networks around ecological governance, with different blocks developing different interpretations of data about the environment (Ghimire et al. 2004). The more modular a social network is, the less trust is demonstrated between different blocks within the network (Borgatti and Foster 2003). Likewise, it is more difficult to transfer tacit and/or complex knowledge (“externalize” scientific knowledge) within social networks with a high degree of betweenness (Reagans and McEvily 2003). In turn, advocacy coalitions characterized by high modularity within the network are prone to fragment, as the removal of a single actor can disengage a block of actors from the rest of the network (Borgatti and Foster 2003).
SNA sometimes assumes that higher centrality gives an actor more power over the other actors, and hence over the behavior of the whole network. However, this assumption is greatly in need of empirical testing in actual policy systems. The policy network approach taken by the Compon project includes measures of actor power in the formation of policy. One measure is created by survey respondents checking off those actors in the list they think to be very powerful within the policy domain (in this case, climate change). This is a reputational measure of power. Another measure involves the actor scoring their degree of satisfaction with the outcome of a policy debate in which they were involved. The higher the satisfaction, the measure assumes, the greater the effective behavioral power of the actor. These power measures can be used to trace the relative influence of different actors, their coalitions, and their ideas and ideologies – in this case about climate change. These measures were developed in earlier policy network studies (Knoke et al. 1996).