SNA has become a popular methodology for a wide variety of applications, and so one major challenge facing researchers is to make sense of the proliferation of network-related results. Bender-deMoll (2008) synthesizes a wide spectrum of SNA research as it pertains to human rights programs. Kilduff and Tsai (2003) provide an even more extensive synthesis of SNA research; they go deeper into the science of SNA and devote an entire book to outlining fruitful avenues for future research pertaining to organizational networks. We recommend both of the above sources to anyone interested in the opportunities for future research that we list below.
"The jury is still out as to whether social capital measured at the individual level does indeed have effects at the community level" according to Kilduff and Tsai (2003). Despite the increasing number of leadership network case studies, there is little comparative research looking at network effects, or systematically linking those effects to desired outcomes (Provan and Milward, 2001). Studies have conclusively linked network effects to individual-level outcomes (e.g., pay-raises and job-promotions), but the contribution of network effects to organization- and community-level outcomes remains unclear. As with other approaches to leadership development evaluation, it is important to recognize that attributing changes in communities to network effects is often difficult. Nevertheless, we think comparative leadership network case studies will significantly strengthen our capacity to understand how networks evolve and function in different contexts, and how they contribute to achieving desired outcomes.
Established standards for evaluating networks do not currently exist. In order for SNA to become a tool that can be applied with validity across different contexts, we need more comparative research on how network metrics are being applied in different contexts and with what results. Such research will enable us to refine our metrics and increase the likelihood that data is being appropriately analyzed and interpreted (Bender Moll, 2008). This research will require integrating different network data sets, which is complicated by the proprietary nature of these data sets. Sharing network data sets can jeopardize both the privacy of individuals described by the data and the professional interests of those who collected the data. Sharing health information involves similar benefits and risks; we hope that efforts to promote health information liquidity (e.g., Lorence et al, 2005) will spur similar innovations in sharing network data.
Collecting network data remains problematic. Using standard survey tools to collect network data is not practical for large networks (e.g., over 200 members). Surveys are also problematic for longitudinal network evaluations, in part because they provide no easy way to manage changes to names. For example, if network member Jill Smith changes her name to Jill Jackson, then any longitudinal network evaluation must recognize that these two names refer to the same person. Whichever way she is named in a survey is open to misinterpretation by her extended network of professional contacts, who do not keep track of her personal status.
The two limitations above are being addressed to some extent by network-specific survey tools that are more streamlined than traditional survey tools and by data-mining techniques that avoid surveys altogether (Tyler et al, 2003). In addition, social software sites such as Facebook and LinkedIn are extremely effective at managing large sets of longitudinal network data; however, these sites tightly control their data, prevent downloading altogether, and so frustrate the would-be evaluator. Evaluators of leadership networks need the best qualities of surveys, data-mining, and social software, all combined in one affordable system.
Finally, we note that popular social software sites demonstrate a useful approach to one of the thorniest privacy issues of SNA: Facebook and LinkedIn users can preview information that others report about them before that information is published. (The "Issues and Risks" section of this paper describes how network surveys handle this issue.) This is another reason why we are hopeful that lessons learned in the social software space will help improve SNA data collection.
We lack constructive guidelines for creating network maps and have only begun to understand how people perceive them. We know of very few papers that have considered how people perceive network maps. Much can be done to expand on research such as that of McGrath and Blythe (2004), which we illustrated in Figure 13. In order to advance our understanding of how people perceive network maps, researchers will first have to overcome three common shortcomings of software used to create network maps: lack of creative control over layouts, difficulty drawing large networks, and a tendency to create maps that are confusing or ambiguous (e.g., by drawing nodes on top of each other and thereby hiding all but the top-most node at that location). The fields of information visualization and human-computer interaction have much to offer this overall area of research. For example, Perer (2008),who addresses SNA from the perspective of these two fields, considers how people perceive network maps, provides tools to draw large networks, and proposes a well-defined process to replace the ad hoc techniques currently used to create network maps. We hope that Perer's work invites more researchers from these fields to apply their skills to the open problems facing SNA.
Structural equivalence has received insufficient attention from the leadership network community, compared to network topics such as centrality and clustering. Netflix has famously offered a million-dollar prize to anyone who can improve its recommendation algorithm, which is just one indication of the large volume of work on structural equivalence that the leadership network community can draw upon. We hope that the examples in this paper of applying structural equivalence to leadership networks will motivate readers to explore the topic of structural equivalence and to build on our work. Mathematical literature on structural equivalence is extensive: Wasserman and Faust (2004) provide an excellent introduction to the topic, and an up-to-date reading list can be found in the bibliography of Luczkovich et al (2005). These sources are more mathematically advanced than typical social network literature. For those who prefer less technical reading, we suggest Hanneman and Riddle's (2005) text and its section on visualizing "two-mode networks" as a helpful next step, in combination with the general introduction to two-mode networks by Borgatti and Everett (1997).
Many issues facing the field of SNA may have important implications for leadership networks. Unresolved issues in the field of SNA include the following:
- SNA represents a "structuralist" approach to organizations, fields, and communities, which complements an "individualist" approach. These two approaches have created two rival camps: "There is a pressing need for non-dogmatic research that explores issues concerning how individual differences in cognition and personality relate to the origins and formations of social networks" (Kilduff and Tsai, 2003).
- The most commonly used centrality metrics, strictly speaking, do not actually model sociological processes of interest; furthermore, many sociological processes that are interesting are not correctly modeled by any available centrality metrics (Borgatti, 2005).
- Further study is needed to understand the benefits and risks of measuring different kinds of network relationships. For example, Rizova (2006) has argued that measuring "seeks advice from" provides significant benefits in some contexts where measuring "works with" or "friends with" provides no benefit. LaBianca and Brass (2006) have pointed out that negative relationships (e.g., "do not like") are under-studied, even though they are often more informative than positive relationships. Cross et al (2003) have shown that positive and negative energy relationships (e.g., "energized by," "de-energized by") are particularly informative.
The dynamics of collective leadership networks deserve further study. Interesting avenues of inquiry include the following:
- What kinds of issues/causes most effectively lead to the formation of collective leadership networks? The general question of what makes something contagious or popular extends beyond the scope of our research, but Salganik et al (2006) suggest that network dynamics make popularity harder to predict than previously thought.
- What kinds of property rights most effectively facilitate the emergence of collective leadership networks? The open source software community has debated this question at length: When someone receives open source property, what rights and responsibilities does that person have? Feller et al (2005) study this and other aspects of the open source community.
- What behavioral norms help build and sustain collective leadership networks? How do people communicate with each other? Evans and Wolf (2005) provide a good starting point for this inquiry. They discuss best practices of the open source software community and the Toyota Production System.
- What kinds of incentives help build and sustain collective leadership networks. How can a sponsor promote "good" behavior? Cheshire (2007) investigates the effects of incentives on information exchange, in the context of wiki contributions.
| < Prev |
|---|
