Policy and Complex Systems
change in professional social networks . The literature on professional social network growth applies a top-down approach to drive changes in the network ’ s dyadic structure by relying on mathematically simulating changes in network statistics . Networks often form through bottom-up processes , with the network becoming an emergent property of agents ’ individual decisions ( Axelrod & Cohen , 2000 ; O ’ Toole , 1997 ). By simulating the actions of organizations in a partnership network , we develop a bottom-up approach to generating the kind of evolving change in inter-organizational partnership networks that has been studied in professional social networks .
Both inter-organizational partnership and interpersonal social networks are an emergent property of organizational and individual agents making their own decisions and taking their own actions to forge or sever links ( Axelrod & Cohen , 2000 ; O ’ Toole , 1997 ). Current methods for forecasting change in social networks rely on reproducing observed variation over time in whole network measures , such as clustering coefficients and node degree distributions . Simulation models add nodes and links to an existing network such that the observed patterns in network measures are reproduced . The evolution is treated as cumulative , with nodes and links remaining in the network once they have been added ( Barabasi et al , 2002 ; Kossinets & Watts , 2006 ). Since networks emerge from the accumulation of individual actions by the separate agents that form the network , efforts to model the growth of networks should take seriously the need replicate that emergence . Exponential random graph modeling ( ERGM ) measures the influence on the final network structure of the differing reasons that agents may use to select collaboration partners , such as finding new partners through existing partners or seeking out particularly well-connected agents ( Goodreau , Kitts , & Morris , 2009 ). Therefore , in extending the research on changing social networks to apply to interorganizational networks , we propose using the results of an ERGM to parameterize an agent based model for bottom-up network growth . This approach provides additional insight into the mechanisms of network change and how the mechanisms of network change can be harnessed to forecast future network growth .
We demonstrate our approach using data from the Vermont Farm to Plate Network . 1 The Vermont Sustainable Jobs Fund organizes and operates the Farm to Plate Network to promote closer interaction between food producers , professional food service , food sellers , and consumers , from government , industry , and the non-profit sector . Members take part in a range of working groups and task forces . Members are free to choose their level of involvement and may enter and exit the Farm to Plate Network at their discretion . The Farm to Plate Network pursues and measures increasing interaction between participants . This effort by the Farm to Plate Network provides the opportunity to study the growth of partnerships by providing data that identify changes in an inter-organizational partnership network over a short period .
1
Details on the Vermont Farm to Plate Network are online at << http :// www . vtfarmtoplate . com />> and << http :// www . vsjf . org >>.
4