Growing Collaborations
jurisdiction in this specific case include areas such as Vermont , New York , and Quebec , as well as the United States and Canada ( Koliba et al , 2015 ). Jurisdictional level records the extent of an organization ’ s geographical reach , indicating whether its operations are relevant throughout a local municipality , the entire state , a sub-state region of the state , including watersheds and regional planning commission districts , or if the organization ’ s reach extends beyond state or national borders ( Comfort , Oh , Ertan , & Scheinert , 2010 ). Appendix A lists the attributes , their conceptual values , and the numeric labels assigned to each .
The main object serves as the Java modeling environment that provides the modeling space for the agent populations . The main object contains a range of state variables , all designed to link the model ’ s settings menu to the model ’ s main algorithm for adding new network links . This includes a separate state variable for each of the estimated ERGM terms ( see appendices B and C ), state variables to determine whether or not the model should run the link accretion algorithm , whether or not the model should run the link decay algorithm , whether or not the model should run the different control experiments described below , which subnetwork to use in the growth and decay algorithms , and how many times the accretion and decay algorithms should be run before comparing the modeled results to the empirically observed results ( see figure 1 ).
Process Overview and Scheduling
The model is built around two central processes , as depicted in figure 1 . In the first processes , a network is updated by adding some new links and removing some old links . This process is labeled as “ Model Operation ” in figure 1 . The model starts by reading in the list of organizations to be found in the network and an initial set of empirically observed links , forming a network . Our data contain multiple types of interactions , forming separate networks . These subnetworks are discussed in more detail below , under Initialization . One subnetwork is chosen and read into memory in the model . The model runs link
9