The data for the state networks are available on the UNC Dataverse. Below I explain how to load networks from this. As in the article, each state-chamber-full-cycle is separated. The networks themselves are in the Edge_Lists.zip while useful information about the nodes for each network are in Metadata.zip.
Each csv in Edge_Lists represents 1 network. The first two columns are the sender/receiver identified with their “EID” (the identifier provided by NIMP). The third column is used to identify what threshold the edge was identified for. For robustness there were 5 thresholds estimated: .9, .95, .975, .99, .995. The number in that column goes from 1 to 5 and reflects each threshold. If an edge is a 3, then it means it was identified at the .975 threshold. When identifying edges you should use all edges at that level and above.
The networks in the article use the .975 threshold and so edges greater than or equal to 3 were included. I recommend using this network for most analysis and then reproducing analysis at other levels for robustness checks.
Each csv file in Metadata contains the node level attributes for a network. EIDs are given to match them. The other variables are:
More information on the nodes can be found on the NIMP website.
The following code provides an example of how to load a network in R using the igraph package for the Colorado lower house network from 2013 to 2014.