Lobbying activities have been an especially scrutinized area as of late in the media and communications field, especially in the wake of the recent decision on net neutrality. Concern abounds that similar victories may be becoming more remote in the face of growing media conglomerates and their ever-fatter wallets. As large corporations appear to gobble up more and more of the market and continue to spend vast amounts on lobbying towards legislative issues, I too have become interested in the ways in which interested parties are engaging in the practice. For this particular network graph, I was hoping to see if I could find any connections within this lobbying world. Were certain lobbyists specializing in communications lobbying? Or did they have a concentrated pool of clients? Did certain firms have more big-spending clients? Are these clients using only one lobbyist or do they have multiple agencies on their docket? And, more generally, who are the big players in the game, the most centrally connected to the overall community?

In order to explore these questions, I first had to gather my data. The senate website, senate.gov, allows you to query lobbying disclosures as mandated by the LDA. The user can search by a combination of registrant information, client information, lobbyist information, affiliated organizations, foreign entities, or filing details. In order to retrieve the data I used in this graph, I filtered the Filing Year to 2014 and the issue area to “communications/broadcasting/radio/TV”. The data returned is fairly simplistic, with only six columns – Registrant name, Client name, Filing type, Amount reported, Date posted, and Filing year. Interestingly, many of the rows also returned nil values for the amount reported, the most salient of the data fields. As I wanted to focus on the amount of lobbying money spent, I deleted these rows from my file. Next, I morphed my data into an edge table by using the lobbying agency as the source and the client name as the target. The amount reported became the weight so that the heaviness of the edge communicated the amount of money the client paid to the lobbyist. The rest of the columns became attributes that would later be used as a way to categorize the data. I also took out any instances in which the client and the lobbyist were the same (in essence, when the client acted as its own lobbyist) as this data was not appropriate for a network.

I then had to clean up a lot of the data as many filings used different name variants. Finally, I specified that all of my relationships were directional, as these were client – provider relationships and would never be mutual. In other words, the agency lobbies on behalf of the client, but the client will never lobby for the agency.

I imported the data into Gephi as an edge table and calculated both the average degree and graph density. The resulting network was incredibly sparse with a graph density of .003.   In fact, it was so sparse that my initial attempt at using the Force Atlas 2 layout resulted in a completely blown out network:

 

I ended up using Yifan Hu, which yielded a much better result. I also sized the nodes by density to make it easier to identify which nodes had the most overall connections, used edge color to differentiate the filing quarters, ranked the edges by weight or filing amount, and added in my labels. I labeled the edges as well as the nodes as the differing edge weights naturally brought up questions of how much lobbying money was being spent through each interaction.

 

The resulting graph is a definite mesh, with many different forms of networks within, the most common being lines and stars. There are a number of different isolated communities and one large, central module. Comcast Corporation is by far the largest node with the highest density, a result that is supported by news reports of their vast lobbying network on the hill.

While most of the isolated networks appear to involve smaller or less invested parties (the Makah Indian Tribe, for example, or the Queens Borough Public Library), Verizon is at the center of a particularly interesting community. Sharing only one connection with another lobbying client (Nokia), Verizon is particularly isolated from the larger network that encompasses many of its counterparts, such as Comcast, Time Warner, and DirectTV. It is no slouch, however, in the number of agencies used totaling 6 different edges, less than DirectTV’s 9 or Comcast’s 21, but still a fair number in comparison to the average degree of 0.842.

I also filtered the graph by filing quarter, interested in seeing whether any quarter seemed particularly dense or active.

Quarter 1:

 

Quarter 2:

Quarter 3:

Quarter 4:

More lobbyists were being used in the fourth quarter than in any other and the weight of the edges in many instances were particularly heavy. This naturally leads to the question of what issues were being debated in the final months of 2014 that were of such concern. Net neutrality may well have been one of them, but additional research would certainly help to contextualize this graph and perhaps explain some of the results.

I think it would also be interesting to explore further attributes other than filing quarter to see if they bring anything new to light – for instance looking into the different categories of clients (are they a large media corporation? A manufacturer? A union or association?), the different specializations of the lobbying firms, the size of the institutions, even the location headquarters. I would be especially interested to see similar graphs on more concentrated issues as you can also query the senate.gov database by specific lobbying issue, perhaps even including the client’s desired result as an attribute (are they for or against?). I wonder how alike the most connected clients are on these issues and what that might mean for the communications world as a whole.