Tracing connections in the State House

BEACON HILL can be an opaque place. With 40 senators and 160 representatives, it is often difficult for those without extensive ties or experience “in the building” to know the true networks of influence in the State House. Luckily, in the first month of each legislative session, the elected officials themselves pull back the curtain a bit, maybe without even knowing it.

The Massachusetts Legislature has an early filing deadline – January 18 this year – for all bills for the session and then a quick turnaround for each legislator to cosponsor bills. Those cosponsorships provide a window into how lawmakers are connected in the State House. By signing on to bills together, publicly stating what issues and what other legislators they agree with or will work with, senators and representatives create a complex network that, with a little math, we can mine for insights.

Using network theory, it’s possible to explore the structures of these social networks. This approach can provide insights into the overall nature of an environment – whether, for example, friendship groups in school break down more along racial or age divisions. Network theory can also reveal information about people within the network. Those who occupy a central place in the network – meaning that they have better or closer connections to most parts of the network – are often the most influential. This approach has been used to show that Torey Krug played a more important role for the Bruins last year on the power play than when the team and its opponent were at even strength. And the approach also showed that Tyrion Lannister is the most influential character in Game of Thrones. He is a Lannister, married a Stark, and works for Daenerys. That puts him at the center of different social groups in Westeros. Centrality is a bit like finding the Kevin Bacon of a social network.

Academics have applied network theory to understanding influence in Congress. Since legislation passes when a majority of legislators support it, the best connected, or most central, legislators could use their strong relationships to win the votes needed for passage, providing them with a measure of influence that cannot be captured in more formal metrics.

This isn’t to say that cosponsorship data is the Rosetta Stone for understanding absolute influence Beacon Hill. In politics, there are many ways for a legislator to have and use influence. In Washington, there is no doubt that Speaker Nancy Pelosi is the most influential member of the House, due to the powers of her office. It is also probably true that Rep. Alexandria Ocasio-Cortez is the most influential member of Congress on social media. That does not translate to greater overall influence than Speaker Pelosi, but it likely has helped her elevate her stature above other freshmen representatives.

Being central in cosponsorship networks is similar to having a high number of Twitter followers. It does not, in itself, grant a legislator control over the political process. But it is an area where a legislator’s power can be grown, and can reflect the influence that members develop by working with colleagues. Most importantly, this centrality helps provide a window into a type of influence – legislators working with other legislators – that is otherwise extremely difficult to measure.

Collecting the data

To explore this area, I scraped all the bills for the House and Senate from the 186th General Court (2009-2011) through the 190th (2017-2018). I counted all the times the same two legislators sponsored or cosponsored bills together during these five sessions and weighted them by the number of cosponsors on that bill. The weighting means the more legislators sign onto a bill, the connection between any two is diluted, so that cosponsoring a bill with one other legislator is worth proportionally more than a bill with 50 cosponsors. The weighted cosponsorships across all bills in the chamber were added up, plotted in a network, and legislators measured for their centrality (technically, eigenvector centrality).

I filtered out members of leadership who did not cosponsor much legislation and special election winners who were not in office long enough to generate understandable data. All legislators were coded for their tenure in office, gender, and their party affiliation.

With that information, and the caveats that come with it, in place, let’s move on to the data.

Women are more central cosponsors

 In the House of Representatives, women are more central in the sponsorship networks for all sessions since 2009 than would be expected if there were no difference between genders. Women are, in fact, the only group (male, female, Democrats, Republicans, newer members, and older members) examined that displayed a consistent trend in every session. This does not mean that women sponsor more bills, but that the bills sponsored and the implied relationships with other members indicate a more central, and therefore, potentially more influential position than we would expect when controlling for party and tenure.

In four of the past five sessions, the most central legislators in sponsorship networks have been women. Rep. Denise Provost of Somerville was the most central twice, while Reps. Elizabeth Poirier of North Attleboro and Kay Khan of Newton were each most central once. The lone male lawmaker to fill that role was the late representative Chris Walsh of Framingham during the 2017-2018 session.

Does this mean that women are the power brokers in the House? Based on what we know of Massachusetts politics, it is more likely that, because positions in leadership have generally not been occupied by women, cosponsoring legislation is one of the avenues of influence available to female legislators. As with Pelosi and Ocasio-Cortez, there are numerous ways to shape the political process, and we could expect that obstacles in one area might lead a legislator to focus on another. Therefore, outsized influence in cosponsorship – at least for this group – might be reflective of historical underrepresentation in leadership.

Whether this will continue to hold true if women assume more leadership positions is a question that we will hopefully find the answer to soon.

Tenure doesn’t matter

 I had assumed that more senior legislators would move into more central positions among sponsoring networks the longer they serve. But the data indicate that they don’t. In fact, the highest members of leadership, Speaker Robert DeLeo, Majority Leader Ronald Mariano, and Assistant Majority Leader Joseph Wagner, had to be filtered from the analysis because they were outliers in how few bills they cosponsored. Longer-tenured members also showed no significant difference than younger members.

 Perhaps members become more selective in what they sponsor as the years go on. Or maybe, as members move into leadership, opening up greater influence in one area means less energy is directed at cosponsorship. For whatever the reason, tenure was the one variable with the least change or explanatory power across the years or chambers.

Party matters – a lot

 We can use an algorithm that detects communities, like friendship groups in a high school, to find if there are groups that consistently cosponsor with each other. For Congress, this type of analysis showed that southern Democrats had traditionally been more closely aligned with Republicans than other Democrats and confirms that there is now growing polarization between the parties.

On Beacon Hill, political leaders often talk about their bipartisan approach to legislation, and suggest they operate very differently from their counterparts in Washington. But the sponsorship analysis suggests otherwise. In the House, Democrats sponsor bills with other Democrats and Republicans sponsor legislation with other members of the GOP. While they may sponsor some bills with members of the other party, they rarely overlap in who they primarily sponsor with. The Senate is a little less split along partisan lines, but the analysis also shows two groups, one with all Republicans, and one with all Democrats and some Republicans.

 The House cosponsorship network in 2011-2012, after a wave of new Republican legislators came in after the 2010 election, illustrated the partisan tendency behind cosponsorships. Republicans during that legislative session cosponsored many bills together and the network analysis showed that a Republican was the most central member, due to the new weight the caucus had in bills filed.

This could be a result of legislators having similar political philosophies, or a result of the limited time available to sponsor legislation. With only a few weeks to consider legislation, bills from members of one’s own party and from lawmakers who entered the Legislature at the same time presumably gain higher attention, and 2011 saw a large number of freshmen Republicans. Whatever the reason, it shows that party is still a major driver in who cosponsors with whom.

Conclusions

 This is only the start of what can be found with this information; there are other ways to analyze networks, more metadata can be added and investigated, and, every two years, more cosponsorship data is created. But the analysis shows that there indeed are insights to be gleaned from data, even as esoteric as cosponsoring legislation, for parsing out the structures underlying Massachusetts politics. As more research into state government is pursued, and as data become more available, the public will hopefully receive a better view of the often-invisible networks at play on Beacon Hill.

Chris Oates is the CEO of Two Lanterns Advisory, a political risk consultancy. He is also a lecturer at the Boston University Pardee School of Global Studies.