CommonWealth magazine yesterday ran my article about using network theory to understand how cosponsorships in legislation indicate a social structure underlying the state government. Basically, we can take the web of legislators who have cosponsored bills with each other, treat that like a social network, and use some sociological methods to tease out some insights about what that social structure means.

Cosponsoring network for 2017–2018 in the Massachusetts House of Representatives

I can be found here and I hope you enjoy it (https://commonwealthmagazine.org/opinion/tracing-connections-in-the-state-house/)

Beyond the findings detailed in the article themselves, this type of approach has some lessons for political risk and its practice.

Don’t overlook overlooked data

Cosponsoring legislation is the odd duck of the legislative process. Nowhere in the Schoolhouse Rock version of politics does it mention that legislators can publicly support a bill that may not make it out of committee or which they may eventually vote against anyway.

Because cosponsorships don’t by themselves make legislation happen, and are at best only an indirect indicator of their passage into law, they are often ignored in political analysis. But in doing so, we walk past a huge wealth of information about how politics functions in a legislature.

Cosponsoring data, as my article, and other academic articles in the field have attempted to show, can be used to understand elements of a political environment that can give us a richer perspective when making forecasts about whether legislation can pass.

Don’t rely solely on unconventional data

One of the temptations in political risk is to find or create numerical indices, and to leave it at that.

Quantitative analysis can lead to precision bias; we think that a conclusion is accurate just because the numbers seem to be precise.

In the MA cosponsorship information, female legislators are more influential in every House session for the past ten years. If I used that data only, I would conclude that the Massachusetts State House is biased towards women, when in fact women have been historically underrepresented in leadership positions. I would conclude one thing, while those with knowledge of how it really works would conclude the opposite.

This led me to look deeper into what might be the causal relationship between cosponsorship and influence. My initial theory is that cosponsorship is an area where non-leadership legislators exert a larger share of their political influence than leadership, who have other places to shape the legislative process. If women have a disproportionate share of non-leadership positions, then they would show up in the data as having a higher consponsorship centrality.

The best analysis leads to more analysis

Political risk is never complete. Whatever the subject area, as long as the world keeps changing, the analysis of it will have to keep up.

The best analysis will always generate a new set of questions for future research. In this case, I think that female legislators’ influence might be a reflection of their underrepresentation in leadership. This leads to questions about whether that will continue as more women enter leadership, or whether it extends to other groups, or to what extent is there a difference in overall power between a highly influential cosponsor and a member of leadership.

These won’t be answered immediately, but they go on a list of useful questions that help direct future research. Political risk can be a daily grind of keeping forecasts updated with the news, but there should also be an longer-term trajectory of improving the practice as a whole and advancing the field. Knowing where to look next is the first step for that.