Campus & Community

Unboxing the Beltway: Stevens Professor Uses Data Analysis to Glean Political Insights from Congressional Emails

Obamacare and Benghazi are finally slipping out of the Congressional eye.

Republicans are more sensitive to public healthcare issues than you might think.

And Secretary of State Hillary Clinton's recent revelations about her use of a private email account for official communications don't appear to concern our elected representatives a bit — at least so far.

Those are just a few of the surprising conclusions of Stevens political science professor Lindsey Cormack, who analyzes mounds of Congressional communications data and teaches her students to marry the modern tools of analytics with time-honored insights into political process.

Her work is somewhat in the vein of former New York Times blogger Nate Silver, whose analytics-driven popular FiveThirtyEight column correctly forecast 99 of 100 Presidential election state outcomes and 66 of 68 Senatorial races during the last two election cycles.

But there's a big difference, too: Cormack's special focus is neither elections nor voting patterns on the floor, but rather on what Congressmen and Senators tell their own constituents they're thinking about and doing.

"Some have said these communications are basically a form of junk mail," Cormack says. "But if it is junk mail, it's about the most important junk mail produced in the nation. Not only is it paid for with tax dollars, but also our elected representatives are dedicating staff and resources to these communications with a full appreciation of the opportunity cost of what else those staff could be doing instead.

"This represents an important look at what our legislators prioritize."

Cormack, a native of Kansas and graduate of the University of Kansas, joined Stevens as a visiting assistant professor in 2014 after obtaining master's and doctoral degrees from NYU and teaching political science at Yeshiva University. In fall 2015, she will become a full faculty member and director of Stevens' political science program.

Her Introduction to Political Science course begins not with a recitation of American presidencies or an analysis of Florida voting-machine chads, but instead a two-week discussion of game theory — in essence, the mathematical modeling and study of human decision-making.

"I see the non-politics majors just light up," she says. "Stevens attracts a certain type of student, oriented toward logic and mathematics, and this approach really whets their appetites. It prepares the way for me to teach them about the political process, which is of course much more than game theory, but this is a sort of portal to that world that they can more easily relate to."

She is also the creator and manager of a unique database, DC Inbox, of all official Congressional e-communications to their constituencies since 2009. (No similar resource exists, to her knowledge.) By writing queries in computing languages — Cormack learned to program in graduate school, and eventually settled on Python as her analytic language of choice — she can quickly answer complex, nuanced questions about the roughly 90 percent of U.S. Senators and Congressmen who deliver regular emails to subscribing voters.

In the fall of 2014, for instance, as Ebola cases began spiking and fears raged in the national media, Cormack mined Congressional emails to determine which Representatives and Senators were communicating more often and clearly about the disease to constituents. Perhaps a bit surprisingly, GOP lawmakers were far more likely to write about the topic: "a whopping 82 percent are from Republicans," she wrote in Washington Post, analyzing a two-month period from early September through early November.

And a deeper dive proved it wasn't just a numbers game.

"Republicans send more messages than Democrats. In the past two months, 69 percent of all e-newsletters have a Republican author," she continued. "Even after accounting for differences in total amounts of e-newsletters sent, however….[M]essages sent by Republicans reference Ebola 24 percent of the time, while messages sent by Democrats only reference Ebola 10 percent of the time."

Interestingly, by December she also found that both parties of Congress had already more or less abandoned the disease as a topic (much as media had done) and moved on.

What's next for DC Inbox? In the near future, Cormack will mine communications and discuss anti-Obama sentiment, which she sees rising sharply — on both sides of the political aisle. She will examine the question of whether the pervasive communications from members of both parties about veterans translate into actual legislative actions by the authors and research legislators' tendency to frame communications in terms of their own occupations: "as a farmer", "as a businesswoman", and so forth.

She will consider a suggestion to begin harvesting state governors' email communication.

And she will closely monitor the changes that will come with the recent rightward shift in power that Congress experienced during midterm elections.

"My theory is, the party in power always has less to prove….so they message fewer times. This may be why Republicans have sent to messages during recent years. It will be a very interesting dynamic to see if the frequency goes up for Democrats next year, or during the run-up to the next election cycle."

She'll also continue to teach both her undergraduates and graduates to wield the power of math when attacking questions about political motivation and messaging.

"The DC Inbox data is a fascinating peek inside Congress," she says. "I hope to make the database, and the methods, even more accepted and accessible in the future."