By Rob Mitchum // December 5, 2013
Politicians do a lot of talking; some might even consider it to be their primary talent. Between campaign events, floor speeches, press releases, television appearances and newsletters, politicians leave behind a mountain of words that can rival datasets from genomics and physics. So it’s no surprise that political scientists are increasingly turning to computational methods to find insight within this big political data. In Justin Grimmer’s talk at the CI on December 4th, he demonstrated how the words politicians use reveal the roles they’ve chosen to play — and how those roles deepen the polarization seen in Washington DC today.
Grimmer, an assistant professor at Stanford University, set up his talk by presenting two representative senators: Jon Tester of Montana, and Sheldon Whitehouse of Rhode Island. While both congressmen are second-term Democrats first elected in 2006, they take very different strategies in communicating their accomplishments to their home constituency, Grimmer said. Tester issues frequent press releases and social media announcements about securing funds for programs in his state, with amounts ranging from millions of dollars down to tens of thousands. By contrast, Whitehouse is outspoken about global issues such as climate change and releases many statements about partisan battles with the Republican Party.
These actions represent two primary strategies of national legislators: credit-claiming and position-taking. To measure how often individual politicians use either of these approaches, Grimmer used a mainstay of political communication, the press release. In the U.S. Senate alone, thousands of press releases are issued annually, with each Senator’s office putting out an average of 212 each year. Grimmer and his colleagues used a statistical topic model to analyze this PR library for subject matter, priorities and style, creating a spectrum of the 100 Senators along the axis of “position takers” to “appropriators.”
Further analysis found that the strategy a politician chooses is probably not just a matter of personal beliefs. When Grimmer and colleagues looked at the political makeup of each Senator’s home state, they found a relationship between the partisanship of the state’s voters and the legislator’s preferred strategy. If a Democrat represented a state with a majority of voters who typically vote Republican (or vice versa), they tended to emphasize their appropriation efforts for the district, like Tester. But if the Senator hailed from a state that significantly favored their party in elections, they were more likely to be a vocal supporter of their party’s policy platform, like Whitehouse.
This relationship has implications for the tenor of political debate in Washington DC, Grimmer said. If only the legislators representing dark red or blue states participate openly in policy discussions — while those from more moderate states keep their heads down and focus on bringing funding to their constituents — then those on the extreme of each party will be left to drive the discussion. Supporting this hypothesis, Grimmer and colleagues found that the hot-button issues from the year they analyzed (2006), such as immigration and the Iraq War, were more polarized politically than the average for all topics. The ensuing increased partisanship and lack of compromise is reflected in recent legislative crises, such as the government shutdown in October.
For more from Grimmer’s presentation, including the results of experiments on how voter sentiment changes in response to different types of press release language and credit-claims, watch the entire talk via Adobe Connect. Grimmer also has two books available, Representational Style in Congress: What Legislators Say and Why It Matters, and The Impression of Influence: How Legislator Communication and Government Spending Cultivate a Personal Vote.
[Photo via Wikimedia Commons.]
