Past Political Methodology Workshops

Over the last forty years, Members of Congress (MCs) have polarized in their legisla- tive behavior, while representing relatively centrist electorates. This lack of anchoring by median preferences highlights a central puzzle: How do polarized candidates run and win elections based on legislative records that are increasingly ‘out of step’ with their districts? Existing research points to two potential sources: a changing balance of electoral forces fa- voring partisan voters and ‘shirking’ by policy-motivated MCs. In this paper, I study forty years of position-taking in congressional elections to assess these competing accounts. In particular, I scale the issue statements aired in 12,692 House and Senate ads from 1968 to 2008, using cosponsorship bill titles to link words in ads to the same dimension of ideo- logical conflict in Congress, and replicate this analysis using ads from the 2008 election. Overall, I find that candidates consistently present themselves as moderates, while portray- ing their opponents as extremists through a process of issue distancing. I find consistent evidence that this distancing may help candidates win votes, by mitigating the fallout from their partisanship. This finding provides additional support for the elite-driven account of a representational disconnect in American politics, suggesting fundamental limits to the ability of voters to hold their representatives accountable in contemporary elections.

American are flocking to social media.  In doing so, we empower friends and acquaintances to influence our information environment, affect how we think about politics, and ultimately how we engage in political activity.  This work tests social media's influence on mass political behavior with a large scale (N=1.6M, 12.3K survey responses), long term (72 day) field experiment that increased the amount of political news treated users saw in their Facebook news feeds during the 2012 campaign.  The scope and scale of the data allow for an extensive analysis of when and for whom these heterogenous treatment effects were most powerful, thanks to behavioral data on Facebook use, exposure to, and consumption of political information (pre-treatment)---obtained by utilizing supervised text classification; incorporating measures of ego-network political identification and behavior; and exploiting conventional demographic variables.  Findings show that political information encountered in social media increased the salience of politics, led to stronger political opinions, and increased voter turnout.

ABSTRACT: I develop a method to measure the ideology of candidates and contributors using campaign nance data. Combined with an expansive dataset of over 81 million contribution records from state and federal elections, the method recovers ideal points for a wide range of political actors. The common pool of contributors that give to campaigns across institutions and levels of politics makes it possible to recover a unied set of ideal points for candidates for Congress, the presidency, state legislatures, governor, and other state-wide o ces, all in a common-space with the interest groups and individual donors that fund their campaigns.

Abstract: Previous research has demonstrated that incumbents who have more “ideologically extreme” roll-call voting records face a higher risk of losing re-election than those with more moderate voting records. I demonstrate that incumbents between 1934 and 1980 faced a higher penalty for extreme voting records than incumbents from 1900 to 1930 or 1984 to 2010. Congressional polarization was much lower between 1934 and 1980 than in other years, yet as incumbents have polarized, they have not become more likely to lose.