Working paper

More than a Feeling

Gary E. Hollibaugh, Jonathan Klingler, and Adam Ramey


Political scientists have long considered the primacy of ideology, party affiliation, and constituency preferences in determining how members of the U.S. Congress make decisions. At the same time, psychologists have held that individuals' immutable personality traits play a central role in individual decision-making. In this paper, we seek to bridge these literatures by offering a rational-choice based characterization of how personality informs legislator decision-making, independently of policy preferences. Specifically, we provide an decision-theoretic grounding for the "Big Five'' personality model and identify how different personality traits affect the legislative behavior -- broadly defined -- of members of Congress. We apply recent advances in computer science to estimate personality across each of the "Big Five" dimensions using speeches recorded in the Congressional Record. In particular, we use Support Vector Machines (SVM) models to connect linguistic cues in legislator speech with known personality correlates of those cues. These results in hand, we study how legislators personality traits affect their behavior in Congress. Our results are substantively fascinating and a significant contribution to the discipline, as they show that personality has a strong relationship with legislator behavior -- even after controlling for legislator ideology.


personality; congress; machine learning; legislative politics; polarization;

JEL codes

  • D72: Political Processes: Rent-Seeking, Lobbying, Elections, Legislatures, and Voting Behavior
  • C1: Econometric and Statistical Methods and Methodology: General
  • D7: Analysis of Collective Decision-Making

See also

Published in

IAST working paper, n. 14-09, March 2014