Directional and Euclidean Theories of Voting Behavior: A Legislative Comparison

Abstract

Recent research by Rabinowitz and Macdonald (1989) claims that voting behavior is better accounted for by a directional model than by a traditional proximity or Euclidean model. This paper compares directional and Euclidean models using congressional roll-call voting data. For these relatively well-informed voters, we can unambiguously reject the directional model in favor of the traditional Euclidean spatial model. We conclude that congressional voting can indeed be very accurately represented by the Euclidean model.