Ranked Choice Voting and Proportional Representation

Abstract

It is widely claimed that ranked choice voting---and single transferable vote (STV) in particular---is conducive to proportional representation. In this paper, we clarify what might be precisely meant by that claim, and then offer confirmation: under a wide range of assumptions about voting behavior and electoral conditions, STV systems will likely secure roughly proportional outcomes, especially when compared to plurality elections in single-member districts (SMD). To do this, we develop novel data-driven methodology for generating ranked ballots in the presence of voter polarization. We demonstrate the methods on case studies from localities in Louisiana, North Carolina, Ohio, and Texas. We find that STV tends to project proportional or slightly higher representation for the relevant minority group in each case, while SMD varies widely in its effectiveness depending on local circumstances. This work brings a statistical modeling toolkit to the questions around ranked choice voting and proportionality, allowing considerations of voter blocs and cohesion to enter the conversation in new ways.

Keywords: ranked choice voting, proportional representation, Plackett-Luce, Bradley-Terry