A Comparison of Three Versions of Ranked Choice Voting using Computer Simulations

ABSTRACT. This paper compares three versions of Ranked Choice Voting (RCV) that differ by their elimination methods: Least First-Place votes, Most Last-Place votes, and Lowest Borda Count. Results are compared to two established measures of majority support—the Condorcet winner and Condorcet loser criteria—and two other criteria that are important to social choice scholars—Independence of Eliminated Alternatives and Reversal Symmetry. R simulations under an Impartial Culture assumption are run for various combinations of candidates and voters. Results show that the Lowest Borda Count outperforms the other two elimination procedures on the Condorcet winner, and Independence of Eliminated Alternatives criteria, while the Most Last-Place method tends to perform at least as well as Borda elimination on reversal symmetry. All three methods perform relatively equally on the Condorcet loser criterion. These findings underscore the need for policymakers and electoral designers to consider the nuances of RCV before adopting the system.