Invisible Fairness: Uncovering Hidden Biases in RCV Through Margin-of-Victory Sensitivity Analysis

Building on Iceland et al. (2024), who highlight the unpredictability of RCV outcomes based on vote sampling, and Clelland (2023), who demonstrates real-world Condorcet failures, this research would develop a sensitivity analysis tool that maps how small shifts in vote rankings (especially in tight races) may lead to disproportionately unfair or paradoxical outcomes. Unlike prior work focusing on overall predictability or isolated case studies, this approach would systematically analyze “near-miss” scenarios—where margins of victory (MoV) are small—and assess whether RCV or similar systems inadvertently amplify demographic, partisan, or geographic biases in these cases. By visualizing and quantifying where and how fairness breaks down in practice, this tool could become a practical auditing mechanism for election administrators, offering transparency and early warning for potential systemic biases that traditional methods miss.

References:

1. Sampling Winners in Ranked Choice Voting. Matthew Iceland, Anson Kahng, Joseph Saber (2024). International Joint Conference on Artificial Intelligence.
2. Ranked Choice Voting And Condorcet Failure in the Alaska 2022 Special Election: How Might Other Voting Systems Compare?. J. Clelland (2023).