Ballot Length in Instant Runoff Voting

Kiran Tomlinson, Johan Ugander, Jon M. Kleinberg

Introduction

Instant Runoff Voting (IRV), or Ranked Choice Voting, is increasingly adopted to mitigate "spoiler" effects and ensure winners possess a broad mandate. While ideally voters rank all candidates, practical constraints often lead jurisdictions to limit rankings to a fixed "ballot length" ($h$), such as the top three or five choices. This truncation is intended to reduce voter fatigue and simplify ballot design.

However, this paper investigates whether such truncation is a neutral administrative choice or a consequential variable. By holding voter preferences constant and varying the allowed ballot length, the authors explore if the IRV algorithm itself produces different winners.

Illustration of ballot truncation where ballot length $h=2$.

Methodology

The researchers employed a multi-pronged approach:

  1. Theoretical Analysis: Deriving lower bounds on voter counts needed for multiple winners and constructing sets of rankings where every possible ballot length yields a unique winner.
  2. Linear Programming (LP): Searching for "pathological" profiles that maximize winner changes using only full ballots.
  3. Simulation & Empirical Study: Millions of elections were simulated using random and "1-Euclidean" (ideological spectrum) preferences.
  4. Real-World Data: Analysis of 168 election datasets from PrefLib, including APA presidential elections and local government contests in San Francisco and Glasgow.

Key Findings

The study reveals that ballot length is a powerful "hidden" variable in IRV elections.

The "Anything Goes" Theorem

The authors prove that for $k$ candidates, it is theoretically possible to produce $k-1$ different winners simply by varying the ballot length. For $k > 3$, as few as $2k^2 - 2k$ voters are sufficient to create such a scenario. Even under structured "single-peaked" preferences, the number of potential winners remains significant, at $\Omega(\sqrt{k})$.

Real-World Sensitivity

Empirical analysis of PrefLib data shows these theoretical concerns have practical weight. In 25% of the 168 real-world elections studied, changing the ballot length would have altered the winner. In some cases, the winner "flip-flopped" between different candidates as the allowed length increased.

Mean and maximum number of truncation winners as a function of the number of candidates $k$.

Simulation Results

In simulated 1-Euclidean elections—often considered the most stable—the probability that the "true" full-ballot winner retains their victory with a short ballot (e.g., $h=3$) falls below 50% once the candidate pool exceeds nine.

Discussion and Future Work

The results suggest that the choice of ballot length is a policy decision that can determine election outcomes. A notable example is the 2009 Burlington Mayoral election, which the authors identify as a "toss-up" at a ballot length of $h=2$, where minor vote shifts could have flipped the result.

A primary limitation of this study is the assumption of sincere truncation: it assumes voters do not change their ranking strategy when faced with a shorter ballot. Future research should investigate strategic behavior under truncation, explore the "margin of victory" sensitivity, and extend this analysis to other voting systems like Borda count or Copeland’s method.

Summary updated

Abstract

Instant runoff voting (IRV) is an increasingly-popular alternative to traditional plurality voting in which voters submit rankings over the candidates rather than single votes. In practice, elections using IRV often restrict the ballot length, the number of candidates a voter is allowed to rank on their ballot. We theoretically and empirically analyze how ballot length can influence the outcome of an election, given fixed voter preferences. We show that there exist preference profiles over k candidates such that up to k-1 different candidates win at different ballot lengths. We derive exact lower bounds on the number of voters required for such profiles and provide a construction matching the lower bound for unrestricted voter preferences. Additionally, we characterize which sequences of winners are possible over ballot lengths and provide explicit profile constructions achieving any feasible winner sequence. We also examine how classic preference restrictions influence our results—for instance, single-peakedness makes k-1 different winners impossible but still allows at least Ω(√k). Finally, we analyze a collection of 168 real-world elections, where we truncate rankings to simulate shorter ballots. We find that shorter ballots could have changed the outcome in one quarter of these elections. Our results highlight ballot length as a consequential degree of freedom in the design of IRV elections.
BibTeX
@inproceedings{tomlinson2023ballot,
  author = {Kiran Tomlinson and Johan Ugander and Jon M. Kleinberg},
  title = {{Ballot Length in Instant Runoff Voting}},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year = {2023},
  doi = {10.1609/aaai.v37i5.25724},
}
Figure