Combatting Gerrymandering with Social Choice: the Design of Multi-member Districts

Nikhil Garg, Wes Gurnee, David Rothschild, David B. Shmoys

Abstract

Partisan gerrymandering remains a persistent challenge to democratic integrity in the United States, exacerbated by the winner-take-all nature of single-member districts (SMDs). This paper investigates the potential of Multi-member Districts (MMDs) combined with Single Transferable Vote (STV) to mitigate these issues. By employing large-scale computational redistricting and social choice theory, the researchers demonstrate that MMDs—specifically those with 3 to 5 members—effectively collapse the "gerrymandering range," ensuring proportional representation and protecting both diffuse and concentrated minority groups. The findings suggest that structural reform, rather than just boundary adjustments, is a viable path toward fairer electoral outcomes.

Introduction

The standard U.S. electoral system relies on SMDs where a single representative is elected by a plurality of the vote. This "winner-take-all" framework is mathematically vulnerable to gerrymandering: partisan mapmakers can "pack" opposing voters into a few districts or "crack" them across many to dilute their influence. Furthermore, even in the absence of malice, "natural gerrymandering" occurs when voters self-segregate geographically (e.g., Democratic concentration in urban centers), leading to non-proportional outcomes.

This study explores whether moving to Multi-member Districts—where larger geographic areas elect multiple representatives via Ranked Choice Voting—can structurally eliminate these vulnerabilities. While the concept is central to the proposed Fair Representation Act, this research provides the first algorithmically rigorous, nationwide analysis of how MMDs interact with partisan incentives.

Methodology

Algorithmic Map Generation

Evaluating the impact of MMDs requires navigating an astronomical number of possible redistricting plans (up to $10^{12}$ for large states). The authors extended the Stochastic Hierarchical Partitioning (SHP) algorithm to meet this challenge. SHP recursively subdivides a state into population-balanced, contiguous regions, creating a hierarchical tree of possible maps. This structure allows for the use of dynamic programming to identify the most extreme partisan plans (the "best" possible Democratic or Republican maps) and the most fair alternatives.

The "Solid Coalitions" Shortcut

A significant hurdle in this research was the computational cost of calculating STV outcomes for millions of simulated maps. To address this, the authors proved Proposition 1: if voters rank all candidates of their own party above those of the opposition (the "solid coalitions" assumption), the STV seat share can be calculated instantly from the district's vote share. This theoretical breakthrough allowed the researchers to integrate STV outcomes directly into the map-optimization loop. To ensure real-world applicability, they validated these results using individual-level voter files to simulate "noisy" behavior, such as cross-party voting.

Data Sources

The study integrated several high-resolution datasets:

  • Precinct-Level Election Returns: Statewide data from 2008 to the present.
  • Census Data: 2010 Census tract-level population and American Community Survey estimates.
  • Individual Voter Files: Anonymized data including demographics and modeled partisan scores for millions of voters.

Key Insights

Collapsing the Gerrymandering Range

The most significant finding is the dramatic reduction in the "gerrymandering range"—the gap between the maximum seats a Republican mapmaker can secure versus a Democratic one. As district size increases, this range collapses. Moving from 1-member to just 2- or 3-member districts provides approximately half of the total potential benefit in reducing proportionality gaps.

The "Massachusetts Problem" and Diffuse Minorities

In states where a minority party is evenly diffused (e.g., Republicans in Massachusetts), it is often impossible to draw a single-member district they can win. MMDs solve this by lowering the "threshold of exclusion." By allowing diffuse minorities to pool their votes across a larger area, MMDs ensure they can reach the quota required to win a seat, which is impossible under winner-take-all SMDs.

Protection for Concentrated Minorities

There is a common concern that MMDs might dilute the influence of geographically concentrated minorities who currently benefit from majority-minority SMDs. However, the authors find that proportional voting rules actually protect these groups more effectively. Under STV, "cracking" a concentrated minority across district lines is much less effective because their votes still contribute toward a quota within the larger MMD.

Results and Implications

  • Proportionality: With 3-member districts and STV, near-perfect proportional outcomes become achievable in every state, even under independent commissions.
  • Partisan Limits: Under 5-member districts, the ability of a party to "steal" seats through gerrymandering is curtailed. Nationally, the Democratic maximal seat share drops from ~60% in SMDs to ~53% in 3-member MMDs.
  • The Fair Representation Act (FRA): The authors’ analysis of the FRA’s 3-to-5 member mandate suggests it strikes an ideal balance between proportionality and maintaining geographic cohesiveness.
  • Responsiveness: Contrary to the fear that MMDs create "safe" seats, the research shows that the average vote shift needed to change a seat's outcome decreases as district size increases, making the system more responsive to shifts in public opinion.

Limitations and Future Research

While the study provides a robust defense of MMDs, several areas remain for future inquiry. The model assumes complete ballot rankings, whereas real-world "ballot exhaustion" (voters ranking only a few candidates) might lead to wasted votes. Additionally, while the study addresses partisan proportionality, further work is needed to model the specific legal requirements of the Voting Rights Act (VRA) within an MMD framework. Finally, the study does not account for how the pool of candidates or strategic voting behavior might evolve under a national RCV system.

Summary updated

BibTeX
@inproceedings{garg2022combatting,
  author = {Nikhil Garg and Wes Gurnee and David Rothschild and David B. Shmoys},
  title = {{Combatting Gerrymandering with Social Choice: the Design of Multi-member Districts}},
  booktitle = {EC},
  year = {2022},
  doi = {10.1145/3490486.3538254},
}
Figure