Multiwinner Voting Methods
Adapting single-winner voting concepts to multiwinner contests presents a significant challenge in political science and mathematics. While single-winner elections focus on a lone victor, multiwinner systems must determine a group of candidates—a committee—that accurately reflects the will of the electorate. These methods generally split into two primary categories: ranked voting, where voters order candidates by preference, and approval voting, where voters indicate which candidates they find acceptable.
Depending on the system, voters may cast a single vote or multiple votes, and candidates may run as individuals or as part of grouped slates (party lists). Some systems prioritize proportional representation, ensuring minority groups gain seats, while others focus on plurality, where the most popular individuals win regardless of party balance.
[ไม่มีภาพประกอบ]
Key Facts
- Multiwinner voting can be based on ranked preferences or simple approval.
- Single Transferable Voting (STV) is designed to create balanced groups and ensure minority representation.
- Thiele's voting rules use weight-based sequences to score committees based on voter approval.
- Computational complexity varies; some methods like MNTV are solvable in polynomial time, while others like PAV are NP-hard.
- Condorcet methods seek a committee that can defeat any other possible committee in a head-to-head majority comparison.
Ranked Voting Systems
Ranked voting allows voters to express a hierarchy of preferences. One of the most prominent examples is Single Transferable Voting (STV). STV aims to elect a mixed, balanced group of members by transferring votes from candidates who are either unelectable or have already secured enough votes to win.
A critical component of STV is the quota, a minimum number of votes required for election. This mechanism prevents a single dominant group from taking all seats unless they hold an overwhelming majority of the total votes or the district magnitude (the number of seats available) is very small.
Approval Voting for Committees
In approval voting, voters simply mark all candidates they approve of. For multiwinner contests, this is often expanded into weight-based rules. In 1895, Thiele introduced a family of rules where a committee's score is determined by a sequence of weights (w1, ..., wk) based on how many approved candidates are in that committee.
Common Thiele-based Rules
- Multiple Non-Transferable Vote (MNTV): Also known as plurality-at-large approval voting, it uses a weight vector of (1, 1, ..., 1).
- Approval-Chamberlin-Courant (ACC): Uses a weight vector of (1, 0, ..., 0), meaning a voter gives one point to a committee if it contains at least one approved candidate.
- Proportional Approval Voting (PAV): Uses a harmonic progression (1, 1/2, 1/3, ..., 1/k) to reward diversity in representation.
Other specialized approval methods include Phragmen's voting rules, the method of equal shares, and minimax approval voting. From a computational standpoint, MNTV is efficient (polynomial time), whereas ACC and PAV are NP-hard, meaning they are computationally difficult to solve as the number of candidates increases.
Positional Scoring Rules
Positional scoring rules assign a numerical value to a candidate based on their rank. In multiwinner contexts, these scores must be adapted to evaluate entire committees.
Implementation Methods
- Single Non-Transferable Vote (SNTV): A generalization of first-past-the-post where each voter casts one vote for one candidate. The top k candidates win. This is computable in polynomial time.
- Multiple Non-Transferable Vote (Bloc Voting): Voters cast up to k votes for their top candidates; the k candidates with the most votes are elected.
- k-Borda: Candidates are scored using the Borda count (a system where points are assigned based on rank), and the top k scorers are selected.
- Borda-Chamberlin-Courant (BCC): Each voter assigns the Borda count of their most preferred candidate within a committee to that committee. This method is NP-hard.
Condorcet Committees
A Condorcet winner is a candidate who would win a head-to-head match against every other candidate. Adapting this to committees has led to three primary theoretical approaches:
- Fishburn's Adaptation: A committee is a Condorcet committee if a majority of voters prefer it over any other possible committee. This often assumes dichotomous preferences (ranking committees by the number of approved members). Checking this is coNP-complete.
- Gehrlein and Ratliff's Adaptation: A committee is a Condorcet committee if every member inside the committee is preferred by a majority over every candidate outside it. This leads to stable rules such as the multiwinner Copeland's method (based on external defeats) and the multiwinner Minimax Condorcet method (based on external opposition).
- Elkind, Lang, and Saffidine's Adaptation: A winning set is defined as one where, for every candidate outside the set, a majority of voters prefer at least one member of the set over that outside candidate.
| Method | Voting Type | Key Characteristic | Complexity |
|---|---|---|---|
| STV | Ranked | Vote transfer & quotas | Varies |
| MNTV | Approval | Equal weights (1, 1, ... 1) | Polynomial |
| PAV | Approval | Harmonic weights (1, 1/2, ... 1/k) | NP-hard |
| SNTV | Ranked/Positional | Single vote for multi-seat | Polynomial |
| BCC | Ranked/Positional | Borda count of top member | NP-hard |
Frequently Asked Questions
What is the main difference between SNTV and STV?
SNTV (Single Non-Transferable Vote) allows each voter only one vote, which prevents one party from sweeping all seats but does not guarantee proportional representation. STV (Single Transferable Vote) uses ranked ballots and transfers votes from unelected or surplus candidates to ensure a more balanced, proportional result.
What does it mean for a voting method to be NP-hard?
In computational terms, NP-hard means that as the number of candidates and voters grows, the time required to find the mathematically optimal winner increases exponentially, making it computationally impractical for very large elections without using approximations.
How does Proportional Approval Voting (PAV) work?
PAV uses a harmonic weight sequence (1, 1/2, 1/3, etc.). This means the first candidate a voter approves of provides a full point, the second provides half a point, and so on. This diminishing return encourages the selection of a diverse committee that satisfies different groups of voters.
What is a Condorcet committee?
A Condorcet committee is a group of winners that satisfies a majority-based criterion. Depending on the definition used, it is either a committee preferred over all other possible committees by a majority of voters, or a set where every member is preferred over every non-member by a majority.