Expanding Representation: Reinventing Congress for the 21st Century
II. Key System Design Choices
Each electoral system is composed of the same constituent parts. The design of those parts can vary significantly, producing different kinds of systems. Proportional systems differ from winner-take-all because of fundamentally different design choices. Within the broad class of proportional representation, different design choices can produce meaningfully different variants.
Policymakers and reformers considering proportional systems in lieu of winner-take-all—whether for federal, state, or local elections—will inevitably face certain choices. To clarify the basic design decisions intrinsic to all electoral systems, the following presents significant benefits and drawbacks to consider, while also offering perspectives on which decisions might be most appropriate for designing proportional systems in the U.S. context. The discussion is intended as directional guidance based on contemporary research and this group’s interdisciplinary expertise.
Electoral systems are made up of four key components, all of which interact to determine how votes are converted into seats.17 Different rule choices give rise to different kinds of systems, and, because choices are expansive across each, no two countries’ electoral systems look exactly alike.
- DISTRICT MAGNITUDE, or the number of seats per district. Districts can be either single- or multimember. The number of seats in a multimember district is limited only by the total number of representatives in a delegation or legislature and so can vary widely.
- ASSEMBLY SIZE, or the total number of representatives in a legislative chamber. Assembly size is also expressed as a ratio of representatives to constituents to reflect the relative size of a legislative body given population size.
- BALLOT STRUCTURE, or how voters can express their preferences on a ballot. Some ballots allow voters to select a single candidate or party, while others allow voters to rank candidates in order of preference.
- ALLOCATION METHOD, or the mathematical rules used to determine winners. Under a plurality rule, a candidate must secure more votes than the next closest rival to win but need not secure an absolute majority. Under a majority rule, a winner must obtain at least 50 percent plus one vote. There are a number of proportional formulas, which generally aim to allocate seats in proportion to votes. These proportional formulas are applicable only in systems that use multimember districts.
What follows is a brief overview of each rule and implications to consider for the design of proportional systems in the United States.
District Magnitude
District magnitude—the number of seats per legislative district—arguably constitutes the most consequential decision for any electoral system. That is because district magnitude, more so than any other variable, determines a system’s potential for proportionality.18 In general, as district magnitude increases, so, too, does the proportionality of electoral outcomes, assuming the use of a proportional allocation method.
However, proportionality is not the only consideration. Comparative research shows that, as district magnitude increases, so does the effective number of political parties.19 That is, the greater the average number of seats per district, the greater the number of competitive parties. “This logic follows something close to a law in political science. Knowing the average number of seats per district, along with the total number of seats in a legislature, will generate a remarkably accurate prediction of the number of nationally competitive political parties in any given country.”20
District Magnitude and Gerrymandering
Gerrymandering is the practice of drawing district lines to generate a bias that favors one party or voter group over others. While gerrymandering is observable across electoral systems, it is especially prevalent under winner-take-all. Single-member districts are uniquely vulnerable to gerrymandering.
One study of fifty-four democracies finds that “not all electoral systems are equally prone to gerrymandering. The problem is inherent in the system of one-seat districts, while it is less serious in multimember districts.”21 As district magnitude decreases, the prevalence of gerrymandering increases; thus, single-members districts, or the lowest possible district magnitude, are especially susceptible. As the number of seats per district increases, drawing lines to advantage one party becomes “prohibitively difficult.”22 Districts with at least five seats appear to be functionally immune to gerrymandering.23
Modeling exercises demonstrate how gerrymandering might be significantly reduced or eliminated by a shift to proportional representation.24 A 2023 study found that “even some of the most gerrymandered maps in the nation would produce fair outcomes if existing gerrymandered districts were combined into multimember districts and seats were allocated to parties proportionally.”25 Illinois’s century-long experiment with cumulative voting, a semiproportional system, offers additional empirical evidence. Researchers in 1920 observed that the system (using three-seat districts for state legislature), introduced in 1870, had virtually eliminated the practice of gerrymandering.26
Proportional systems, because of their use of multimember districts, tend to generate multiparty systems. But variability across countries is significant. Low district magnitudes correspond to fewer political parties, whereas higher district magnitudes correspond to more. Consider that in a district with, say, five seats, up to five political parties have the opportunity to secure one seat; with ten seats, even more parties could potentially win and so may be motivated to form and compete. As we explore in greater detail below, a more dynamic and competitive party system could decrease polarization, improve representation, and facilitate better governance, among other effects. But scholarship also cautions against the risk of fragmentation, when too many parties hamper effective governance through overly “broad and fractious coalitions.”27
Scholars have long debated an “optimal” district magnitude range.28 One study using data from 610 election outcomes in eighty-one countries from 1945 to 2006 concluded that average district magnitudes of between four and eight tend to maximize proportionality in outcomes (that is, gains in proportionality become minimal thereafter) while limiting party fragmentation. They also surpass both winner-take-all systems and proportional systems with higher district magnitudes on various indicators of government performance.29 Other research finds that magnitudes below this range—and especially, two-seat districts30—are neither likely to increase proportionality nor create space for more parties.31
So, what is the right district magnitude to adopt for U.S. congressional elections? The working group recommends drawing districts that have three to eight seats, but it also discussed allowing states some flexibility to experiment slightly beyond the upper end of this range. This window is larger than the electoral “sweet spot” of four to eight seats identified in the academic literature and takes into account particular features of the American polity. A baseline district size of three is helpful to secure the benefits of a proportional system while permitting the broadest set of states to participate. Meanwhile, while comparative evidence indicates a higher district magnitude could lead to greater party fragmentation, features of the U.S. system—including an unusually strong Senate compared to other countries’ upper chambers, and an especially powerful executive—will probably continue to advantage the two largest parties. So, too, will existing restrictive state rules on party formation that hinder the emergence of new parties. Thus, while eight-member districts generally are an advisable ceiling, going above this range in the U.S. context is less likely to result in severe downsides.
Assembly Size
The size of a legislative assembly performs a similar function to district magnitude. Larger assemblies increase proportionality in both electoral outcomes and the effective number of parties in a system.32
Assembly sizes are typically expressed in relation to population size as a ratio of constituents to each representative. The constituent-representative ratio in the United States is nearly 800,000:1, placing it on the extreme end of the assembly-size spectrum globally. Only one other country (India) has a smaller lower chamber relative to population. The U.S. ratio is six times larger than the average democracy.33 State legislatures across the United States also tend to be small compared to their populations.34
Cross-national and domestic research suggests various drawbacks for assemblies that are too small relative to their population (that is, that have very large ratios of constituents per representative), such as weaker electoral connections between representatives and constituents, poorer legislative deliberation, and decreased trust among constituents in their representatives.35 Comparative observations and studies of the United States show that representatives with a larger number of constituents are also more likely to take more extreme positions and prioritize wealthier and narrower interests.36 Research from U.S. state legislatures finds that relatively smaller assemblies result in more negative evaluations of representative government.37
At the country’s founding, each member on average represented thirty thousand constituents, and until 1929, when Congress capped its size, the House gradually expanded to roughly keep pace with population growth. In the Our Common Purpose report, the Commission recommended that the House initially expand by 150 seats for a total of 585, and that, as the framers intended, it continue to expand as the U.S. population grows. This initial increase would still fall below global averages and well below historical ratios in the United States. Nonetheless, while modest, the increase would help to restore representation and place the House back on track to gradually expand. (For a discussion of why 150 seats would help and how this number was arrived at, see The Case for Enlarging the House of Representatives.)38
This group underscores that recommendation. In addition to the benefits described in the prior Our Common Purpose report, this working group considers larger assemblies especially desirable when coupled with proposals for adopting electoral systems that are more proportional. Indeed, the two reforms are often packaged together,39 in particular because an expanded House would allow more states to experiment with proportional systems.40
Consider that Alaska, Delaware, North Dakota, South Dakota, Vermont, and Wyoming each have only a single representative. Multimember districts are foreclosed to each—and so, in turn, is the ability to adopt a proportional system. Another seven states have only two representatives, below the minimum district magnitude recommended here to yield the benefits of a proportional system. Pursuing proportional representation outside this limited set of states is certainly possible absent changes to the size of the House, but increasing the size of the House would at least marginally expand opportunities for implementation and experimentation. The proposal to add 150 new seats, for example, would likely bring Hawaii, West Virginia, and Idaho across this threshold. However, most proposals to increase the size of the House would still leave many of these states with delegations too small to make the switch to a proportional system.
Ballot Structure
Ballots are the point at which voters most directly engage with an electoral system. Ballot structure refers to the kind of basic information a system gathers from a voter, with different structures collecting different kinds of information from voters. It is the vehicle through which an electoral system collects the electorate’s preferences. This is different from ballot design, which refers to the way that information is presented to a voter.
Ballot structures can generally be classified as either categorical or ordinal. Categorical ballots, used in most U.S. elections, allow voters to express a single choice. Voters are given a list of candidates and simply select their top preference. Ordinal ballots, by contrast, allow voters to express multiple choices in order of preference—as with ranked-choice voting. From a list of candidates, voters rank them according to their first choice, second choice, and so on.
In proportional systems, ballots are still categorical or ordinal—selecting one choice or multiple—but the options from there expand.
Transferable systems, like the system recommended in OCP, use an ordinal ballot. Typically, all candidates are listed together, with the order determined by state law. Voters rank candidates in order or preference, and the top-ranked candidates are elected.
By contrast, in most of the world, proportional systems present voters with lists of candidates grouped by party. For instance, Democratic Party candidates running in a district would be listed together; next to them, would be a list of Republican Party candidates; and next to them, perhaps a list of Moderate Party candidates.
In open list systems, voters select a single candidate from among these lists. The number of votes a party’s candidates receive is totaled together. If a party then receives, say, 50 percent of the vote in a six-seat district, three of its candidates win. Which three is determined by the voters: if candidates B, D, and E on the list received the most votes, those candidates are seated. In closed list systems, voters are still presented with lists of candidates—but voters select a party, not individual candidates. If a party receives 50 percent of the vote in a six-seat district, three of its candidates will again win. But this time, candidates A, B, and C on the list are seated, because the party itself chose the order of candidates. Elections in proportional systems that use party lists are akin to combining the primary and general elections.
Certain options are more like current U.S. voting experiences and election administration than others. For instance, open list mirrors the categorical ballot experience used for most U.S. elections today: voters simply select a single candidate, and administrators tally the totals. (The difference lies in how seats are allocated based on those totals.)41 As ordinal ballots become more common with the adoption of ranked-choice voting in many U.S. jurisdictions, transferable systems would replicate the ranking experience. By contrast, closed list would mark a meaningful departure by asking voters to select only a party, not a candidate, and so may generally be considered a poorer match for the American context. Still, among the recent proposals for a proportional system in the United States is a closed list system for the Wyoming state legislature.42 Familiarity may not be the only appropriate measure for ballot structure decisions. For example, policymakers may want to consider the ease of interpreting outcomes, the preferences of election administrators, the costs of implementation, the impact on voter confidence, the cognitive load on voters, and so on.
A robust comparative literature exists that grapples with the benefits and drawbacks of each possible ballot structure: how different options strengthen or weaken political parties, facilitate or hamper coalition-building, enforce accountability among elected officials, and other outcomes. Political scientists hardly have a consensus view, and the group responsible for this paper similarly has no uniform perspective favoring one ballot structure over another, with many group members advocating that states be allowed to experiment.
Allocation Method
When ballots are tabulated, how does a system determine who gets seated? Allocation methods are the rules used to convert votes into a determination of winners and an allocation of seats.
In winner-take-all systems, winners are decided by either a plurality or a majority of votes. For example, in a three-way race in a single-member district, a plurality rule requires that a candidate secure one-third of the vote plus one to win. A majority rule would instead require that a candidate secure at least 50 percent plus one vote to win. In a three-way race, though, a third possible outcome is that no candidate reaches this threshold, and so systems have different approaches to solve for a lack of an initial majority vote. For example, some require run-off elections in which the top-two vote-getters face off in a second round. Instant run-off (also called ranked-choice voting) similarly seeks to ensure that the candidate who wins secures a majority of votes. U.S. state and federal elections use both plurality and majority rules.
When an election has only a single winner, allocation methods, as described above, are straightforward. With multiwinner races, less so. In a six-seat race, a party securing 48 percent of the vote cannot win exactly 48 percent of the district’s seats; seats cannot be fractionalized. Practically, that party should win three of the six seats, given that it nearly won half of the votes. But some standard allocation method is required to make that determination. Proportional systems generally use one of two types of allocation methods: quotas with largest remainders or divisors (also called highest average methods). Both approaches aim to ensure that a party’s share of seats approximates its share of votes.43 STV requires the establishment of a quota for the sequential allocation of seats. Ireland, a well-known example, employs an allocation method termed the “Droop Quota.”44 Different methods impact the proportionality of results, with effects more noticeable when district magnitude is moderate or small (below eight). Certain methods also tend to favor larger parties or groups while others favor smaller ones.45
As with ballot structure, scholars have not reached consensus as to a “best” method—nor did the members of this working group. Instead, what matters as a first-order principle is the necessity of using a standard proportional allocation method from within one of these classes. Given the history of some U.S. states’ use of multimember districts together with a nonproportional (plurality or majority) allocation method—producing a bloc vote system, effectively moving in the opposite direction from what is intended by a switch to proportional representation—policymaking should at minimum explicitly prohibit the use of winner-take-all allocation methods with multimember districts in federal elections. In their place, lawmakers could specify a menu of options from which states could choose.
Proportional Representation and Primary Elections
In almost every state, congressional elections begin with a primary election: a separate publicly held election in which voters decide which candidates each political party will nominate. Proportional representation systems have important implications for how voters can select candidates within a party. As such, shifting to proportional representation, including the STV-type system recommended in OCP, can have significant consequences for how primary elections are conducted, or whether they are necessary at all.
Under either STV or open list proportional representation, a separate primary election is less necessary because voters have the opportunity to choose among several candidates affiliated with each political party during the general election. For example, open list proportional representation can effectively combine the primary election with the general election by allowing voters to vote for any candidate on a party’s list, and thereby collectively decide not only how many seats each party will earn but also which candidates will fill those seats.
Because proportional representation allows the primary election to be folded into the general election in this way, instead of holding a separate primary election to choose nominees or winnow the field, candidates could simply petition to be placed on the general election ballot, with party labels or lists determined by candidate affiliation or by consent of the parties. States could continue to hold primary elections, albeit for multiple nominees instead of just one, but proportional representation makes it possible to combine both rounds into one competitive general election. Doing so could reduce costs for state and local governments. More important, since primaries tend to have lower turnout than general elections and primary electorates differ from general election voters in significant ways, moving to a single all-encompassing election could mean that more voices are factored into the selection of candidates. This latter impact could also change the types of candidates who represent the parties.
The working group did not take a position on whether to forgo primary elections as part of a shift to a proportional system but noted the importance of thinking about primaries as one factor deserving further study.
Constitutionality
Under the Constitution, Congress has the power to prescribe the manner of holding elections for the House and Senate and has exercised that power numerous times.46 The law that mandates single-member districts was an act of Congress in 1967, and Congress itself could repeal or amend it, such as by permitting or requiring that states employ a more proportional system for electing members of the House.47
Additionally, legal scholars argue that proportional representation is compatible with the Voting Rights Act—and is likely to outperform single-member districts at achieving improved minority representation.48 Proportional representation interests this working group precisely because it is a tool for increasing voter voice and representation. As a result, compliance with the Voting Rights Act can be considered the floor and not the ceiling for meeting the needs of an electoral system that provides meaningful avenues for representation. (See “Proportional Representation and the Voting Rights Act” for further discussion of the Voting Rights Act.)
Some variants of proportional representation may not be constitutional. For example, a national party list system may be incompatible with the requirement that House members come from state delegations rather than be elected nationally. However, the proportional multimember districts discussed in this paper are all states or subsets of states and would not present that issue. Likewise, the working group did not discuss the use of proportional representation for Senate elections, as it would likely conflict with the Constitution’s instructions on the allocation of two seats per state elected on separate cycles.