Yesterday I argued that one of the benefits of expanding the House of Representatives would be to make it easier to use proportional representation, but I didn't go into detail on how that might work or what its benefits would be. Today I will demonstrate how single-member districts are susceptible to gerrymandering and inherently unfair to political minority voters. I will show how proportional representation using single transferable voting gives much fairer results, works for all kinds of elections, and complements its better-known single-winner variant.
Gerrymandering and Wasted Votes
To begin with, lets round up a list of why the current system of single-member plurality is ill-suited for electing the House and state legislatures. The most obvious reason, as the map above shows, is that it is easy for the people who draw districts to manipulate them in order to favor one party or the other. This is particularly damaging for Democrats because we tend to cluster in the major cities. It is easy for people who want to help Republicans to pack Democratic voters into one or a few districts where they will win by huge majorities, leaving Republicans with majorities in the suburban and rural districts. This is also why allocating electoral votes by Congressional districts is a terrible idea and should absolutely never be called “proportional” as some media outlets seem to think. As UC Davis political scientist Matthew Shuggart puts it, “It is a power grab especially when employed only in states where the Republican candidate tends to have a better geographical spread of the votes in the state.”
Gerrymandering and so-called proportional Electoral College schemes work because single-member plurality voting treats the barest plurality and the largest majority in exactly the same way. You don't earn extra seats in Congress by winning a district with 80% instead of 51%, and you don't even have to earn 50% + 1 vote, just more than any other candidate. Democratic voters who find themselves in a suburban or rural district with a Republican majority are unrepresented, and they may as well stay home on Election Day. Likewise, Republican voters in Democratic districts need not bother to vote. Excess votes for sure winners and all votes for sure losers simply don't affect the outcome of the election; they are, in a quite literal sense, wasted. Slate’s Reihan Salam puts it bluntly: “As a conservative living in New York City, my vote for Congress is essentially a socially approved form of venting.” Many voters, then, have no incentive to vote for their candidate of choice or even to bother voting at all. And yet, shouldn't our voting system let people vote for their preferred candidate and treat all votes equally, so far as possible? In an ideal voting system, it shouldn't matter where you vote, just whom you vote for.
Proportional Representation
First of all, there are many forms of proportional representation (PR), but all share certain features. The most important is that districts elect more than one representative. In this way, vote wastage is limited because surplus votes can elect extra representatives, and votes for less popular candidates or parties can still elect some representatives if they meet the threshold. This makes it difficult or impossible to use gerrymandering to affect the outcome of an election and encourages voters to choose their actual preferred party or candidate without fear of casting a useless vote.
Looking at this map produced by electoral reform organization FairVote, it is clear that most democracies around the world either use straight proportional representation or at least include a proportional element in their national elections. The major exceptions are the US, UK and countries influenced by the Westminster system. Even in the Westminster world, though, there are subnational elections using PR. All of the UK's devolved parliaments and the London Assembly use proportional systems, as do local elections in Northern Ireland and Scotland. Australia's Senate uses PR, as do the upper chambers of all the state parliaments except Tasmania, where it is used in the lower chamber. That leaves the US and Canada as just about the only full democracies that do not and have never used PR for any major elections.
Single Transferable Vote: PR American Style
The question then becomes which system would be the best fit for American elections. Looking at FairVote's map, it seems like the most common choices are open and closed party lists. In a party list system, voters are presented with a ballot like this one designed by Professor Douglas J. Amy of Mount Holyoke College for a US House election.
In an open list PR election, votes for candidates pool to their respective parties. Party vote totals are then run through a simple algorithm to determine how many seats each party will be awarded. More votes, more seats, but every significant party should get a seat. Seats are then filled by the strongest candidates determined by their number of votes.
Open list PR is the obvious choice for parliamentary elections where party matters more than personality. It ensures that the number of seats won by each party reflects its strength in the electorate while allowing voters some control over exactly who ends up in parliament. But the US is emphatically not a parliamentary democracy, and we do care exactly who ends up in Congress. Besides, at the district magnitudes (number of people elected from a district) that we would be working with, list PR tends to waste a lot of votes if there are more than two significant parties.
Better for US purposes is the single transferable vote system (STV). You may already be familiar with the single-winner variant, which goes by many names including instant runoff voting (IRV) and ranked-choice voting (RCV). Whatever you call it, the system is straightforward. Voters rank the candidates 1-2-3 in their order of preference. If a candidate earns 50% +1 vote, she wins outright. If not, the weakest candidate is eliminated, and his votes go to the next preferred candidate. The process continues until one candidate has a majority or there are only two candidates left. This process works well for electing single winners to offices like governor or president, but it does little to alleviate gerrymandering and wasted votes.
In the multi-winner variant, the number of votes needed to win is equal to 1/(s+1) + 1 vote. So for a district electing 3 members of Congress, it is 1/(3+1) + 1 vote or 25% + 1 vote. Any candidate who earns that many votes on the first count is elected, and votes over that amount are transferred. This way, candidates only keep as many votes as they need, and wasteful surplus votes become a nonissue. It is difficult to explain the process in words, so for the visually-inclined here is YouTube's explainer-in-chief CGP Grey.
STV has many advantages for American elections. It continues our proud tradition of voting for candidates, not parties. It can handle small district magnitudes thanks to the use of preferences and transfers, and it treats independent candidates on the same footing as those running on party labels. It's main disadvantage is that it takes some explaining, but the gist is simple enough. You vote your preferences, and the system does its best to elect the candidate or candidates you like.
STV is used successfully in elections in Ireland, both the Republic and Northern Ireland, Scotland, Malta, New Zealand, Australia, India (for the indirectly-elected upper chamber), and Cambridge, Mass. Historically, it was used by many more city councils in the US:
The first U.S. city to adopt at-large ranked choice voting for its city council was Ashtabula, Ohio in 1915. During the first half of the 20th century, ranked choice voting spread rapidly as part of the progressive movement. At its peak, some two-dozen cities adopted it, including Cincinatti, Cleveland, Boulder, Sacramento, and even New York City. New York City continued to use ranked choice voting for its school board until 2002 when those school boards were abolished.
Naturally, fairer results angered the establishment, leading to repeal efforts where PR let undesirables (e.g. black people and minor parties) win representation.
Implementing STV
FairVote has already drawn up plans for multi-member districts in every state. Let's have another look at Ohio, where Republicans swept House elections in a very evenly-split state.
All FairVote did was combine Ohio's 16 existing districts into 4 superdistricts. 2 would elect 5 representatives, and 2 would elect 3. Using partisanship data, they project that Ohioans would elect 7 Republicans and 7 Democrats, with the remaining 2 seats as swing. Every district would be represented by at least 1 Republican and at least 1 Democrat. That's a far cry from the actual 2014 result, which returned 12 Republicans and only 4 Democrats and left rural Democrats and urban Republicans unrepresented. Note that FairVote's plans use existing, heavily gerrymandered districts. Under STV, district lines just matter less.
STV works for elections at every level, from city councils and school boards to the federal House of Representatives. It works for nonpartisan and partisan races and provides good representation for both. Even at just 3 or 5 seats per district, it is vastly more fair than our existing system. It is also the natural complement to its single-winner variant, which has been gaining momentum from California to Maine. Expanding the House would ensure that superdistricts are manageable and allow more 5-member districts. STV is easy to use. Voters just pick their favorite candidates, and the system does the rest, translating votes into seats with minimal waste.