Preference Voting vs. Cumulative Voting

by Rob Richie

The Center for Voting and Democracy -


In recent years, proportional representation voting systems have moved from being an interesting political theory to discuss to viable political options for American elections. Thus, it becomes more important to make sure that one implements the best system for a certain level of elections.

For local elections, the main choice likely will be between two candidate-based systems: cumulative voting and preference voting (e.g., single transferable vote). Cumulative voting is easy to explain briefly, and thus a good choice to illustrate a PR system. But preference voting is the system local activists should seek to implement. Here is why.

Voters' Perspective: Preference Voting is Easier

Some people believe preference voting is too complicated, but the evidence strongly suggests otherwise. As one example, Northern Ireland used preference voting for a few elections in the 1920s; in the first election, there was an 89% voter turnout and a less than 1% invalid ballot rate -- lower than many elections here today. This history helps explain why the February 1995 peace accord proposes to restore preference voting for new legislative elections; it already is used in Northern Ireland for local and European parliamentary elections and has helped diffuse tension between Catholics and Protestants.

If voting "1, 2, 3" is as easy as it would seem, what does it mean in comparison to cumulative voting? In a five-seat race, the voter's calculation with preference voting is a simple one: which candidate do I like best, which do I like next best and so on, knowing that a lower choice will never help defeat a higher choice and also knowing that ranking a lower choice might help that candidate defeat a candidate you dislike.

With cumulative voting, it's not so easy. You like one candidate best and yes, you could put all your votes on that candidate. But what if you also like some other candidates? How do you divvy up your votes? What if you think your favorite candidate might not win? What if two strong candidates are appealing to the same community of voters that only has enough votes to win one seat?

Preference Voting Makes Votes Count

The last question of the possibility of "too many" candidates points to a serious problem with cumulative voting: wasted votes. If voters make the "wrong" calculation -- and how can they avoid it if not part of a disciplined political organization that is rare in U.S. politics -- then one candidate could end up with far too many votes and others with too few. We have already seen this happen in some cumulative voting elections, like the 1991 city council election in Peoria, Illinois, where two black liberals split the black vote.

One way to think of it is to compare cumulative voting to a traditional party list proportional system. With party list, if a party gets more votes, then it can earn more seats -- there is a direct correlation. But when voting for individual candidates, more votes can do no more than just elect that one candidate.

Preference voting avoids this problem by creating a dynamic similar to party list PR. Votes beyond what is necessary to win will simply be transferred onto philosophically similar candidates (as determined by individual voters) and still count.

Preference Voting Encourages Competitiveness

One natural result of the wasted vote dynamic is to have political forces try to run only as many candidates as they think can win, or perhaps one extra. When Illinois had cumulative voting in three-member districts for state legislative elections, there were usually only four candidates in the general election (until they made a rule change requiring parties to nominate two people, there often had been only three for three seats!).

However, as much of a problem that such candidate limitation is in partisan elections, it is more of one in non- partisan elections. First, it is harder for loosely organized groups to do than for parties. Second, if only some political forces are doing it, then they get an unfair boost. Third, if everyone does it, then voters will not be too thrilled, even if the results are "fair": ratifying pre-election choices of party-like leaders is not much fun.

Finally, cumulative voting has a pro-incumbent bias in both partisan and non-partisan elections. Supporters of a party or interest group are likely to discourage challenges to incumbents favoring their position and not to risk giving votes to the challengers who do run.

Preference Voting Helps Women Candidates

When there is no party discipline that leads to a party running a certain number of candidates and. asking supporters to vote for all of them, cumulative voting naturally leads to many voters putting all their votes on one candidate -- just as many only vote for one ("bullet vote") in at-large elections, voters often will only vote for one or two candidates because they do not want a lesser choice to help defeat their top choices.

Bullet voting, just like single-member districts, historically hurts women candidates. When voters have only one vote, men have gotten more of those votes. That might be changing, but it still could be a problem for women candidates.

Preference Voting Builds Coalitions

Bullet voting is more likely to have other negative effects. No matter how diverse a city is, its voters will tend to be multi-dimensional and tend to have shared interests with others who in other ways are different from them. With preference voting, coalition building is an obvious result. After voters cast their first and/or second choices on those candidates most like them, they probably will calculate which of the remaining candidates is the one they like next best. The result is that candidates will reach beyond their base, voters will look beyond candidates most like them and groups of voters too small to elect someone on their own will find at least some candidates responsive to them.

Cumulative voting has no such incentives. In the racially polarized southern county of Chilton County (AL), for example, almost all black voters put their seven votes on the black candidates. It was an obvious choice, and both black candidates and white candidates knew it would likely happen. The winning black candidate did reach out to white voters, but received votes from only 1.5% of them,

Preference voting would create a clear incentive for at least some white candidates to actively court black voters in order to pick up transfer votes in Chilton County, while white voters would take a closer look at black candidates. Such outreach would simply be smart campaigning.

Preference Voting Discourages Negative Campaigns

If many voters are likely to put all their votes on one candidate, then two candidates seeking the support of those voters have an unfortunate incentive to trash their opponent -- it is an all-or-nothing game, just like single-member districts.

With preference voting, candidates will still need to differentiate themselves from the other candidate to gain support, but they cannot be too negative if they want to gain the second preferences of these voters. And they might very well run with other candidates on a slate, telling supporters to make sure to put one or the other of them first and others next; in the non-partisan preference voting elections in Cambridge, slates are very important.

Popularity Among Voters

The limited record of cumulative voting in the U.S. is a decent one -- not remarkable, but good. But there is one important test: it was defeated 80% to 20% when put before voters in Cincinnati in 1993. There were many reasons for the lopsided vote, but it is instructive that preference voting only was defeated 55% to 45% the two times it faced those same voters in Cincinnati in 1988 and 1991.

Preference voting was adopted by referendum in two dozen U.S. cities earlier this century. Of the first 25 attempts to repeat it in these cities -- by political forces that often had majority support among voters - only two were successful. But eventually the anti-reform forces outlasted the reformers, helped by running negative campaigns against unpopular minorities and by the long ballot-count. Now we can computerize the count, and I believe there is more tolerance of diversity.

Those focusing on the comparable winning thresholds with cumulative voting and preference voting might overlook some of these more subtle campaign dynamics that point in preference voting's favor. But elections do more than establish who wins representation-. the campaigns themselves are times for building community, sharing information and identifying and solving problems,

Yet even from the simple standard of winning seats, preference voting is better. Cumulative voting is not a semi- proportional system because of a higher winning threshold. Rather, cumulative voting inevitably wastes more votes. "Semi-proportional" really means "semi-fair." We can do better.

Politics watch