You are here

Mutliple candidate voting

Continuing our discussion of the goals of voting systems, today I want to write about ballots that let you vote for more than one candidate in the same race. Many people have seen Preferential voting where you rank the candidates in order of how much you like them. This is used in Australia, and many private elections such as for the Hugo Awards. The most widely known preferential ballot is Single Transferable Vote and its cousin the instant-runoff. Many election theorists, however view these as the worst possible system. I prefer the Condorcet method with the modification that the cases where it fails, it is declared a tie, or a second type of election is used to break the tie. While it has been demonstrated that all preferential ballots have failure modes where they choose somebody that seems illogical based on the voters' true desires, this does not have to be true when a tie is possible.

Multiple candidate votes would provide a dramatic improvement in the US -- they are already used in many other places. They would have entirely eliminated the question of minor candidates "splitting" or spoiling the vote. There would have been no question in Florida of 2000, with Al Gore defeating George W. Bush (and at least by the popular vote, some feel that Bill Clinton would have lost to George Bush the elder, and there's strong evidence the electoral margin would have at least been smaller.) This is in fact what prevents them from being used -- there is always somebody in power who is going to conclude they would have lost has there been a multi-candidate ballot in place. Such people will fight it harder than advocates push it.

Small party candidates want it because it gives them a chance to be heard. Voters who like them can safely express that preference without fear of "spoiling" the race among the frontrunners. Given that, small candidates can eventually become frontrunners. In the 2 party system, as we've seen, any time a minor candidate like Ralph Nader gets popular enough that he might actually make a difference, the result is cries of "Ralph, don't run" and a dropping of support from those who fear that problem. However, the key factor to note is that many of these systems do cry out for computerized voting machines. I have been involved in some STV elections done on paper and have discovered that voters routinely get confused by them. They get confused about how they work, and they get confused about how to mark their ballots. They routinely make errors such as trying to rank two candidates the same (most systems demand a fully ordered ranking) engaging in defensive voting and other moves that indicate a lack of understanding of the system. In spite of the fact that preferential ballot systems are designed to allow one party to run more than one candidate, this rarely happens. In the Hugo awards, writers who have managed two nominations will still sometimes pull one even though, in theory this is foolish. (In practice, the misunderstanding of the system by voters sometimes makes it not so foolish!)

Computerized voting can solve a lot of this. For example, the instant runoff can be demonstrated to them right on the screen.

  • Ask the voter to pick their first choice.
  • Then remove that choice from the menu and say, "If that person is eliminated, who would you pick next?"
  • Repeat

At the end they can see their choices clearly ordered and finally confirm. They can also be informed of what it signifies to not rank a candidate (which varies by the rules of the election.) This provides the best opportunity for correct voting by an electorate that is informed about how they are making their selections. On paper this doesn't seem to happen.

One of the reasons I like Condorcet, as well, is it is easier to explain to voters than most of the other systems. "The winner beats every other candidate in a 2 way race based on how many people ranked them higher than the other candidate and vice versa. If nobody beats everybody else, we go into the tie situation among the 3 or more who beat all but one." The ties are rare so the rules are easy to understand. Voter ability to understand how the votes will be counted is important.

For this reason I am becoming more and more of a fan of the simplest of all the multi-candidate ballots, known as Approval voting. You simply list the choices you find acceptable, as many as you like, with no ordering. This is so simple that it also works just fine with simple paper ballots, and is also very easy to understand, both very positive features.

The main strike against Approval in most people's minds is that it seems to deliberately throw away important information. You can vote for Al Gore and Ralph Nader but you can't specify that you might much prefer one to the other. Right now people will seek multi-candidate ballots for a couple of reasons. One is to say, "My real choice is Al Gore, but I want to send a message that I support Nader and the Green party" Another is to say "Nader is who I really want, but because I know he can't win I want to send my support to Gore." Approval does not allow this to be expressed, though it generally is going to produce the right result, because in an election like 2000, Nader's approvals would have had no chance to beat the two major candidates. However, we would not have learned how many were in each of the two camps above. That knowledge can actually be very helpful in supporting the growth of 3rd party candidates to contender status.

On voting machines it is also possible to allow voters to effectively do both -- rank the candidates for the opinion poll, but count them unranked for the real election. In that case the main issue is that you clearly should not rank somebody who might win if you think they would be unacceptable for the position.

Computerized voting machines continue to gain traction because they can explain more than just more intricate voting systems. They can also assist the disabled, and point out errors to voters. For example, while there is some controversy over how to do this fairly, it could have made a lot of sense in the recent Sarasota election if machines told users, "You did not cast a choice in the race for member of the house of representatives. That race is traditionally the most widely voted in in this district, so please confirm you don't wish to make a choice in that race." Limited error checks like these can compensate for errors in ballot design, though of course we must not allow biased error checks even though they might make sense, such as, "We notice you have voted entirely for candidates of the Democratic party, except for President where you voted for Patrick Buchanan."

While unauditable DRE voting machines are clearly a bad idea, voting machines which generate real paper ballots and assist the voter continue to have real potential for improving how well we meet our goals for election systems.


I'm also long a fan of approval, as you know.

Condorcet methods have the slight advantage that if you rank two candidates the same, it's no problem.

> The most widely known preferential ballot is Single Transferable Vote and its cousin the instant-runoff.
> Many election theorists, however view these as the worst possible system.

Can you give me a reference for this? You correctly point out that STV is pretty complicated and can confuse voters; does it have worse problems than that? Several countries are using it; if there are serious problems it seems like someone should have run into an example by now.

Wikipedia notes that strange things can happen if a party fields too many or too few candidates, but (particularly in the case of presidential elections!) this doesn't seem like a serious bug.

Ping's link above has some good material on problems with STV.

Now the reality is that all the systems fail (if they need to provide just one winner) and most of the failures are in somewhat pathological cases, but some fail more than others, so it is better to use systems that fail less often in real cases, as well as systems that are easier to understand. Approval is easiest to understand and implement. Condorcet is also fairly easy to understand (except in ties.) Weighted systems are probably too complex for too many voters.

It is however worth noting that some of the "failures" are definitional. For example, the Condorcet Criterion (that the winner beats everybody else in a head to head race) is not met by many of the non-Condorcet systems. Some systems will choose a winner who is everybody's 2nd choice and others won't, and it's subjective if this is a bug or a feature. (IRV does not choose a candidate who is everybody's 2nd choice but few people's first choice, while it does choose that candidate if they are not last in the first round.) Some criteria, like monotonicity make sense to everybody, and IRV fails this. Supporters of IRV just feel it has other virtues to make up for the non-monotonicity. However, IRV is inferior to Condorcet in all criteria. See Voting system.

Add new comment

Subscribe to Comments for "Mutliple candidate voting"