Range voting "Encourages honesty"?

(Executive summary)

In almost all voting systems, including range voting, voters can gain "strategical advantages" by being dishonest in their vote.

For example in the plurality system used in the 2000 USA presidential election, voters who thought Nader was best were urged not to be strategically stupid and instead to vote (dishonestly) for either Gore or Bush.

Polls showed that about 90% of the voters who thought Nader was best in fact did vote for somebody else. I.e. they agreed with that strategy argument! Nevertheless, about 10% of Nader-wanting voters wanted him so strongly they felt compelled to actually be honest and say so in their vote. They paid a big price for that honesty! A lot more among the 97488 Nader voters in the critical state of Florida, favored Gore than favored Bush. But Bush won Florida, and hence the whole USA, by an official margin of 537 votes. If even 5% of those 97488 Nader voters had decided to strategically lie instead of be honest, that would have easily tipped the election to Gore and (most) of them would have ended up a lot happier! (Picture)

With plurality voting you often feel forced to either be a liar, or a fool! That's a heck of a choice!

Range voting, while not perfect in this respect, is comparatively good.

To provide perspective, Allan Gibbard showed a famous impossibility theorem saying that no single winner voting system exists that

  1. handles 3-candidate elections
  2. defines a "vote" to be a rank-ordering of the candidates,
  3. has the property that an honest vote is always a strategically-best vote, i.e. smart voters never feel the urge to lie.
In other words: for every possible voting system based on rank-orderings as votes, there exists a 3-candidate election in which it pays for you to lie in your vote. So there is no perfect voting system in this respect; the best we can hope for is to reduce the degree of imperfection.

(See paper #79 here for an online survey paper on such impossibility theorems, which states, proves, and/or cites these theorems and references the original papers by Gibbard & co.)

Let us discuss some ways Range Voting fulfills that hope.

Range voting still "encourages honesty" in a somewhat weaker than absolute sense. (I use the word "encourage" in a civilian rather than military manner!)

Honesty Theorem. In a 3-candidate range election, nobody can ever gain strategic advantage by dishonestly range-voting as though Alice>Bob, if their honest opinion is Bob>Alice.

This is not a state of perfection. But it is still good and useful, and it is better than

Contrasting Theorem (follows from Gibbard). In a 3-candidate election conducted using any method whatever whose votes are (unlike in range voting) rank-orderings of the candidates, there is always some election scenario in which a voter can obtain a better election result (from his point of view) by dishonestly voting as though Alice>Bob, even if his honest opinion is Bob>Alice.

So range voting encourages honesty better than every rank-ordering-based voting system, including Borda, IRV, and Condorcet methods.

Experimental Fact: Range voting encourages voter honesty

The above arguments have been purely mathematical. But what also matters is psychological and experimental, i.e, how actual human voters behave.

Some critics told me they thought that, in practice, range voters will "strategically" always max-out or min-out every candidate's score to 99 or 0. This exaggeration would effectively convert range voting into approval voting. (And that would not be so bad, since approval voting is still a pretty good voting system.)

But, experimentally, that just does not happen: Most range voters do not vote approval style. Most of them choose to be honest (or at least partially honest) rather than strategically optimal. Specifically, in our range and approval voting exit-poll study of random voters in the 2004 US presidential election (#82 here) we found that the percentage of range voters who vote approval-style (with optional blanks) is 26% as a central estimate (and ≤33% with confidence 90%). In other words:

Experimental fact: At least 74% of range voters vote "honestly" or at least partially so.

This fact stands in gigantic contrast to the fact (top of this page) that about >90% of Nader-favoring plurality voters vote dishonestly! It also has very major (also experimentally verified) consequences: the "nursery effect" is one of them. This giant effect may cause third parties never to be able to survive past infancy under approval voting, but to be able to with range voting. Don't you prefer the latter?


Return to main page