Tom McIntyre said: > The alternatives are to enforce > strict ranking of the candidates the voter chooses to list, or to > enforce strict ranking of all candidates on the ballot. <snip> > but I'd like to know if there's consensus here on which of > these alternatives would be better in actual practice.
Well, the consequences of allowing equal rankings and how to deal with them have been discussed ad nauseum, particularly in the context of Condorcet methods (more on that later). In general, the opinion seems to be that if a voter feels that the best way to protect his interests is by voting two candidates equal then more power to him. Indeed, many people on this list like Approval Voting, in which you sort the candidates into two categories: approved and not approved. Some also like Cardinal Ratings: Give each candidate however many points you like on a scale of 0 to 4 or 0 to 10 or -1 to +1 or whatever other scale is decided upon. It's often argued that such methods are equivalent to Approval Voting, in that your best strategy is to give each candidate either the top rating or the bottom rating. In any case, even though most people here hold that Cardinal Ratings and Approval are strategically equivalent, if the public really preferred Cardinal Ratings and the _option_ of giving intermediate scores I don't think many of us would complain. In Condorcet methods, the handling of equal rankings becomes tricky. If somebody considers two candidates equal then in pairwise contests that person obviously has no say. If there's a candidate who wins all of his pairwise contests then there's no problem. But, if nobody wins all pairwise contests then there's a subtle but important question to resolve. Say we use a method where we drop the weakest pairwise defeats until somebody is "undefeated", and we have a situation where A>B>C>A. Say also that there are 100 voters. If A beats B 51-49, B beats C 45-40, and C beats A 55-45, A clearly suffers the most decisive defeat. But, who suffers the weakest defeat? Some would say B suffers the weakest the weakest defeat. He was beaten by a margin of 2 points. Others say "That's true, but an absolute majority came out against B. An absolute majority did NOT come out against C. Sure, B won a significant victory among those who expressed a preference between B and C. Still, because a lot of people sat out that contest, it doesn't carry the same weight as the ABSOLUTE MAJORITY who came out against B in the A-B contest." Those who express the above sentiment then say that instead of looking at margins of defeat, we should look at the absolute number of people who came out against a candidate. In that case, the 45 who came out against C constitute a weaker defeat for C than the 51 who came out against B. Looking at the absolute number of people who voted against a candidate is known as the "Winning Votes" method measuring the strength of a pairwise defeat. Looking at the margin is known as the margins method. The above discussion of absolute majorities is the normative, or political argument in favor of the "winning votes" (wv) method. The political argument in favor of margins is essentially that winning by a wide margin is a mark of excellence. I'm probably not making the case for margins very well, mostly because I'm not familiar with the intricacies of the debate. There are also technical arguments concerning which method gives more incentives to vote insincerely. I'll stay on the sidelines of that argument. Anyway, you won't find much in the archives on equal rankings per se, but you'll find a lot of discussion of matters related to equal rankings. Alex ---- For more information about this list (subscribe, unsubscribe, FAQ, etc), please see http://www.eskimo.com/~robla/em
