3

It seems that in most real life examples of range voting, they are in a niche field where either:

  • There are objective measurable criteria for how to determine ranges (physical distance, time distance), or

  • Complicated scoring system that at least pretends to be objective via complicated scoring rules.

How do the proponents of range voting propose to deal with the complication of "what does 0-10 mean objectively" for political applications of range voting, where the scoring criteria is basically a fully subjective "like-dislike" or "approve-disapprove" continuum?

| improve this question | |
  • 1
    If voters want to vote optimally they should give every candidate either the maximum or minimum, otherwise they can use whatever subjective scoring they want, I don't think there is any need for additional "objective scorings". – Fela Winkelmolen Jan 5 '13 at 13:47
  • @FelaWinkelmolen - then you devolve to simple approval voting. What you stated seems to merely argue that range voting should not be used at all, and have approval voting instead, 100% of the time. Hardly an argument I would expect from the proponent of range voting? :) – user4012 Jan 5 '13 at 13:54
  • either that or it is okay to have non-optimal subjective voting – Fela Winkelmolen Jan 5 '13 at 13:55
  • 1
    @DVK: RV does not have to "devolve to simple approval voting". For example, I could find on the ballot Mr. IdeologicallyPerfectButUnelectable, Mrs. AbsolutelyRepugnant, and Ms. NotExactlyGreatButElectableSoSheCanAtLeastAdvanceMyAgenda. I would give, respectively, 10 out of 10, 7 out of 10, and 0 out of 10. – xuinkrbin. Oct 16 '13 at 14:30
  • 1
    @xuinkrbin. I think your numbers are out of order. Mrs. Repugnant should get 0, no? – endolith Jan 4 '17 at 19:09
2

I would also like to hear an answer from a hardcore Score Voting enthusiast, but here are some possible partial answers:

  • iSideWith.com will give you a list of candidates and the percentage you agree with their platforms/positions on issues, which could be converted to a list of scores to put on your ballot. Presumably if score voting were adopted by some jurisdiction, similar websites would pop up to help you figure out what your (honest) ballot should be, and the results would be comparable between voters.

  • There are several variants under the umbrella of Score/Range voting, and I think systems that use descriptive words for the scale instead of numbers can still be considered score systems, like using a Likert scale from "Strong oppose" to "Strong support". I think this would make votes more comparable between voters, and this is similar to Majority Judgment, a sibling of Score, which uses a scale from "Excellent" to "Poor", for instance. However, this long page claims that:

    the "best" scale for human voters should have 10 levels and consist entirely of nonnegative numbers ordered increasing from left to right, and equispaced, with the two endpoints only of the scale "anchored" with descriptive words

    I haven't had time to read it and decide whether I agree or not.

  • I think reducing the number of categories or decreasing the resolution of the ballots makes them more comparable between voters? "3 out of 5 stars" means pretty much the same thing regardless of who voted it. "63% approval" doesn't necessarily mean the same thing from one voter to the next.
  • Preferential ballots are even less comparable between voters, because their endpoints of "favorite" and "least favorite" are not objective measures, and they destroy information about the degree of preference between the endpoints, condensing it all into a flat list. Score ballots of A=10 B=9 C=0 and A=10 B=1 C=0 and A=6 B=5 C=4 are all flattened to A > B > C in a preferential system, but they aren't equivalent.
    • In psychometrics, psychologists measure the attitudes of people using questionnaires, and it's understood that preferential questions ("ipsative measures") cannot be meaningfully compared between people, while ratings questions ("normative measures") can. I think we can apply the same principles to ballots which measure voter attitudes about candidates.

    • "In summary, one may state that scores originally obtained as ipsative measures may legitimately be employed only for purposes of intraindividual comparisons. Normative measures may be employed for either interindividual or intraindividual comparisons." http://psycnet.apa.org/psycinfo/1971-01501-001

    • Early Krosnick research ("Maximizing Questionnaire Quality", 1999) saw ranking questions as having greater predictive validity, but a number of studies since, include his own later research, show rating questions as having greater validity (Krosnick, Thomas, and Shaeffer, 2003; Maio, Roese, Seligman, Katz, 1996). http://blog.verint.com/ranking-questions-vs-rating-questions

| improve this answer | |
1

I used to support a Likert-style approach from strongly-disapprove to strongly-approve, but that still aims for some objectivity of meaning in that a pool of all acceptable candidates would get all medium-to-high scores, giving extra weight to those willing to normalize their scores.

Normalizing is the process of taking the lowest scoring candidate and giving them a 0 and giving the highest scoring the top score and adjusting everything else relative to that.

The sensible way to score is to simply encourage / educate voters that they will maximize their influence by normalizing their scores. Basically: just explain that the scores are worst-to-best relative to this pool of candidates, not relative to any objective anything.

| improve this answer | |

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .