The long-awaited results of the Nov. 2004 range-voting presidential pseudo-election are now available. Paper also includes an approval-voting pseudo-election and some other things! Packed with data from the real world!
http://math.temple.edu/~wds/homepage/works.html go to #82. Study conducted by WD Smith, JN Quintal, and DS Greene. Comments/suggestions welcomed. Important lessons learned include the following: 1. Range voting should be carried out with a range at least as large as [0,20] (assuming integer votes) and perhaps as large as [0,99999]. The choice [0,99] is superior to [0,100]. 2. For either range or approval voting, voters should be given the option of leaving a candidate's numerical vote *blank*, and the total score of each candidate should then be got by *averaging* his non-blank numerical votes. 3. Both plurality and approval voting greatly distort election results (by comparison with range voting), by artificially shrinking the final scores of all candidates besides the top two (or in the case of approval, besides the top three). We're talking one or more orders of magnitude shrinkage. 4. Present-day voters in the USA, if given the choice between plurality and range (or approval) voting, and given 1 minute to think about it, will choose to stay with the inferior plurality system. (They may be more biased against range than approval, but this is not clear.) This proves that considerable media attention and voter education would be required to get range or approval voting enacted. People's main stated reason for disdaining range voting is its perceived complexity, so all voting systems more complicated than range voting are *definitely* expected to be unwanted. 5. At least during initial years, about 80% of approval voters will vote plurality-style. This has successfully been used as ammunition by those opposed to approval voting, to return to the plurality system. But only about 20% of range voters will choose to vote plurality style, making the analogous ammunition unavailable for use against range voting. 6. For psychological reasons, range voters unexpectedly exhibit much more honesty and much less strategy than plurality voters. 7. (The "SQG law"?) Respondents to polls in plurality-voting democracies usually *lie* in such a way as to exaggerate perceived support for third-party candidates. ---- Election-methods mailing list - see http://electorama.com/em for list info