Bart Ingles wrote:
It would be interesting to know how many ballots were exhausted because the voter voluntarily ranked only one or two candidates, versus the number exhausted because all three permitted choices were eliminated.
Put another way, how many three-ranked ballots did not contain a vote for either the winner or the runner-up?
I believe the numbers you are looking for are contained within the published results at:
http://www.sfgov.org/site/uploadedfiles/election/results.htm
The RCV tally was only needed in Districts 1, 5, 7 & 11.
In District 1, ~15% of the ballots were exhausted with 7 candidates. In District 5, ~33% of the ballots were exhausted with 22 candidates. In District 7, ~30% of the ballots were exhausted with 13 candidates. In District 11, ~26% of the ballots were exhausted with 8 candidates.
This shows what I would have considered an expected correlation between the numbers of candidates and the percent of exhausted ballots...i.e. as the number of candidates increase, so do the number of exhausted ballots when the voters are limited to the number of candidates they are allowed to rank.
Of course, a sample size of 4 is far to small to make this claim with any degree of certainty.
What I would consider an interesting analysis is whether the failure of Independence of Clones in this implementation of IRV because the voters were not allowed to rank all available candidates would have changed the result in any of the elections.
The hard question to answer is how to come up with a credible case that two or more candidates were clones and then to handle the case where four or more clones were in the same race.
I've got all the relevant data & links archived in a convenient location on my website.
http://www.ericgorr.net/library/tiki-index.php?page=SanFrancisco2004
-- == Eric Gorr ========= http://www.ericgorr.net ========= ICQ:9293199 == "Therefore the considerations of the intelligent always include both benefit and harm." - Sun Tzu == Insults, like violence, are the last refuge of the incompetent... === ---- Election-methods mailing list - see http://electorama.com/em for list info
