[EM] Plurality with Condorcet polling is equivalent to Condorcet. Condorcet for 2012!

MIKE OSSIPOFF nkklrp at hotmail.com
Sun Oct 23 13:18:11 PDT 2011


Juho--
 
Of course there won't be many polling locations Right now, I know of only one--my own local poll, which
might turn out to be the only one. In that case, I'll have no choice but to infer about the entire country
from a poll in my own town.
 
With luck there might be a few (three or four?) other people on EM who are willing to conduct polls in their
cities around the country. Maybe, maybe not. I'm not counting on it. If mine is the only poll, then that's just
how it is.
 
About the count method, the important thing is to just try to find a CW. If there is, instead, a circular tie, then
maybe examination of the rankings will reveal offensive order-reversal, and allow inference about a CW.
 
If there's a natural circular tie, then I'll look at first choices in the rankings, and, using an estimate of relative
positions in the left-right spectrum, I'll find the candidate such that equal numbers of 1st choice ballots have
been cast for candidates to his/her left and right. That candidate will stand in for the CW.
 
The purpose is to inform strategic voting in Plurality. Will the CW be a Democrat or Republican? Maybe not.
 
Look at all the Internet polls in which Nader consistently won. 
 
Selection bias? Don't people with more money
have more computers? Aren't people with more money more conservative?
 
Ballot-stuffing:
 
Arguably, more honest people prefer more honest candidates. More cynical people who vote for more cynical candidates
would arguably be more likely to ballot-stuff.
 
I suggest that Nader's consistent wins are likely to represent genuine winnability for a candidate with Nader's 
policy positions 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.electorama.com/pipermail/election-methods-electorama.com/attachments/20111023/09ca9511/attachment-0003.htm>


More information about the Election-Methods mailing list