[EM] 12/19/02 - Ballots from an Approval election:

[James Gilmour]

> 12/19/02 - Ballots from an Approval election:
Donald Davison wrote:
> The great thinker, Tom Ruen, has kindly given me the ballots from an
> election he conducted last year.  I share them with you, being as Approval
> ballots are hard to come by.
>
> You Condorcet people will not be able to work the ballots as a Condorcet
> election because the ballots are not ranked, that's the problem with
> Approval.

I'm not sure what Donald meant by this comment, but I doubt the validity of Tom
Ruen's "IRV" analysis because of the assumptions he made in the absence of actual
rankings.

> Please notice in the last chart that forty-four percent of the voters made
> two or less choices.

This should be no surprise because these electors were voting in an Approval
election.  I would expect a very different result from a real IRV election with 12
candidates.

We do know from real public elections that voter behaviour changes when the voting
system is changed.  In particular, with preferential voting, electors will mark
several or many preferences, not just one or two.  The only way to make a valid
comparison is to ask the same voters to vote for the same set of candidates by
each of the different methods under comparison.  There have been one or two
limited attempts to do this as "exit polls" alongside public elections in the UK,
but I am not aware of any full comparison being made.

James

----
For more information about this list (subscribe, unsubscribe, FAQ, etc), 
please see http://www.eskimo.com/~robla/em