[EM] poll fever

[Gervase Lam]

> Date: Wed, 11 Feb 2004 06:12:09 -0800
> From: James Green-Armytage
> Subject: [EM] poll fever

> I'd suggest that we give the candidates cardinal ratings from one to a
> hundred too, but I guess that may as well be optional, since I don't
> want to make this any more complicated than it is already. I'm talking
> about *sincere* cardinal ratings, if we can do that; I'd be interested
> to see how the CR results match up with the other results.

I think having CR is a very good idea.  From CR you can easily get 
cardinal utilities.  As result, it would add something extra with regards 
to the analysis of the results.  You can obviously get the rankings from 
the CRs.

I would suggest having a CR scale of 0 to 10 and no more, with only whole 
number ratings being allowed.  The reason I say this is because it is what 
most people are used to (i.e. mark each candidate out of 10).  Also it is 
due to a post I remember seeing a while back about the how much humans can 
categorise things:

<http://lists.electorama.com/pipermail/election-methods-electorama.com/2001-September/006624.html>

The fact that the whole number 0 to 10 CR ballot will be converted 
directly into a ranked ballot will at least discourage voters from voting 
just at extremes of the CR scale.

Also, in order to elicit more sincere CRs, may be the method that Kevin 
Venzke came up with late September last year could be used:

<http://lists.electorama.com/pipermail/election-methods-electorama.com/2003-September/010975.html>

The way the method processes each ballot means that you don't seem to gain 
very much from giving it insincere CRs.  So you therefore might as well 
give it sincere CRs.

There was also a thread between Kevin and Forest with regards to the 
method about cut-off points and always approving/disapproving the 
highest/lowest cardinally rated candidates regardless of anything that 
happens:

<http://lists.electorama.com/pipermail/election-methods-electorama.com/2003-September/011008.html>

Thanks,
Gervase.