[EM] Manipulability stats for more poll methods (fixed footnotes)

[Kristofer Munsterhjelm]

On 2024-05-05 12:50, Michael Ossipoff wrote:
> Oh, I see what you meant. You didn’t use the Above-Mean strategy. I just 
> assumed that that was your intention.
> 
> Instead you renormalized the ballots to 0-1, & rounded off a candidate’s 
> renormalized rating to the nearest integer.
> 
> Forgive me for jumping to conclusions. It was just that Above-Mean is 
> the assumed Approval-strategy that I’ve heard of in simulations. That of 
> course doesn’t make it better.
> 
> I hadn’t heard of it being done as you did it.

I guess I was being imprecise. To restate what I said in the other post, 
but in fewer words:

- For Range (Smith//Range, STAR, etc): I use a generalization of the 
midpoint approach.
- For Approval (Smith//Approval explicit implicit etc): I use mean 
utility cutoffs.

These seem to be the obvious ways to do it for the respective domains.

-km