After reading the post on measuring multiwinner goodness (and writing a
reply to it), I started to think of how to determine how good the
different multiwinner methods actually are.
One way to do that is by criterion compliance. But there is another:
while proportionality can't be expressed in terms of majoritarian
quality (preference), we can still determine the Pareto front, given a
way to measure proportionality and preference.
There are many ways to do so - I have chosen one combination (normalized
Sainte-Lague index and Bayesian regret). The results are here:
http://munsterhjelm.no/km/elections/multiwinner_tradeoffs/
May it be of interest! :-)
----
Election-Methods mailing list - see http://electorama.com/em for list info
