[Election-Methods] PR favoring racial minorities (Gilmour/Dopp?)

[Kristofer Munsterhjelm]

Warren Smith wrote:
> I agree in principle with Gilmour that some PR systems - such as my
> own "asset voting"
> and "reweighted range voting" as well as PR-STV systems - see
> 
> http://www.rangevoting.org/Asset.html
> http://www.rangevoting.org/RRV.html
> http://www.rangevoting.org/PropRep.html
> 
> for background - should cure underrepresentation of
> racial minorities.
> 
> BUT this conclusion is NOT based on real world data. Neither asset nor
> RRV have ever been tried
> as far as I know; and while PR-STV has been tried, it plainly has NOT
> led to tremendously representative legislatures.
> 
> Instead it is based on the mathematical theorem that
> if people vote in a racial manner, then if X percent of people vote
> black, and enough black candidates are available, then we get X percent
> black winners.

For what it's worth, my proportionality test doesn't judge RRV very 
well. The test seems to have a small-party bias, but RRV is still scored 
worse than, say, STV.

To my understanding, RRV is constructed so that it reduces to D'Hondt if 
people stick to party lines and vote max for all within the party and 
min to the rest. Is that true? If so, there's another method that can be 
derived from party list methods that could be better - QPQ. Just count 
cardinal rating votes as fractional votes of the maximum, contributing 
to the hopeful candidate according to (rating of next higher rated 
candidate) - (rating of current candidate). But simplicity is lost and 
this doesn't reduce to Range voting, but rather to an RV equivalent of IRV.

Perhaps RRV could work better if it was made to generalize to 
Sainte-Laguë instead of D'Hondt. The RRV explanation you link to by Ivan 
Ryan gives a way of doing so.

But maybe it could be made even better, by somehow using your new 
apportionment method instead of Sainte-Laguë. For party list, P_k would 
be the number of people voting for party K (analogous to the 
"population" of "state" K). However, in a party-neutral method, there is 
no party for which we know the "population" of. If there is a simple way 
to generalize your apportionment method to a party-neutral election 
system, I can't see it; still, it would be interesting if such a method 
existed.

In any case, the left-right problem would still be a limitation to RRV, 
where ballots are set so that RV (and any sensible method) would elect 
Center first, but where electing an assembly of two should elect Left 
and Right. For some reason, that problem doesn't appear in party list, 
but I don't know why.

> Party-list-based PR systems should also lead in principle to representativeness,
> e.g. if X percent vote for the "black party" then X percent of the
> seats will be won by that party.
> 
> But, for example, Israel with a party-list PR system has an extremely
> ethnically biased legislature.
> Far more so than the  USA's plurality-single-winner-district legislature.
> So in practice this mathematical conclusion too can fail to be realized.
> I believe that is mainly because Israel's system is rigged in various
> ways reminiscent of the
> USA's "Jim Crow" era and South Africa "apartheid."

I don't think that's unintuitive. Israel uses closed list PR, which puts 
the power in the hands of the list-makers, usually the party elite, 
meaning that change from within a party is limited. For other reasons 
(like those you refer to), change from without is also limited.

> Finally I've been asked (since I advocate RANGE VOTING) what that
> would do for/against racial
> minorities.  I must admit that I currently do not have the  faintest idea.

Range voting wouldn't have much of a direct effect, I think. For 
single-member districts, any exaggeration on part of the majority can be 
met with exaggeration by the minority, until the system reduces to 
approval voting. At that point, the further dynamics of the system 
depends on whether some majority voters are friendly towards minority 
candidates (approve of them too), and on spatial distribution of 
minority and majority members. Even with a good voting system, 
gerrymandering could have some effect, but probably less of an effect 
than bad voting systems (like Plurality) since the gerrymandering would 
have to focus more on separating the voters than on altering the 
distribution to fit with a way of exploiting the election method.

For multimember districts, one wouldn't use range voting, but a PR 
method. Short of SNTV-type balance, using majoritarian RV on a 
multimember district would give the entire district to the majority. 
It's possible to pervert multimember districts so that they are even 
more majoritarian than single-member ones (like Singapore has done with 
their GRCs), but  whoever would be doing such would probably not be 
interested in Range in the first place.

Range might have an indirect effect. If Range voting for single-member 
positions (president, governor, etc) lets third parties survive, which 
one would imagine good election methods would do, then some of these 
parties could become minority-oriented and thus make the minority more 
visible. Such a party could also contest legislative seats and have a 
greater chance of getting minority members elected if the "big parties" 
rely too much on a minority-hostile elite and independents aren't 
sufficiently visible.