[fvc-wat-disc] "Local Proportional Representation"

[Evan Rosamond]

Hi Eleanor,

Yes, your concerns are accurately expressed. The AllVotesCount.ca 
explanation seems a bit incomplete. So I got curious and tried to create 
a test case based on the 5 ridings (Guelph-Oakville etc.) in the web 
site's examples and the 2015 election results. To help me guess at 2nd 
and following preferences, I assumed that most voters would continue to 
vote for the same party until all the candidates were gone, and that 
voters would stop assigning preference numbers at some time before 
voting for all candidates.

In these 5 ridings, 3 Libs and 2 Cons were elected with FPTP. Using LPR, 
the same 5 people would probably have been elected. The best 
proportional result would have been 2 Libs, 2 Cons and 1 NDP. The only 
weird happening was that Lisa Raitt would be the first candidate 
elected, even though she never achieved the quota. All the other Milton 
candidates were eliminated early, so Raitt was in. That's because Milton 
had fewer votes cast than the other ridings.

So it seems like the results under LPR aren't going to much more 
proportional than FPTP. Perhaps Byron Weber Becker can calculate a 
Gallagher Index for us.

Still curious, I tried the same experiment with the 5 ridings in 
Waterloo Region. This produced some bizarre results. With FPTP in 2015, 
4 Libs and 1 Con were elected. The best proportional result would be 2 
Libs, 2 Cons and 1 NDP. The LPR system generated those ideal numbers! 
But the way it did that is going to cause a lot of howling from 
candidates. Bardish Chagger got the most 1st preference votes in the 5 
ridings, but was not elected in my LPR model. That's because the lower 
preferences of defeated candidates (NDP and Green) pushed Diane Freeman 
over the quota. So Chagger loses because there were still 2 other Libs 
in the race from whom she couldn't get any lower preference votes, 
whereas with Freeman, all the other NDP and Green candidates had been 
defeated. This kind of thing happens, and it's why STV is sort of 
proportional sometimes, but I bet the candidates won't like it.

According to the ERRE report, there are 2 other ways to count ranked 
ballots. One is to assign points to the candidates based on the 
preference number assigned. You get the number of points by subtracting 
the preference number from the number of candidates. E.g. with 26 
candidates, a voter who writes 1,2,3 on his ballot will be giving his 
candidates 25,24,23 points. then the computer adds up all the points to 
get a total for each candidate. This system has the advantage of 
counting all the preferences. Candidates are not eliminated. There is 
only 1 totalling of points, instead of many rounds of counting. I'm not 
sure how this will work with LPR, it might make the result the same as 
FPTP, but I think somebody should try to figure this out.

Another thing that might upset candidates is that they will have to rely 
on votes coming from the other ridings in their cluster. They are 
actually being elected/defeated by all the voters in the 5 ridings, but 
they will be MP of only one if they win.

I hope this answers as many questions as it raises.

Evan



On 3/3/2017 3:35 PM, Eleanor Grant wrote:
> Hi Fair Voters,
>
> Cathy and Salim and I went to the meeting in Guelph Thurs evening 
> about the new initiative called Local PR.  Their Liberal MP Lloyd 
> Longford is prepared to push it forward if they get enough 
> signatures.  They'll be at Guelph market Sat March 11.
>
> The speaker, Steve Dyck, said that Local PR has the support of Nathan 
> Cullen, Eliz May, and Jean-Pierre Kingsley - and even our Anita N.  It 
> has the advantage that it can be implemented more quickly than other 
> PR systems because it would not require new riding boundaries or more 
> MPs.  In this way we could call Trudeau's bluff, that there's "no 
> consensus on a system" and "too hasty" to do by 2019.
>
> Web site AllVotesCount.ca
> Twitter @AllVotesCount , @SteveDyck .
>
> The system was not quite explained to my satisfaction.  I couldn't see 
> how it's really different from STV - you just have *clusters* of 
> ridings instead of larger ridings.  And every riding has to have one 
> MP - so I couldn't see where it would actually work out be fair to the 
> small parties.  Depending on the number of ridings in the cluster, the 
> *threshold* at which a smaller party would get a seat would vary: the 
> bigger the cluster, the lower the threshold.
>
> It appears to be less than true PR, but perhaps more palatable to 
> those who are hesitant about change, and definitely an improvement 
> over FPTP.
>
> Cath and Salim, and Anita, do you think I've explained it accurately?
>
> If Guelph succeeds in spear-heading Local PR, there'll be a push to 
> get support across Canada in a very short time frame - like 2 months.
>
> FYI,
> Eleanor
>
>
>
>
>
> _______________________________________________
> This is the fvc-wat-disc mailing list
> Post a message: fvc-wat-disc at listserv.thinkers.org
> Unsubscribe: http://listserv.thinkers.org/mailman/listinfo/fvc-wat-disc_listserv.thinkers.org


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.thinkers.org/pipermail/fvc-wat-disc_listserv.thinkers.org/attachments/20170303/92fd1855/attachment-0002.htm>