[EM] Filling Unscheduled Vacancies With PR

Stephane Rouillon stephane.rouillon at sympatico.ca
Thu Nov 6 06:03:02 PST 2003


I disagree with recounting methods of replacement from
next persons on closed list because the point of having a
vacancy election is not only to replace gone representatives,
but it aims toward an update of the electorate opinion.

So to take that form of recall element (on a governmental basis, not
on an individual) in account is a major element of democracy. It helps
not electing monarchs for several years.

Now how to do it while preserving a PR view?
First let's assume the last by-election gives the best image of
the party distribution wanted by the electorate. So this is the goal,
exactly like the first distribution was our goal for the full chamber.
Now instead of having all the degrees of freedom given by the
complete chamber, we have to preserve existing representatives.
So we need to build a seat attribution process that would minimize
the final PR while giving free seats.

An example:

general election (100 seats):
Rep 53%
Dem 39%
Green 8%

Now suppose 1 green, 2 democrats and 4 reps are gone.
The by-election gives the following distribution:
Rep 40%
Dem 55%
Green 5%

So we have Rep(49 seats) that sould go to 40, Dem(37 seats) that should go to
55
and Greens (7 seats) that should go to 5 seats. Because we do not remove seats
to
any party the Dems get the 7 seats and the new chamber is still leaded by
Republicans
Rep: 49 seats
Dem: 44 seats
Green: 7 seats

If two or more party are to receive seats, I suggest using norm-1 to minimize
the
Pr error: Sigma(i) |Xi-Si|.

This shows PR can very well be used for the american senate election even if
only
a third of senators are replaced.

Steph.

Alex Small a écrit :

> Say that a legislator resigns or dies in the middle of his term.  If we
> are electing the legislators from single-member districts then the way to
> fill the vacancy is pretty easy:  Use the same method as for regular
> elections (except perhaps making modifications if the lack of partisan
> primaries creates a large field of candidates and the election method
> works best for smaller fields, but that's a separate issue).
>
> Now, suppose we elect the legislators via PR.  How to fill the vacancy?
>
> If we just hold a standard single-winner election and the resigning
> legislator is from a smaller party, it's almost certain that the
> replacement will be from a larger party.  The whole point of PR is to give
> the various portions of the electorate representation according to their
> size, so filling the vacancy in this way upsets proportionality.
>
> (At this point somebody will probably point out that organized political
> parties are not the only way to divide up the electorate, so we should use
> candidate-based methods rather than party lists.  While I more or less
> agree with that view, in theoretical discussions the concept of a party is
> still useful.  Implicit in proportionality is that various ideas and
> factions are represented according to their size, so we need to think
> about these factions to see if the method is achieving proportionality,
> and the word "party" is as good a way as any to label those factions.  So
> bear with me here.)
>
> Now, if we used party lists then the resigning legislator could simply be
> replaced by somebody from his party, perhaps whoever on his list got the
> most votes without being elected (in an open list method).  But if we use
> candidate-based methods (e.g. STV, PAV, cumulative voting, etc. etc.) then
> we're saying that, while political parties will undoubtedly exist in
> practice, they have no fundamental role in the design of our electoral
> system.  So it would seem dubious to let party leaders pick the
> replacement.
>
> What do people here think is the best way to replace legislators in the
> middle of a term while retaining proportionality?  What solutions have
> various countries devised?
>
> Alex
>
> ----
> Election-methods mailing list - see http://electorama.com/em for list info




More information about the Election-Methods mailing list