[EM] (no subject)

Bart Ingles bartman at netgate.net
Sun Oct 5 15:06:03 PDT 2003


Dgamble997 at aol.com wrote:
> 
> Bart Ingles wrote:
> 
> >I think when mentioning criteria, it's a good idea to also state why
> >those criteria might be important. For example, to me criteria such as
> >FBC and Participation are important because they relate to a voting
> >system's immunity to the Duverger effect.


> In my opinion the first criterion a single seat method must satisfy in
> order to be acceptable is :
> If a candidate is the first choice of a majority of voters (50% +)
> that candidate must win.

If this was a response to my post, then you ignored my point.  Why do
you consider "majority" to be more important that all other criteria?


> IRV, Condorcet, Bucklin and Plurality pass this criterion, Approval
> and Borda fail it.

That's debatable, depending how you determine compliance.  If basing it
on actual voted ballots, then Approval complies with the criterion (if
only one candidate is approved on a majority of ballots, that candidate
wins).  If determining compliance based on sincere preference orders,
then compliance depends on the voters casting sincere ballots.

In theory at least, it is possible under all of these systems for bad
polling data to induce voters to abandon a majority first choice
candidate, resulting in a win for some unnecessary compromise.  This may
sound unlikely when considering a ranked ballot system, but is also
precisely what is required for an instance of noncompliance under
Approval.

I view the majority criterion as important when applied to actual voted
ballots, since failure is a direct contradiction to a reasonable
interpretation of the ballot information (just as a Condorcet winner
failure contradicts ranked ballot info).  I view MC as applied to
sincere preference orders as less important than other criteria.


> As regards Duverger he said the following:
> 
> > To these socio-economic and historical factors a technical factor
> > must be added: the electoral system. I expressed its effects in 1946
> > in the formulation of three sociological laws: (1) a majority vote
> > on one ballot is conducive to a two-party system; (2) proportional
> > representation is conducive to a multiparty system; (3) a majority
> > vote on two ballots is conducive to a multiparty system, inclined
> > toward forming coalitions.
> 
> http://www.janda.org/c24/Readings/Duverger/Duverger.htm

It is my understanding that item (1) is what is usually referred to as
"Duverger's Law" or "Duverger's Principle".  This law or principle is
stated in terms of a cause (majority vote on one ballot) and effect (a
two-party system).  Since I am interested in which other voting systems
are conducive to a two-party system, I refer to second half of
Duverger's Law-- the "effect" part.  Hence "Duverger Effect".  Does
anyone else find this term objectionable?


> Majority vote on two ballots (the electoral system of the French Third
> and Fifth Republics) is extremely similar to IRV and top two runoff.

In Duverger's text, he refers to a system where candidates withdraw
after the first ballot, or where voters switch their votes to the
stronger candidate.  There was no explicit mention of "elimination". 
But he does mention the Third and Fifth Republics.  You say the system
used there was "similar" to top-two runoff-- do you mean "identical", or
were there differences?

In any case, I don't deny that the pressures in favor of a two-party
system are somewhat weaker with a runoff than with plain plurality, but
the question is how much weaker.  In fact the kind of multi-party system
that Duverger sees arising from two ballots is not much different from a
two-party system.  To quote from the link above: 

>>> The sharpest difference with the system of proportional
>>> representation concerns electoral alliances.  A coalition
>>> system par excellence, the two ballot regime can sometimes
>>> permit the formation of a dual system of alliances, introducing
>>> a sort of two-party system in the midst of a multi-party
>>> situation. This phenomenon was quite evident in France during
>>> the Third and the Fifth Republics, and in Germany from 1870 to 1914.

I should make it clear that I wouldn't expect Approval or Condorcet to
produce a multi-party system comparable to PR, but would at least expect
to see more "rotation" or evolution among multiple parties depending on
which issues happen to be relevant at a given time.  In other words
something more complex than the two-way pendulum swing we have now.


> How is participation (surely the winner of the September objective
> criterion of the month prize under any method) relevant to Duverger ?

I thought I had withdrawn that comment.  At least the mere fact of
compliance seems to have no bearing on the two-party system, since Borda
and Plurality comply, and Condorcet doesn't.  There may be some bearing
when you consider frequency of non-compliance among other systems, but
there are other criteria more relevant to the Duverger Effect.

Bart



More information about the Election-Methods mailing list