[EM] electoral methods - US and Europe

[Toplak Jurij]

Yes, the in US the methods are used for allocation of 435 seats to 50 states
and in Europe they are used for allocation of seats to the parties. However,
mathematically, problem is the same.
With "ideal proportionality" I mean that party (or state in a case of US)
would get exactly the same percentage of seats as it received votes (or
people in a case of US). Most of the PR methods are supposed to try to come
as close as possible, but, of course, none of them can achieve it (because a
party cannot be assigned 5.43 seats, for example :)

Altough it is often thought that D'Hondt is superior to Saint-Lague, as you
say, all the research shows that D'Hondt systematically and consisently
favors large parties, while Saint-Lague does not favour either large or
small ones. (see Lijphart 1985, Balinski & Young 2001). For this reason
D'Hondt was also abandoned in US in 1832 (after 40 years of use), therefore
46 years before D'Hondt "invented" it.

Jure
ps. si stands for Slovenia


"Toplak Jurij" <jure.toplak at uni-mb.si> writes:
>I am finishing an article on apportionment/seat allocation and I would
>like to make some things clear. Please correct me if I am wrong in any of
>the facts below. I also would apreciate any comment on this, because I
>find it quite strange.
>
>I understand that in 19th century European scientists were not aware of
>American research. But what really surprises me is that in 1990 European
>research and even American Political Science research did not know (at
>list they don't cite it) that it has all been invented already. Therefore:
>
>1. Jefferson Method (1791) was reinvented in Europe by Victor D'Hondt
>(1878).
>2. Hamilton Method (1791) was reinvented in Europe by Hare.
>3. Webster Method (1832) was reinvented in Europe by Saint-Lague.
>4. The discussion about the "most proportional" method followed in
>Europe, too, and...
>
>5. American National Academy of Sciences in 1920 concuded that none of
>the methods is "most" proportional, because it all depends on how we
>measure the discrepancies from ideal proportionality.
>European research came to exactly same conclusion around 1985-1991
>without even mentioning the one from 1920.
>(see Gallagher (1991). Proportionality, Disproportionality and Electoral
>Systems. Electoral Studies 10:1, 33-51. Something also Lijphard (1985)The
>Field of Electoral Systems Research: A Critical Survey. Electoral Studies
>4:1, 3-14).
>
>I'll be grateful for any comments.
>
>Jure

That's really interesting. I don't know if I can add anything, but
hopefully somebody from the list can. Am I right in thinking that the
American methods you mention are for representation of states in Congress,
and the European methods are for proportional representation of parties? I
guess that they are the same problem in some fundamental ways, but there
are some substantial differences too. For one, every voter in a state
should be accounted for as being a member in a district, but not every
voter in a party list PR election will see someone elected from the party
they voted for. In general, these are two political problems with
different aims, despite very similar mathematical elements. Perhaps that
partly explains why the two bodies of research didn't pool together
properly. Or maybe it's just because voting systems have never gotten
enough attention to become a really organized field. Or maybe all social
science fields are that disorganized. Are they?
What exactly do you mean by "ideal proportionality"? That's interesting.
I think that there are some very substantial reasons why the D'Hondt or
Jefferson divisor method, along with its near-equivalent the Droop quota
largest remainder method, is clearly superior to the Saint-Lague divisors
or Hare quota... at least for the purposes of proportional representation.
I haven't really studied seat allocation by region, and I don't know if it
has a set of paradoxes which I'm unaware of.

James

P.S. By the way, just out of curiosity, where are you writing from? I'm
not familiar with the si suffix.