[EM] bicameral design poll

Alex Small asmall at physics.ucsb.edu
Mon Jan 5 12:01:50 PST 2004

Jeffrey ONeill said:
> Hi Alex,
> Interesting question.  I agree with your answer except for the size of
> the districts.  According to Duverger, a district size of 10 supports 11
> parties.  If each party runs a full slate, then there would be 110
> candidates.  I would prefer a district size of 4-5 to reduce the number
> of candidates.

In principle that may be possible, although I've only heard Duverger's law
discussed in the context of single-member districts elected by plurality. 
In practice, few countries with N-member districts have N+1 parties
winning seats.  The number of parties varies widely from country to
country, and may be more an indicator of political culture than district
size, at least once the district size exceeds a certain threshold (I have
no idea what that threshold is).

My hunch is that with 10 members per district and hence a quota of roughly
9%-10% (depending on the method used), we'd get 3-4 major parties and a
handful of minor parties in the US.  The 2 major parties would probably
each divide in half, but they'd still be good at catering to the center
(200 years of experience) and hence building big coalitions.


More information about the Election-Methods mailing list