[EM] Cretney's compromise method

[Blake Cretney]

" " <donald at mich.com (Donald E. Davison)>, on the subject of '[EM]
Cretney's compromise method', is quoted as:
>- - - - - - - - - - - - - - - - - - - - - - - - - - - 11-16-00
>Dear Blake Cretney and the EM list,
>
>     It is possible to divide the seats according to the party quotas
and
>it is also possible to divide the seats according to the quotas of
votes
>cast in each district.
>     But, the two divisions are not the same. How are you going to
decide
>which party is to receive the seats in a district?
>
>     Suppose a district is assigned three seats because three quotas
of
>votes were cast in the district. Suppose also there are ten candidates
and
>eight parties in the district, with the leading candidate having only
a
>half a quota of votes. If you give the three seats to the three
leading
>candidates this will not maintain the party proportionality of the
entire
>election.

Districts are assigned seats based on eligible voters, or more
precisely, the total area is divided into roughly equal districts based
on eligible voters.  Just as is done now.  This happens before the
election.

This, of course, means that if there is a great discrepancy between
voter turn-out in different districts, the national party percentages
may be different than what you would get by giving each party its local
percentage.  This isn't necessarily the case, though.  If this happens,
the formula appears to handle it by compromising the party percentages
achieved in the low voter turn-out districts.

As for your question, the method assigns the parties a fixed number of
national seats based on national vote totals, using St. Lague.  It then
tries to place candidates in districts according to their support, also
using St. Lague.  So, the leading 3 parties in a district may not all
win (especially if they get under 1/6 support).  Of course, internal to
a party, the candidates with the most votes always win.

---
Blake Cretney